From de2aeaa887aff666e7e8e98c1d4733efa790900f Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 13:30:02 -0400
Subject: [PATCH 01/66] Create cmake.yml

Added cmakeaction
---
 .github/workflows/cmake.yml | 38 +++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
 create mode 100644 .github/workflows/cmake.yml

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
new file mode 100644
index 0000000..e282e38
--- /dev/null
+++ b/.github/workflows/cmake.yml
@@ -0,0 +1,38 @@
+name: CMake
+
+on:
+  push:
+    branches: [ master ]
+  pull_request:
+    branches: [ master ]
+
+env:
+  # Customize the CMake build type here (Release, Debug, RelWithDebInfo, etc.)
+  BUILD_TYPE: Release
+
+jobs:
+  build:
+    # The CMake configure and build commands are platform agnostic and should work equally
+    # well on Windows or Mac.  You can convert this to a matrix build if you need
+    # cross-platform coverage.
+    # See: https://docs.github.com/en/free-pro-team@latest/actions/learn-github-actions/managing-complex-workflows#using-a-build-matrix
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v2
+
+    - name: Configure CMake
+      # Configure CMake in a 'build' subdirectory. `CMAKE_BUILD_TYPE` is only required if you are using a single-configuration generator such as make.
+      # See https://cmake.org/cmake/help/latest/variable/CMAKE_BUILD_TYPE.html?highlight=cmake_build_type
+      run: cmake -B ${{github.workspace}}/build -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}}
+
+    - name: Build
+      # Build your program with the given configuration
+      run: cmake --build ${{github.workspace}}/build --config ${{env.BUILD_TYPE}}
+
+    - name: Test
+      working-directory: ${{github.workspace}}/build
+      # Execute tests defined by the CMake configuration.  
+      # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
+      run: ctest -C ${{env.BUILD_TYPE}}
+      

From e24e49c8e7bbcebb770c2eec91748cafc62c4051 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 13:31:12 -0400
Subject: [PATCH 02/66] Create msbuild.yml

Added msbuild
---
 .github/workflows/msbuild.yml | 32 ++++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 .github/workflows/msbuild.yml

diff --git a/.github/workflows/msbuild.yml b/.github/workflows/msbuild.yml
new file mode 100644
index 0000000..29b6ace
--- /dev/null
+++ b/.github/workflows/msbuild.yml
@@ -0,0 +1,32 @@
+name: MSBuild
+
+on: [push]
+
+env:
+  # Path to the solution file relative to the root of the project.
+  SOLUTION_FILE_PATH: .
+
+  # Configuration type to build.
+  # You can convert this to a build matrix if you need coverage of multiple configuration types.
+  # https://docs.github.com/actions/learn-github-actions/managing-complex-workflows#using-a-build-matrix
+  BUILD_CONFIGURATION: Release
+
+jobs:
+  build:
+    runs-on: windows-latest
+
+    steps:
+    - uses: actions/checkout@v2
+
+    - name: Add MSBuild to PATH
+      uses: microsoft/setup-msbuild@v1.0.2
+
+    - name: Restore NuGet packages
+      working-directory: ${{env.GITHUB_WORKSPACE}}
+      run: nuget restore ${{env.SOLUTION_FILE_PATH}}
+
+    - name: Build
+      working-directory: ${{env.GITHUB_WORKSPACE}}
+      # Add additional options to the MSBuild command line here (like platform or verbosity level).
+      # See https://docs.microsoft.com/visualstudio/msbuild/msbuild-command-line-reference
+      run: msbuild /m /p:Configuration=${{env.BUILD_CONFIGURATION}} ${{env.SOLUTION_FILE_PATH}}

From 412c6c271181a506df7fb0c87e27185625329a70 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 13:32:23 -0400
Subject: [PATCH 03/66] cmake

cmake
---
 .github/workflows/c-cpp.yml | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)
 create mode 100644 .github/workflows/c-cpp.yml

diff --git a/.github/workflows/c-cpp.yml b/.github/workflows/c-cpp.yml
new file mode 100644
index 0000000..e323326
--- /dev/null
+++ b/.github/workflows/c-cpp.yml
@@ -0,0 +1,23 @@
+name: C/C++ CI
+
+on:
+  push:
+    branches: [ master ]
+  pull_request:
+    branches: [ master ]
+
+jobs:
+  build:
+
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: configure
+      run: ./configure
+    - name: make
+      run: make
+    - name: make check
+      run: make check
+    - name: make distcheck
+      run: make distcheck

From 0947a5615ff6e90613418f0b1d837e2909a2f3a0 Mon Sep 17 00:00:00 2001
From: VLT Media <info@vltmedia.com>
Date: Mon, 7 Jun 2021 17:44:12 +0000
Subject: [PATCH 04/66] updated mem.h and cmake

---
 CMakeLists.txt |   1 +
 include/mem.h  | 362 +++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 363 insertions(+)
 create mode 100644 include/mem.h

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 2771644..a50c16f 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -8,6 +8,7 @@ include_directories(include)
 
 set(SOURCE_FILES
         include/stb_image.h
+        include/mem.h
         src/main.c
         src/stb_image.c
         src/normalmap.c
diff --git a/include/mem.h b/include/mem.h
new file mode 100644
index 0000000..3ecb601
--- /dev/null
+++ b/include/mem.h
@@ -0,0 +1,362 @@
+/*
+ * Copyright (c) 2013-present, Yann Collet
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef MEM_H_MODULE
+#define MEM_H_MODULE
+
+#if defined (__cplusplus)
+extern "C" {
+#endif
+
+/*-****************************************
+*  Dependencies
+******************************************/
+#include <stddef.h>     /* size_t, ptrdiff_t */
+#include <string.h>     /* memcpy */
+
+
+/*-****************************************
+*  Compiler specifics
+******************************************/
+#if defined(_MSC_VER)   /* Visual Studio */
+#   include <stdlib.h>  /* _byteswap_ulong */
+#   include <intrin.h>  /* _byteswap_* */
+#endif
+#if defined(__GNUC__)
+#  define MEM_STATIC static __inline __attribute__((unused))
+#elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
+#  define MEM_STATIC static inline
+#elif defined(_MSC_VER)
+#  define MEM_STATIC static __inline
+#else
+#  define MEM_STATIC static  /* this version may generate warnings for unused static functions; disable the relevant warning */
+#endif
+
+/* code only tested on 32 and 64 bits systems */
+#define MEM_STATIC_ASSERT(c)   { enum { MEM_static_assert = 1/(int)(!!(c)) }; }
+MEM_STATIC void MEM_check(void) { MEM_STATIC_ASSERT((sizeof(size_t)==4) || (sizeof(size_t)==8)); }
+
+
+/*-**************************************************************
+*  Basic Types
+*****************************************************************/
+#if  !defined (__VMS) && (defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
+# include <stdint.h>
+  typedef   uint8_t BYTE;
+  typedef  uint16_t U16;
+  typedef   int16_t S16;
+  typedef  uint32_t U32;
+  typedef   int32_t S32;
+  typedef  uint64_t U64;
+  typedef   int64_t S64;
+#else
+  typedef unsigned char      BYTE;
+  typedef unsigned short      U16;
+  typedef   signed short      S16;
+  typedef unsigned int        U32;
+  typedef   signed int        S32;
+  typedef unsigned long long  U64;
+  typedef   signed long long  S64;
+#endif
+
+
+/*-**************************************************************
+*  Memory I/O
+*****************************************************************/
+/* MEM_FORCE_MEMORY_ACCESS :
+ * By default, access to unaligned memory is controlled by `memcpy()`, which is safe and portable.
+ * Unfortunately, on some target/compiler combinations, the generated assembly is sub-optimal.
+ * The below switch allow to select different access method for improved performance.
+ * Method 0 (default) : use `memcpy()`. Safe and portable.
+ * Method 1 : `__packed` statement. It depends on compiler extension (i.e., not portable).
+ *            This method is safe if your compiler supports it, and *generally* as fast or faster than `memcpy`.
+ * Method 2 : direct access. This method is portable but violate C standard.
+ *            It can generate buggy code on targets depending on alignment.
+ *            In some circumstances, it's the only known way to get the most performance (i.e. GCC + ARMv6)
+ * See http://fastcompression.blogspot.fr/2015/08/accessing-unaligned-memory.html for details.
+ * Prefer these methods in priority order (0 > 1 > 2)
+ */
+#ifndef MEM_FORCE_MEMORY_ACCESS   /* can be defined externally, on command line for example */
+#  if defined(__GNUC__) && ( defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) || defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6Z__) || defined(__ARM_ARCH_6ZK__) || defined(__ARM_ARCH_6T2__) )
+#    define MEM_FORCE_MEMORY_ACCESS 2
+#  elif defined(__INTEL_COMPILER) || defined(__GNUC__)
+#    define MEM_FORCE_MEMORY_ACCESS 1
+#  endif
+#endif
+
+MEM_STATIC unsigned MEM_32bits(void) { return sizeof(size_t)==4; }
+MEM_STATIC unsigned MEM_64bits(void) { return sizeof(size_t)==8; }
+
+MEM_STATIC unsigned MEM_isLittleEndian(void)
+{
+    const union { U32 u; BYTE c[4]; } one = { 1 };   /* don't use static : performance detrimental  */
+    return one.c[0];
+}
+
+#if defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==2)
+
+/* violates C standard, by lying on structure alignment.
+Only use if no other choice to achieve best performance on target platform */
+MEM_STATIC U16 MEM_read16(const void* memPtr) { return *(const U16*) memPtr; }
+MEM_STATIC U32 MEM_read32(const void* memPtr) { return *(const U32*) memPtr; }
+MEM_STATIC U64 MEM_read64(const void* memPtr) { return *(const U64*) memPtr; }
+MEM_STATIC size_t MEM_readST(const void* memPtr) { return *(const size_t*) memPtr; }
+
+MEM_STATIC void MEM_write16(void* memPtr, U16 value) { *(U16*)memPtr = value; }
+MEM_STATIC void MEM_write32(void* memPtr, U32 value) { *(U32*)memPtr = value; }
+MEM_STATIC void MEM_write64(void* memPtr, U64 value) { *(U64*)memPtr = value; }
+
+#elif defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==1)
+
+/* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */
+/* currently only defined for gcc and icc */
+#if defined(_MSC_VER) || (defined(__INTEL_COMPILER) && defined(WIN32))
+    __pragma( pack(push, 1) )
+    typedef struct { U16 v; } unalign16;
+    typedef struct { U32 v; } unalign32;
+    typedef struct { U64 v; } unalign64;
+    typedef struct { size_t v; } unalignArch;
+    __pragma( pack(pop) )
+#else
+    typedef struct { U16 v; } __attribute__((packed)) unalign16;
+    typedef struct { U32 v; } __attribute__((packed)) unalign32;
+    typedef struct { U64 v; } __attribute__((packed)) unalign64;
+    typedef struct { size_t v; } __attribute__((packed)) unalignArch;
+#endif
+
+MEM_STATIC U16 MEM_read16(const void* ptr) { return ((const unalign16*)ptr)->v; }
+MEM_STATIC U32 MEM_read32(const void* ptr) { return ((const unalign32*)ptr)->v; }
+MEM_STATIC U64 MEM_read64(const void* ptr) { return ((const unalign64*)ptr)->v; }
+MEM_STATIC size_t MEM_readST(const void* ptr) { return ((const unalignArch*)ptr)->v; }
+
+MEM_STATIC void MEM_write16(void* memPtr, U16 value) { ((unalign16*)memPtr)->v = value; }
+MEM_STATIC void MEM_write32(void* memPtr, U32 value) { ((unalign32*)memPtr)->v = value; }
+MEM_STATIC void MEM_write64(void* memPtr, U64 value) { ((unalign64*)memPtr)->v = value; }
+
+#else
+
+/* default method, safe and standard.
+   can sometimes prove slower */
+
+MEM_STATIC U16 MEM_read16(const void* memPtr)
+{
+    U16 val; memcpy(&val, memPtr, sizeof(val)); return val;
+}
+
+MEM_STATIC U32 MEM_read32(const void* memPtr)
+{
+    U32 val; memcpy(&val, memPtr, sizeof(val)); return val;
+}
+
+MEM_STATIC U64 MEM_read64(const void* memPtr)
+{
+    U64 val; memcpy(&val, memPtr, sizeof(val)); return val;
+}
+
+MEM_STATIC size_t MEM_readST(const void* memPtr)
+{
+    size_t val; memcpy(&val, memPtr, sizeof(val)); return val;
+}
+
+MEM_STATIC void MEM_write16(void* memPtr, U16 value)
+{
+    memcpy(memPtr, &value, sizeof(value));
+}
+
+MEM_STATIC void MEM_write32(void* memPtr, U32 value)
+{
+    memcpy(memPtr, &value, sizeof(value));
+}
+
+MEM_STATIC void MEM_write64(void* memPtr, U64 value)
+{
+    memcpy(memPtr, &value, sizeof(value));
+}
+
+#endif /* MEM_FORCE_MEMORY_ACCESS */
+
+MEM_STATIC U32 MEM_swap32(U32 in)
+{
+#if defined(_MSC_VER)     /* Visual Studio */
+    return _byteswap_ulong(in);
+#elif defined (__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ >= 403)
+    return __builtin_bswap32(in);
+#else
+    return  ((in << 24) & 0xff000000 ) |
+            ((in <<  8) & 0x00ff0000 ) |
+            ((in >>  8) & 0x0000ff00 ) |
+            ((in >> 24) & 0x000000ff );
+#endif
+}
+
+MEM_STATIC U64 MEM_swap64(U64 in)
+{
+#if defined(_MSC_VER)     /* Visual Studio */
+    return _byteswap_uint64(in);
+#elif defined (__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ >= 403)
+    return __builtin_bswap64(in);
+#else
+    return  ((in << 56) & 0xff00000000000000ULL) |
+            ((in << 40) & 0x00ff000000000000ULL) |
+            ((in << 24) & 0x0000ff0000000000ULL) |
+            ((in << 8)  & 0x000000ff00000000ULL) |
+            ((in >> 8)  & 0x00000000ff000000ULL) |
+            ((in >> 24) & 0x0000000000ff0000ULL) |
+            ((in >> 40) & 0x000000000000ff00ULL) |
+            ((in >> 56) & 0x00000000000000ffULL);
+#endif
+}
+
+MEM_STATIC size_t MEM_swapST(size_t in)
+{
+    if (MEM_32bits())
+        return (size_t)MEM_swap32((U32)in);
+    else
+        return (size_t)MEM_swap64((U64)in);
+}
+
+/*=== Little endian r/w ===*/
+
+MEM_STATIC U16 MEM_readLE16(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_read16(memPtr);
+    else {
+        const BYTE* p = (const BYTE*)memPtr;
+        return (U16)(p[0] + (p[1]<<8));
+    }
+}
+
+MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val)
+{
+    if (MEM_isLittleEndian()) {
+        MEM_write16(memPtr, val);
+    } else {
+        BYTE* p = (BYTE*)memPtr;
+        p[0] = (BYTE)val;
+        p[1] = (BYTE)(val>>8);
+    }
+}
+
+MEM_STATIC U32 MEM_readLE24(const void* memPtr)
+{
+    return MEM_readLE16(memPtr) + (((const BYTE*)memPtr)[2] << 16);
+}
+
+MEM_STATIC void MEM_writeLE24(void* memPtr, U32 val)
+{
+    MEM_writeLE16(memPtr, (U16)val);
+    ((BYTE*)memPtr)[2] = (BYTE)(val>>16);
+}
+
+MEM_STATIC U32 MEM_readLE32(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_read32(memPtr);
+    else
+        return MEM_swap32(MEM_read32(memPtr));
+}
+
+MEM_STATIC void MEM_writeLE32(void* memPtr, U32 val32)
+{
+    if (MEM_isLittleEndian())
+        MEM_write32(memPtr, val32);
+    else
+        MEM_write32(memPtr, MEM_swap32(val32));
+}
+
+MEM_STATIC U64 MEM_readLE64(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_read64(memPtr);
+    else
+        return MEM_swap64(MEM_read64(memPtr));
+}
+
+MEM_STATIC void MEM_writeLE64(void* memPtr, U64 val64)
+{
+    if (MEM_isLittleEndian())
+        MEM_write64(memPtr, val64);
+    else
+        MEM_write64(memPtr, MEM_swap64(val64));
+}
+
+MEM_STATIC size_t MEM_readLEST(const void* memPtr)
+{
+    if (MEM_32bits())
+        return (size_t)MEM_readLE32(memPtr);
+    else
+        return (size_t)MEM_readLE64(memPtr);
+}
+
+MEM_STATIC void MEM_writeLEST(void* memPtr, size_t val)
+{
+    if (MEM_32bits())
+        MEM_writeLE32(memPtr, (U32)val);
+    else
+        MEM_writeLE64(memPtr, (U64)val);
+}
+
+/*=== Big endian r/w ===*/
+
+MEM_STATIC U32 MEM_readBE32(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_swap32(MEM_read32(memPtr));
+    else
+        return MEM_read32(memPtr);
+}
+
+MEM_STATIC void MEM_writeBE32(void* memPtr, U32 val32)
+{
+    if (MEM_isLittleEndian())
+        MEM_write32(memPtr, MEM_swap32(val32));
+    else
+        MEM_write32(memPtr, val32);
+}
+
+MEM_STATIC U64 MEM_readBE64(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_swap64(MEM_read64(memPtr));
+    else
+        return MEM_read64(memPtr);
+}
+
+MEM_STATIC void MEM_writeBE64(void* memPtr, U64 val64)
+{
+    if (MEM_isLittleEndian())
+        MEM_write64(memPtr, MEM_swap64(val64));
+    else
+        MEM_write64(memPtr, val64);
+}
+
+MEM_STATIC size_t MEM_readBEST(const void* memPtr)
+{
+    if (MEM_32bits())
+        return (size_t)MEM_readBE32(memPtr);
+    else
+        return (size_t)MEM_readBE64(memPtr);
+}
+
+MEM_STATIC void MEM_writeBEST(void* memPtr, size_t val)
+{
+    if (MEM_32bits())
+        MEM_writeBE32(memPtr, (U32)val);
+    else
+        MEM_writeBE64(memPtr, (U64)val);
+}
+
+
+#if defined (__cplusplus)
+}
+#endif
+
+#endif /* MEM_H_MODULE */
\ No newline at end of file

From 34033b43b65828ac511b05995d777c4c3f04518a Mon Sep 17 00:00:00 2001
From: VLT Media <info@vltmedia.com>
Date: Mon, 7 Jun 2021 17:46:55 +0000
Subject: [PATCH 05/66] attempted to fix strcmp

---
 src/main.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/src/main.c b/src/main.c
index 871e319..06018be 100644
--- a/src/main.c
+++ b/src/main.c
@@ -46,19 +46,19 @@ int main(int argc, const char **argv) {
     }
 
     const char *ext = get_filename_ext(outfile);
-    if (strcmpi("png", ext) == 0) {
+    if (strcmp("png", ext) == 0) {
         if (!stbi_write_png(outfile, x, y, 4, image_out, 0)) {
             fprintf(stderr, "Unable to write PNG %s", outfile);
         }
-    } else if (strcmpi("bmp", ext) == 0) {
+    } else if (strcmp("bmp", ext) == 0) {
         if (!stbi_write_bmp(outfile, x, y, 4, image_out)) {
             fprintf(stderr, "Unable to write BMP %s", outfile);
         }
-    } else if (strcmpi("tga", ext) == 0) {
+    } else if (strcmp("tga", ext) == 0) {
         if (!stbi_write_tga(outfile, x, y, 4, image_out)) {
             fprintf(stderr, "Unable to write TGA %s", outfile);
         }
-//    } else if (strcmpi("hdr", ext) == 0) {
+//    } else if (strcmp("hdr", ext) == 0) {
 //        if (!stbi_write_hdr(outfile, x, y, 4, image_out)) {
 //            fprintf(stderr, "Unable to write HDR %s", outfile);
 //        }

From 4e2da798bf1a4cfd5dd9466ac3cdbf02162f2953 Mon Sep 17 00:00:00 2001
From: VLT Media <info@vltmedia.com>
Date: Mon, 7 Jun 2021 18:00:35 +0000
Subject: [PATCH 06/66] added linker for math.h

---
 CMakeLists.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index a50c16f..d260308 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -13,5 +13,6 @@ set(SOURCE_FILES
         src/stb_image.c
         src/normalmap.c
         include/normalmap.h include/stb_image_write.h src/stb_image_write.c)
+target_link_libraries(normalmap PRIVATE m)
 
 add_executable(normalmap ${SOURCE_FILES})
\ No newline at end of file

From acd41404ab04298cd932bb72e92b5e5e6d7f2c94 Mon Sep 17 00:00:00 2001
From: VLT Media <info@vltmedia.com>
Date: Mon, 7 Jun 2021 18:02:12 +0000
Subject: [PATCH 07/66] tryin gto fix linker

---
 CMakeLists.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index d260308..1b5bab1 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -13,6 +13,6 @@ set(SOURCE_FILES
         src/stb_image.c
         src/normalmap.c
         include/normalmap.h include/stb_image_write.h src/stb_image_write.c)
-target_link_libraries(normalmap PRIVATE m)
 
-add_executable(normalmap ${SOURCE_FILES})
\ No newline at end of file
+add_executable(normalmap ${SOURCE_FILES})
+target_link_libraries(normalmap PRIVATE m)

From dcec29b85f6084ddd21bb6ea29c8c84eae9679b3 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 14:36:39 -0400
Subject: [PATCH 08/66] Added Buildcheck

Added Buildcheck
---
 .github/workflows/cmake.yml | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index e282e38..563118a 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -36,3 +36,10 @@ jobs:
       # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
       run: ctest -C ${{env.BUILD_TYPE}}
       
+    
+    - name: CheckBuildFolder
+      working-directory: ${{github.workspace}}/build
+      # Execute tests defined by the CMake configuration.  
+      # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
+      run: ls
+

From 505e92ef518e21c9f2f4883346c55d25b51ad578 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 14:42:18 -0400
Subject: [PATCH 09/66] Test Normalmap Exec

Test Normalmap Exec
---
 .github/workflows/cmake.yml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 563118a..1d21a21 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -30,11 +30,11 @@ jobs:
       # Build your program with the given configuration
       run: cmake --build ${{github.workspace}}/build --config ${{env.BUILD_TYPE}}
 
-    - name: Test
+    - name: Test Normalmap Exec
       working-directory: ${{github.workspace}}/build
       # Execute tests defined by the CMake configuration.  
       # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
-      run: ctest -C ${{env.BUILD_TYPE}}
+      run: ${{github.workspace}}/build/normalmap -h
       
     
     - name: CheckBuildFolder

From a47f55fe9e00f60bb9f54d3b27cb0849376ef0c2 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:03:29 -0400
Subject: [PATCH 10/66] added release linux

added release linux
---
 .github/workflows/cmake.yml | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 1d21a21..8b2d1b4 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -43,3 +43,12 @@ jobs:
       # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
       run: ls
 
+    - name: Release Linux
+      uses: softprops/action-gh-release@v1
+      if: startsWith(github.ref, 'refs/tags/')
+      with:
+        files: |
+            ${{github.workspace}}/build/normalmap
+            ${{github.workspace}}/README.md
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

From 37c3da41c68d76b880e2c7718164b8aa0fa2a353 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:04:07 -0400
Subject: [PATCH 11/66] removed normalmap check

removed normalmap check
---
 .github/workflows/cmake.yml | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 8b2d1b4..7380f5c 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -30,13 +30,7 @@ jobs:
       # Build your program with the given configuration
       run: cmake --build ${{github.workspace}}/build --config ${{env.BUILD_TYPE}}
 
-    - name: Test Normalmap Exec
-      working-directory: ${{github.workspace}}/build
-      # Execute tests defined by the CMake configuration.  
-      # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
-      run: ${{github.workspace}}/build/normalmap -h
-      
-    
+
     - name: CheckBuildFolder
       working-directory: ${{github.workspace}}/build
       # Execute tests defined by the CMake configuration.  

From a8cc3fffacdaa05dcf43dd79ac8ec8f7e7ec753a Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:07:05 -0400
Subject: [PATCH 12/66] Changed Release linux to another

Changed Release linux to another
---
 .github/workflows/cmake.yml | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 7380f5c..5db9d12 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -37,12 +37,15 @@ jobs:
       # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
       run: ls
 
+
     - name: Release Linux
-      uses: softprops/action-gh-release@v1
-      if: startsWith(github.ref, 'refs/tags/')
+      uses: "marvinpinto/action-automatic-releases@latest"
       with:
+        repo_token: "${{ secrets.GITHUB_TOKEN }}"
+        automatic_release_tag: "latest"
+        prerelease: true
+        title: "Development Build"
         files: |
-            ${{github.workspace}}/build/normalmap
-            ${{github.workspace}}/README.md
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          ${{github.workspace}}/build/normalmap
+          ${{github.workspace}}/README.md
+         

From 8c8c41a65a700a40281d105a77429af094eec5cd Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:15:30 -0400
Subject: [PATCH 13/66] added naming change

---
 .github/workflows/cmake.yml | 18 +++++++++++++++---
 1 file changed, 15 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 5db9d12..2eee95c 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -36,15 +36,27 @@ jobs:
       # Execute tests defined by the CMake configuration.  
       # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
       run: ls
-
-
+      
+    - name: Extract tag name
+      id: tag
+      uses: actions/github-script@0.2.0
+      with:
+        github-token: ${{ secrets.GITHUB_TOKEN }}
+        script: |
+          return context.payload.ref.replace(/\/refs\/tags\//, '');
+          
+    - name: Extract branch name
+      shell: bash
+      run: echo "##[set-output name=branch;]$(echo ${GITHUB_REF#refs/heads/})"
+      id: extract_branch
+  
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: "Development Build"
+        title: ${{ steps.tag.outputs.result + steps.tag.outputs.result }}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From b864629c394da5d7438f8fad8d20bfc26311a51a Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:16:31 -0400
Subject: [PATCH 14/66] Updated title

---
 .github/workflows/cmake.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 2eee95c..d006c53 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -56,7 +56,7 @@ jobs:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: ${{ steps.tag.outputs.result + steps.tag.outputs.result }}
+        title: ${{ steps.tag.outputs.result }}${{steps.tag.outputs.result }}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From 76a0136bb6cf484968c6bc6d0dc11e0397bc2822 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:19:06 -0400
Subject: [PATCH 15/66] added new getbranch action

added new getbranch action
---
 .github/workflows/cmake.yml | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index d006c53..f2c46a1 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -37,18 +37,11 @@ jobs:
       # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
       run: ls
       
-    - name: Extract tag name
-      id: tag
-      uses: actions/github-script@0.2.0
-      with:
-        github-token: ${{ secrets.GITHUB_TOKEN }}
-        script: |
-          return context.payload.ref.replace(/\/refs\/tags\//, '');
-          
-    - name: Extract branch name
-      shell: bash
-      run: echo "##[set-output name=branch;]$(echo ${GITHUB_REF#refs/heads/})"
-      id: extract_branch
+    - name: Use action
+      id: get-branch
+      uses: ypicard/get-branch-name-github-action@v1
+    - name: Echo outpus
+      run: echo "name=${{ steps.get-branch.outputs.branch }} - tag=${{ steps.get-branch.outputs.tag }}"
   
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"
@@ -56,7 +49,7 @@ jobs:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: ${{ steps.tag.outputs.result }}${{steps.tag.outputs.result }}
+        title: ${{ steps.get-branch.outputs.name }}${{steps.get-branch.outputs.tag }}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From c58aed2872c03858f8076b2d24682a49ac0ca073 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:25:29 -0400
Subject: [PATCH 16/66] changed get Branch Info

---
 .github/workflows/cmake.yml | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index f2c46a1..71cebeb 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -37,19 +37,17 @@ jobs:
       # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
       run: ls
       
-    - name: Use action
-      id: get-branch
-      uses: ypicard/get-branch-name-github-action@v1
-    - name: Echo outpus
-      run: echo "name=${{ steps.get-branch.outputs.branch }} - tag=${{ steps.get-branch.outputs.tag }}"
-  
+    - name: Get branch names
+      id: branch-name
+      uses: tj-actions/branch-names@v4.3
+        
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: ${{ steps.get-branch.outputs.name }}${{steps.get-branch.outputs.tag }}
+        title: ${{ steps.branch-name.outputs.current_branch}}${{steps.branch-name.outputs.tag}}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From 0be0f1f0ea8bc117fd5e9e83bc8855375962ebbe Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:27:51 -0400
Subject: [PATCH 17/66] Updated Tags release

---
 .github/workflows/cmake.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 71cebeb..9280067 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -47,7 +47,7 @@ jobs:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: ${{ steps.branch-name.outputs.current_branch}}${{steps.branch-name.outputs.tag}}
+        title: ${{ steps.branch-name.outputs.current_branch}{steps.branch-name.outputs.tag}}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From 4d46fec15d807c32e37f50fe71a19236b23ff005 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:28:33 -0400
Subject: [PATCH 18/66] Update cmake.yml

---
 .github/workflows/cmake.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 9280067..1ce41cb 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -47,7 +47,7 @@ jobs:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: ${{ steps.branch-name.outputs.current_branch}{steps.branch-name.outputs.tag}}
+        title: ${{ steps.branch-name.outputs.tag}}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From 64f89b97f8892c9a4c224158b574dcaa5df6a0b0 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:38:21 -0400
Subject: [PATCH 19/66] added tag combine

---
 .github/workflows/cmake.yml | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 1ce41cb..dc186d0 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -40,6 +40,9 @@ jobs:
     - name: Get branch names
       id: branch-name
       uses: tj-actions/branch-names@v4.3
+      
+    - name: Set env
+      run: echo ::set-env name=STEPVAR::${{ steps.branch-name.outputs.tag }}-${{ steps.branch-name.outputs.current_branch }}
         
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"
@@ -47,7 +50,7 @@ jobs:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: ${{ steps.branch-name.outputs.tag}}
+        title: ${{STEPVAR}}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From 33f37c936e7b2e02b85b45d0a8e9c0bf72b26bab Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:39:22 -0400
Subject: [PATCH 20/66] added env.

---
 .github/workflows/cmake.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index dc186d0..b8186b8 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -50,7 +50,7 @@ jobs:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: ${{STEPVAR}}
+        title: ${{env.STEPVAR}}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From fa6e4a7bf74d2feb728ecd3b0882cc9a7e2f2487 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:47:26 -0400
Subject: [PATCH 21/66] updated set envname

---
 .github/workflows/cmake.yml | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index b8186b8..b81b57e 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -42,7 +42,8 @@ jobs:
       uses: tj-actions/branch-names@v4.3
       
     - name: Set env
-      run: echo ::set-env name=STEPVAR::${{ steps.branch-name.outputs.tag }}-${{ steps.branch-name.outputs.current_branch }}
+      id: envname
+      run: echo ::set-output name=pr_title::"release candidate - ${{ steps.branch-name.outputs.tag }}-${{ steps.branch-name.outputs.current_branch }}"
         
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"
@@ -50,7 +51,7 @@ jobs:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "latest"
         prerelease: true
-        title: ${{env.STEPVAR}}
+        title: ${{steps.envname.outputs.pr_title}}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From e6d46c31dfc37c7142110159c747a06bb5ce425e Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:50:12 -0400
Subject: [PATCH 22/66] updated linux build env

---
 .github/workflows/cmake.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index b81b57e..1edef8b 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -43,7 +43,7 @@ jobs:
       
     - name: Set env
       id: envname
-      run: echo ::set-output name=pr_title::"release candidate - ${{ steps.branch-name.outputs.tag }}-${{ steps.branch-name.outputs.current_branch }}"
+      run: echo ::set-output name=pr_title::"Linux build - ${GITHUB_REF#refs/*/}-${{ steps.branch-name.outputs.current_branch }}"
         
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"

From 1da7dd90e0a514949e39d52e578c7b085fc43c6c Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:54:30 -0400
Subject: [PATCH 23/66] added new tag reader

---
 .github/workflows/cmake.yml | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 1edef8b..2fc03d3 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -37,13 +37,19 @@ jobs:
       # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
       run: ls
       
+    - uses: little-core-labs/get-git-tag@v3.0.1
+      id: tagName
+      with:
+        tagRegex: "v*"  # Optional. Returns specified group text as tag name. Full tag string is returned if regex is not defined.
+        tagRegexGroup: 1 # Optional. Default is 1.
+          
     - name: Get branch names
       id: branch-name
       uses: tj-actions/branch-names@v4.3
       
     - name: Set env
       id: envname
-      run: echo ::set-output name=pr_title::"Linux build - ${GITHUB_REF#refs/*/}-${{ steps.branch-name.outputs.current_branch }}"
+      run: echo ::set-output name=pr_title::"Linux build - ${{ steps.tagName.outputs.tag}}-${{ steps.branch-name.outputs.current_branch }}"
         
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"

From c671077c6b8689481ade85471a5321ba9c80c24e Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:55:44 -0400
Subject: [PATCH 24/66] removed automatic release tag

---
 .github/workflows/cmake.yml | 1 -
 1 file changed, 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 2fc03d3..c6814b0 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -55,7 +55,6 @@ jobs:
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
-        automatic_release_tag: "latest"
         prerelease: true
         title: ${{steps.envname.outputs.pr_title}}
         files: |

From 69dcc6296be65abce329ca5eb98bbf4b1ead5fd8 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 15:59:25 -0400
Subject: [PATCH 25/66] added automatic release tag

added automatic release tag
---
 .github/workflows/cmake.yml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index c6814b0..0f773eb 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -55,6 +55,7 @@ jobs:
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
+        automatic_release_tag: "v2.0.0"
         prerelease: true
         title: ${{steps.envname.outputs.pr_title}}
         files: |

From 0b1c83652bbcde4b61dc85a9751ab88854cc7b4e Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 16:06:28 -0400
Subject: [PATCH 26/66] added new tag name

---
 .github/workflows/cmake.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 0f773eb..8016133 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -49,7 +49,7 @@ jobs:
       
     - name: Set env
       id: envname
-      run: echo ::set-output name=pr_title::"Linux build - ${{ steps.tagName.outputs.tag}}-${{ steps.branch-name.outputs.current_branch }}"
+      run: echo ::set-output name=pr_title::"Linux build - ${{ github.event.release.tag_name }} - ${{ steps.branch-name.outputs.current_branch }}"
         
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"

From a3920d5947298d14fed6d382dcc66696f08750c3 Mon Sep 17 00:00:00 2001
From: "info@vltmedia.com" <info@vltmedia.com>
Date: Mon, 7 Jun 2021 16:22:55 -0400
Subject: [PATCH 27/66] updated formatting in readme

---
 .vscode/settings.json | 8 ++++++++
 README.md             | 8 ++++++++
 2 files changed, 16 insertions(+)
 create mode 100644 .vscode/settings.json

diff --git a/.vscode/settings.json b/.vscode/settings.json
new file mode 100644
index 0000000..e57cf3e
--- /dev/null
+++ b/.vscode/settings.json
@@ -0,0 +1,8 @@
+{
+    "workbench.colorCustomizations": {
+        "activityBar.background": "#531C02",
+        "titleBar.activeBackground": "#742803",
+        "titleBar.activeForeground": "#FFFBFA"
+    },
+    "window.title": "${dirty}${activeEditorShort}${separator}${rootName}${separator}${appName} ${separator} [Branch: master]"
+}
\ No newline at end of file
diff --git a/README.md b/README.md
index e803044..bb6d9f9 100644
--- a/README.md
+++ b/README.md
@@ -33,3 +33,11 @@ The underlying normal map generation code is inherited from The GIMP and is cons
   * true "normalized" normal
 
 Note that the effect of du/dv output being unnormalized is that normal vectors constructed by sampling the map will require normalization in your shader or image loader, which means in turn that the effect of the scale parameter above is less pronounced than it would be otherwise.
+
+
+
+
+
+
+
+

From fa49e83f27df8cfbb7957e0a84cd5bf2ca996e0f Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 16:28:45 -0400
Subject: [PATCH 28/66] added GetTagCheck

---
 .github/workflows/GetTagCheck.yml | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)
 create mode 100644 .github/workflows/GetTagCheck.yml

diff --git a/.github/workflows/GetTagCheck.yml b/.github/workflows/GetTagCheck.yml
new file mode 100644
index 0000000..eec1619
--- /dev/null
+++ b/.github/workflows/GetTagCheck.yml
@@ -0,0 +1,30 @@
+name: CMake
+
+on:
+  push:
+    branches: [ master ]
+  pull_request:
+    branches: [ master ]
+
+env:
+  # Customize the CMake build type here (Release, Debug, RelWithDebInfo, etc.)
+  BUILD_TYPE: Release
+
+jobs:
+  build:
+    # The CMake configure and build commands are platform agnostic and should work equally
+    # well on Windows or Mac.  You can convert this to a matrix build if you need
+    # cross-platform coverage.
+    # See: https://docs.github.com/en/free-pro-team@latest/actions/learn-github-actions/managing-complex-workflows#using-a-build-matrix
+    runs-on: ubuntu-latest
+
+    steps:
+    - name: Get tag
+  # You may pin to the exact commit or the version.
+  # uses: dawidd6/action-get-tag@66f78e9796155499bd32b3bc4993b793c34096be
+      uses: dawidd6/action-get-tag@v1.0.0
+      
+    - name: Use tag
+      run: echo ${{steps.tag.outputs.tag}}
+      
+      

From 5cef22dda8e655c30a5a50797cdf0648b11a525f Mon Sep 17 00:00:00 2001
From: "info@vltmedia.com" <info@vltmedia.com>
Date: Mon, 7 Jun 2021 16:29:46 -0400
Subject: [PATCH 29/66] added How To Update

---
 README.md | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/README.md b/README.md
index bb6d9f9..fa48999 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,5 @@
 # normalmap
+
 Command line utility to convert height maps to DU-DV normal maps.
 
 ## Platforms
@@ -15,6 +16,7 @@ The height will be taken from the red channel of the source image. RGBA PNGs are
 
 The output will be an unfiltered, 8-bit unsigned du/dv map, which can be used as a normal map in rendering applications. The heightmap sampler is a 2x2 box.
 
+## How to Update
 The underlying normal map generation code is inherited from The GIMP and is considerably more flexible. In this version, you must update `main.c` and rebuild to change the following settings:
 
 * Scale *default = 2.0*

From 135a3bc0d98b45e9de413daaf3a2dc154725b2ab Mon Sep 17 00:00:00 2001
From: "info@vltmedia.com" <info@vltmedia.com>
Date: Mon, 7 Jun 2021 16:31:25 -0400
Subject: [PATCH 30/66] Revert "added How To Update"

This reverts commit 5cef22dda8e655c30a5a50797cdf0648b11a525f.
---
 README.md | 2 --
 1 file changed, 2 deletions(-)

diff --git a/README.md b/README.md
index fa48999..bb6d9f9 100644
--- a/README.md
+++ b/README.md
@@ -1,5 +1,4 @@
 # normalmap
-
 Command line utility to convert height maps to DU-DV normal maps.
 
 ## Platforms
@@ -16,7 +15,6 @@ The height will be taken from the red channel of the source image. RGBA PNGs are
 
 The output will be an unfiltered, 8-bit unsigned du/dv map, which can be used as a normal map in rendering applications. The heightmap sampler is a 2x2 box.
 
-## How to Update
 The underlying normal map generation code is inherited from The GIMP and is considerably more flexible. In this version, you must update `main.c` and rebuild to change the following settings:
 
 * Scale *default = 2.0*

From 7dd88e62cf5d9b46c2be5f84d1e3c3144ac62ae4 Mon Sep 17 00:00:00 2001
From: "info@vltmedia.com" <info@vltmedia.com>
Date: Mon, 7 Jun 2021 16:34:49 -0400
Subject: [PATCH 31/66] add How to Update

---
 README.md | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/README.md b/README.md
index bb6d9f9..fa48999 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,5 @@
 # normalmap
+
 Command line utility to convert height maps to DU-DV normal maps.
 
 ## Platforms
@@ -15,6 +16,7 @@ The height will be taken from the red channel of the source image. RGBA PNGs are
 
 The output will be an unfiltered, 8-bit unsigned du/dv map, which can be used as a normal map in rendering applications. The heightmap sampler is a 2x2 box.
 
+## How to Update
 The underlying normal map generation code is inherited from The GIMP and is considerably more flexible. In this version, you must update `main.c` and rebuild to change the following settings:
 
 * Scale *default = 2.0*

From 24f918f359ec32fc5a4c2c85fff46e800347fd8f Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 16:36:38 -0400
Subject: [PATCH 32/66] updated hardcoded v0.2.0 tag

---
 .github/workflows/cmake.yml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 8016133..e4df81b 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -49,13 +49,13 @@ jobs:
       
     - name: Set env
       id: envname
-      run: echo ::set-output name=pr_title::"Linux build - ${{ github.event.release.tag_name }} - ${{ steps.branch-name.outputs.current_branch }}"
+      run: echo ::set-output name=pr_title::"Linux build - v0.2.0 - ${{ steps.branch-name.outputs.current_branch }}"
         
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
-        automatic_release_tag: "v2.0.0"
+        automatic_release_tag: "v0.2.0"
         prerelease: true
         title: ${{steps.envname.outputs.pr_title}}
         files: |

From 108a4cfa187846fbb62cc945b7463e872958de80 Mon Sep 17 00:00:00 2001
From: "info@vltmedia.com" <info@vltmedia.com>
Date: Mon, 7 Jun 2021 16:37:09 -0400
Subject: [PATCH 33/66] Deleted uneeded Github Actions

---
 .github/workflows/GetTagCheck.yml | 30 -----------------------------
 .github/workflows/c-cpp.yml       | 23 ----------------------
 .github/workflows/msbuild.yml     | 32 -------------------------------
 3 files changed, 85 deletions(-)
 delete mode 100644 .github/workflows/GetTagCheck.yml
 delete mode 100644 .github/workflows/c-cpp.yml
 delete mode 100644 .github/workflows/msbuild.yml

diff --git a/.github/workflows/GetTagCheck.yml b/.github/workflows/GetTagCheck.yml
deleted file mode 100644
index eec1619..0000000
--- a/.github/workflows/GetTagCheck.yml
+++ /dev/null
@@ -1,30 +0,0 @@
-name: CMake
-
-on:
-  push:
-    branches: [ master ]
-  pull_request:
-    branches: [ master ]
-
-env:
-  # Customize the CMake build type here (Release, Debug, RelWithDebInfo, etc.)
-  BUILD_TYPE: Release
-
-jobs:
-  build:
-    # The CMake configure and build commands are platform agnostic and should work equally
-    # well on Windows or Mac.  You can convert this to a matrix build if you need
-    # cross-platform coverage.
-    # See: https://docs.github.com/en/free-pro-team@latest/actions/learn-github-actions/managing-complex-workflows#using-a-build-matrix
-    runs-on: ubuntu-latest
-
-    steps:
-    - name: Get tag
-  # You may pin to the exact commit or the version.
-  # uses: dawidd6/action-get-tag@66f78e9796155499bd32b3bc4993b793c34096be
-      uses: dawidd6/action-get-tag@v1.0.0
-      
-    - name: Use tag
-      run: echo ${{steps.tag.outputs.tag}}
-      
-      
diff --git a/.github/workflows/c-cpp.yml b/.github/workflows/c-cpp.yml
deleted file mode 100644
index e323326..0000000
--- a/.github/workflows/c-cpp.yml
+++ /dev/null
@@ -1,23 +0,0 @@
-name: C/C++ CI
-
-on:
-  push:
-    branches: [ master ]
-  pull_request:
-    branches: [ master ]
-
-jobs:
-  build:
-
-    runs-on: ubuntu-latest
-
-    steps:
-    - uses: actions/checkout@v2
-    - name: configure
-      run: ./configure
-    - name: make
-      run: make
-    - name: make check
-      run: make check
-    - name: make distcheck
-      run: make distcheck
diff --git a/.github/workflows/msbuild.yml b/.github/workflows/msbuild.yml
deleted file mode 100644
index 29b6ace..0000000
--- a/.github/workflows/msbuild.yml
+++ /dev/null
@@ -1,32 +0,0 @@
-name: MSBuild
-
-on: [push]
-
-env:
-  # Path to the solution file relative to the root of the project.
-  SOLUTION_FILE_PATH: .
-
-  # Configuration type to build.
-  # You can convert this to a build matrix if you need coverage of multiple configuration types.
-  # https://docs.github.com/actions/learn-github-actions/managing-complex-workflows#using-a-build-matrix
-  BUILD_CONFIGURATION: Release
-
-jobs:
-  build:
-    runs-on: windows-latest
-
-    steps:
-    - uses: actions/checkout@v2
-
-    - name: Add MSBuild to PATH
-      uses: microsoft/setup-msbuild@v1.0.2
-
-    - name: Restore NuGet packages
-      working-directory: ${{env.GITHUB_WORKSPACE}}
-      run: nuget restore ${{env.SOLUTION_FILE_PATH}}
-
-    - name: Build
-      working-directory: ${{env.GITHUB_WORKSPACE}}
-      # Add additional options to the MSBuild command line here (like platform or verbosity level).
-      # See https://docs.microsoft.com/visualstudio/msbuild/msbuild-command-line-reference
-      run: msbuild /m /p:Configuration=${{env.BUILD_CONFIGURATION}} ${{env.SOLUTION_FILE_PATH}}

From ad60c8344542a0a5a69ac887edd3cc74d2156adb Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 16:42:32 -0400
Subject: [PATCH 34/66] added Windows build

added Windows build
---
 .github/workflows/cmake_windows.yml | 58 +++++++++++++++++++++++++++++
 1 file changed, 58 insertions(+)
 create mode 100644 .github/workflows/cmake_windows.yml

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
new file mode 100644
index 0000000..291b3e9
--- /dev/null
+++ b/.github/workflows/cmake_windows.yml
@@ -0,0 +1,58 @@
+name: CMake
+
+on:
+  push:
+    branches: [ master ]
+  pull_request:
+    branches: [ master ]
+
+env:
+  # Customize the CMake build type here (Release, Debug, RelWithDebInfo, etc.)
+  BUILD_TYPE: Release
+
+jobs:
+  build:
+    # The CMake configure and build commands are platform agnostic and should work equally
+    # well on Windows or Mac.  You can convert this to a matrix build if you need
+    # cross-platform coverage.
+    # See: https://docs.github.com/en/free-pro-team@latest/actions/learn-github-actions/managing-complex-workflows#using-a-build-matrix
+    runs-on: windows-latest
+
+    steps:
+    - uses: actions/checkout@v2
+
+    - name: Configure CMake
+      # Configure CMake in a 'build' subdirectory. `CMAKE_BUILD_TYPE` is only required if you are using a single-configuration generator such as make.
+      # See https://cmake.org/cmake/help/latest/variable/CMAKE_BUILD_TYPE.html?highlight=cmake_build_type
+      run: cmake -B ${{github.workspace}}/build -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}}
+
+    - name: Build
+      # Build your program with the given configuration
+      run: cmake --build ${{github.workspace}}/build --config ${{env.BUILD_TYPE}}
+
+
+    - name: CheckBuildFolder
+      working-directory: ${{github.workspace}}/build
+      # Execute tests defined by the CMake configuration.  
+      # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
+      run: ls
+          
+    - name: Get branch names
+      id: branch-name
+      uses: tj-actions/branch-names@v4.3
+      
+    - name: Set env
+      id: envname
+      run: echo ::set-output name=pr_title::"Windows build - v0.2.0 - ${{ steps.branch-name.outputs.current_branch }}"
+        
+    - name: Release Linux
+      uses: "marvinpinto/action-automatic-releases@latest"
+      with:
+        repo_token: "${{ secrets.GITHUB_TOKEN }}"
+        automatic_release_tag: "v0.2.0"
+        prerelease: true
+        title: ${{steps.envname.outputs.pr_title}}
+        files: |
+          ${{github.workspace}}/build/normalmap
+          ${{github.workspace}}/README.md
+         

From 5fe1371182af61aef9ee23bd5067f96322705559 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 16:43:34 -0400
Subject: [PATCH 35/66] updated naming on the github action

---
 .github/workflows/cmake_windows.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 291b3e9..25c90d4 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -1,4 +1,4 @@
-name: CMake
+name: CMake_Windows
 
 on:
   push:

From f39483ae10296d0a1355737a4e4d8dda1019ee53 Mon Sep 17 00:00:00 2001
From: "info@vltmedia.com" <info@vltmedia.com>
Date: Mon, 7 Jun 2021 17:14:17 -0400
Subject: [PATCH 36/66] Added Windows src folder and fixed the struct for
 windows usage

---
 .gitignore                      |    2 +
 CMakeLists.txt                  |    2 +-
 win64/CMakeLists.txt            |   18 +
 win64/include/mem.h             |  362 ++
 win64/include/normalmap.h       |   89 +
 win64/include/stb_image.h       | 7099 +++++++++++++++++++++++++++++++
 win64/include/stb_image_write.h | 1048 +++++
 win64/src/main.c                |   77 +
 win64/src/normalmap.c           |  971 +++++
 win64/src/stb_image.c           |    2 +
 win64/src/stb_image_write.c     |    2 +
 11 files changed, 9671 insertions(+), 1 deletion(-)
 create mode 100644 win64/CMakeLists.txt
 create mode 100644 win64/include/mem.h
 create mode 100644 win64/include/normalmap.h
 create mode 100644 win64/include/stb_image.h
 create mode 100644 win64/include/stb_image_write.h
 create mode 100644 win64/src/main.c
 create mode 100644 win64/src/normalmap.c
 create mode 100644 win64/src/stb_image.c
 create mode 100644 win64/src/stb_image_write.c

diff --git a/.gitignore b/.gitignore
index 123b13b..bf99a55 100644
--- a/.gitignore
+++ b/.gitignore
@@ -34,6 +34,8 @@
 
 # CMake
 cmake-build-*/
+build/
+
 
 # IDE
 .idea/
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 1b5bab1..eb84658 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -15,4 +15,4 @@ set(SOURCE_FILES
         include/normalmap.h include/stb_image_write.h src/stb_image_write.c)
 
 add_executable(normalmap ${SOURCE_FILES})
-target_link_libraries(normalmap PRIVATE m)
+# target_link_libraries(normalmap PRIVATE m)
diff --git a/win64/CMakeLists.txt b/win64/CMakeLists.txt
new file mode 100644
index 0000000..eb84658
--- /dev/null
+++ b/win64/CMakeLists.txt
@@ -0,0 +1,18 @@
+cmake_minimum_required(VERSION 3.6)
+project(normalmap)
+
+set(CMAKE_C_STANDARD  11)
+set(CMAKE_CXX_STANDARD 11)
+
+include_directories(include)
+
+set(SOURCE_FILES
+        include/stb_image.h
+        include/mem.h
+        src/main.c
+        src/stb_image.c
+        src/normalmap.c
+        include/normalmap.h include/stb_image_write.h src/stb_image_write.c)
+
+add_executable(normalmap ${SOURCE_FILES})
+# target_link_libraries(normalmap PRIVATE m)
diff --git a/win64/include/mem.h b/win64/include/mem.h
new file mode 100644
index 0000000..3ecb601
--- /dev/null
+++ b/win64/include/mem.h
@@ -0,0 +1,362 @@
+/*
+ * Copyright (c) 2013-present, Yann Collet
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+#ifndef MEM_H_MODULE
+#define MEM_H_MODULE
+
+#if defined (__cplusplus)
+extern "C" {
+#endif
+
+/*-****************************************
+*  Dependencies
+******************************************/
+#include <stddef.h>     /* size_t, ptrdiff_t */
+#include <string.h>     /* memcpy */
+
+
+/*-****************************************
+*  Compiler specifics
+******************************************/
+#if defined(_MSC_VER)   /* Visual Studio */
+#   include <stdlib.h>  /* _byteswap_ulong */
+#   include <intrin.h>  /* _byteswap_* */
+#endif
+#if defined(__GNUC__)
+#  define MEM_STATIC static __inline __attribute__((unused))
+#elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
+#  define MEM_STATIC static inline
+#elif defined(_MSC_VER)
+#  define MEM_STATIC static __inline
+#else
+#  define MEM_STATIC static  /* this version may generate warnings for unused static functions; disable the relevant warning */
+#endif
+
+/* code only tested on 32 and 64 bits systems */
+#define MEM_STATIC_ASSERT(c)   { enum { MEM_static_assert = 1/(int)(!!(c)) }; }
+MEM_STATIC void MEM_check(void) { MEM_STATIC_ASSERT((sizeof(size_t)==4) || (sizeof(size_t)==8)); }
+
+
+/*-**************************************************************
+*  Basic Types
+*****************************************************************/
+#if  !defined (__VMS) && (defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
+# include <stdint.h>
+  typedef   uint8_t BYTE;
+  typedef  uint16_t U16;
+  typedef   int16_t S16;
+  typedef  uint32_t U32;
+  typedef   int32_t S32;
+  typedef  uint64_t U64;
+  typedef   int64_t S64;
+#else
+  typedef unsigned char      BYTE;
+  typedef unsigned short      U16;
+  typedef   signed short      S16;
+  typedef unsigned int        U32;
+  typedef   signed int        S32;
+  typedef unsigned long long  U64;
+  typedef   signed long long  S64;
+#endif
+
+
+/*-**************************************************************
+*  Memory I/O
+*****************************************************************/
+/* MEM_FORCE_MEMORY_ACCESS :
+ * By default, access to unaligned memory is controlled by `memcpy()`, which is safe and portable.
+ * Unfortunately, on some target/compiler combinations, the generated assembly is sub-optimal.
+ * The below switch allow to select different access method for improved performance.
+ * Method 0 (default) : use `memcpy()`. Safe and portable.
+ * Method 1 : `__packed` statement. It depends on compiler extension (i.e., not portable).
+ *            This method is safe if your compiler supports it, and *generally* as fast or faster than `memcpy`.
+ * Method 2 : direct access. This method is portable but violate C standard.
+ *            It can generate buggy code on targets depending on alignment.
+ *            In some circumstances, it's the only known way to get the most performance (i.e. GCC + ARMv6)
+ * See http://fastcompression.blogspot.fr/2015/08/accessing-unaligned-memory.html for details.
+ * Prefer these methods in priority order (0 > 1 > 2)
+ */
+#ifndef MEM_FORCE_MEMORY_ACCESS   /* can be defined externally, on command line for example */
+#  if defined(__GNUC__) && ( defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) || defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6Z__) || defined(__ARM_ARCH_6ZK__) || defined(__ARM_ARCH_6T2__) )
+#    define MEM_FORCE_MEMORY_ACCESS 2
+#  elif defined(__INTEL_COMPILER) || defined(__GNUC__)
+#    define MEM_FORCE_MEMORY_ACCESS 1
+#  endif
+#endif
+
+MEM_STATIC unsigned MEM_32bits(void) { return sizeof(size_t)==4; }
+MEM_STATIC unsigned MEM_64bits(void) { return sizeof(size_t)==8; }
+
+MEM_STATIC unsigned MEM_isLittleEndian(void)
+{
+    const union { U32 u; BYTE c[4]; } one = { 1 };   /* don't use static : performance detrimental  */
+    return one.c[0];
+}
+
+#if defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==2)
+
+/* violates C standard, by lying on structure alignment.
+Only use if no other choice to achieve best performance on target platform */
+MEM_STATIC U16 MEM_read16(const void* memPtr) { return *(const U16*) memPtr; }
+MEM_STATIC U32 MEM_read32(const void* memPtr) { return *(const U32*) memPtr; }
+MEM_STATIC U64 MEM_read64(const void* memPtr) { return *(const U64*) memPtr; }
+MEM_STATIC size_t MEM_readST(const void* memPtr) { return *(const size_t*) memPtr; }
+
+MEM_STATIC void MEM_write16(void* memPtr, U16 value) { *(U16*)memPtr = value; }
+MEM_STATIC void MEM_write32(void* memPtr, U32 value) { *(U32*)memPtr = value; }
+MEM_STATIC void MEM_write64(void* memPtr, U64 value) { *(U64*)memPtr = value; }
+
+#elif defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==1)
+
+/* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */
+/* currently only defined for gcc and icc */
+#if defined(_MSC_VER) || (defined(__INTEL_COMPILER) && defined(WIN32))
+    __pragma( pack(push, 1) )
+    typedef struct { U16 v; } unalign16;
+    typedef struct { U32 v; } unalign32;
+    typedef struct { U64 v; } unalign64;
+    typedef struct { size_t v; } unalignArch;
+    __pragma( pack(pop) )
+#else
+    typedef struct { U16 v; } __attribute__((packed)) unalign16;
+    typedef struct { U32 v; } __attribute__((packed)) unalign32;
+    typedef struct { U64 v; } __attribute__((packed)) unalign64;
+    typedef struct { size_t v; } __attribute__((packed)) unalignArch;
+#endif
+
+MEM_STATIC U16 MEM_read16(const void* ptr) { return ((const unalign16*)ptr)->v; }
+MEM_STATIC U32 MEM_read32(const void* ptr) { return ((const unalign32*)ptr)->v; }
+MEM_STATIC U64 MEM_read64(const void* ptr) { return ((const unalign64*)ptr)->v; }
+MEM_STATIC size_t MEM_readST(const void* ptr) { return ((const unalignArch*)ptr)->v; }
+
+MEM_STATIC void MEM_write16(void* memPtr, U16 value) { ((unalign16*)memPtr)->v = value; }
+MEM_STATIC void MEM_write32(void* memPtr, U32 value) { ((unalign32*)memPtr)->v = value; }
+MEM_STATIC void MEM_write64(void* memPtr, U64 value) { ((unalign64*)memPtr)->v = value; }
+
+#else
+
+/* default method, safe and standard.
+   can sometimes prove slower */
+
+MEM_STATIC U16 MEM_read16(const void* memPtr)
+{
+    U16 val; memcpy(&val, memPtr, sizeof(val)); return val;
+}
+
+MEM_STATIC U32 MEM_read32(const void* memPtr)
+{
+    U32 val; memcpy(&val, memPtr, sizeof(val)); return val;
+}
+
+MEM_STATIC U64 MEM_read64(const void* memPtr)
+{
+    U64 val; memcpy(&val, memPtr, sizeof(val)); return val;
+}
+
+MEM_STATIC size_t MEM_readST(const void* memPtr)
+{
+    size_t val; memcpy(&val, memPtr, sizeof(val)); return val;
+}
+
+MEM_STATIC void MEM_write16(void* memPtr, U16 value)
+{
+    memcpy(memPtr, &value, sizeof(value));
+}
+
+MEM_STATIC void MEM_write32(void* memPtr, U32 value)
+{
+    memcpy(memPtr, &value, sizeof(value));
+}
+
+MEM_STATIC void MEM_write64(void* memPtr, U64 value)
+{
+    memcpy(memPtr, &value, sizeof(value));
+}
+
+#endif /* MEM_FORCE_MEMORY_ACCESS */
+
+MEM_STATIC U32 MEM_swap32(U32 in)
+{
+#if defined(_MSC_VER)     /* Visual Studio */
+    return _byteswap_ulong(in);
+#elif defined (__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ >= 403)
+    return __builtin_bswap32(in);
+#else
+    return  ((in << 24) & 0xff000000 ) |
+            ((in <<  8) & 0x00ff0000 ) |
+            ((in >>  8) & 0x0000ff00 ) |
+            ((in >> 24) & 0x000000ff );
+#endif
+}
+
+MEM_STATIC U64 MEM_swap64(U64 in)
+{
+#if defined(_MSC_VER)     /* Visual Studio */
+    return _byteswap_uint64(in);
+#elif defined (__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ >= 403)
+    return __builtin_bswap64(in);
+#else
+    return  ((in << 56) & 0xff00000000000000ULL) |
+            ((in << 40) & 0x00ff000000000000ULL) |
+            ((in << 24) & 0x0000ff0000000000ULL) |
+            ((in << 8)  & 0x000000ff00000000ULL) |
+            ((in >> 8)  & 0x00000000ff000000ULL) |
+            ((in >> 24) & 0x0000000000ff0000ULL) |
+            ((in >> 40) & 0x000000000000ff00ULL) |
+            ((in >> 56) & 0x00000000000000ffULL);
+#endif
+}
+
+MEM_STATIC size_t MEM_swapST(size_t in)
+{
+    if (MEM_32bits())
+        return (size_t)MEM_swap32((U32)in);
+    else
+        return (size_t)MEM_swap64((U64)in);
+}
+
+/*=== Little endian r/w ===*/
+
+MEM_STATIC U16 MEM_readLE16(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_read16(memPtr);
+    else {
+        const BYTE* p = (const BYTE*)memPtr;
+        return (U16)(p[0] + (p[1]<<8));
+    }
+}
+
+MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val)
+{
+    if (MEM_isLittleEndian()) {
+        MEM_write16(memPtr, val);
+    } else {
+        BYTE* p = (BYTE*)memPtr;
+        p[0] = (BYTE)val;
+        p[1] = (BYTE)(val>>8);
+    }
+}
+
+MEM_STATIC U32 MEM_readLE24(const void* memPtr)
+{
+    return MEM_readLE16(memPtr) + (((const BYTE*)memPtr)[2] << 16);
+}
+
+MEM_STATIC void MEM_writeLE24(void* memPtr, U32 val)
+{
+    MEM_writeLE16(memPtr, (U16)val);
+    ((BYTE*)memPtr)[2] = (BYTE)(val>>16);
+}
+
+MEM_STATIC U32 MEM_readLE32(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_read32(memPtr);
+    else
+        return MEM_swap32(MEM_read32(memPtr));
+}
+
+MEM_STATIC void MEM_writeLE32(void* memPtr, U32 val32)
+{
+    if (MEM_isLittleEndian())
+        MEM_write32(memPtr, val32);
+    else
+        MEM_write32(memPtr, MEM_swap32(val32));
+}
+
+MEM_STATIC U64 MEM_readLE64(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_read64(memPtr);
+    else
+        return MEM_swap64(MEM_read64(memPtr));
+}
+
+MEM_STATIC void MEM_writeLE64(void* memPtr, U64 val64)
+{
+    if (MEM_isLittleEndian())
+        MEM_write64(memPtr, val64);
+    else
+        MEM_write64(memPtr, MEM_swap64(val64));
+}
+
+MEM_STATIC size_t MEM_readLEST(const void* memPtr)
+{
+    if (MEM_32bits())
+        return (size_t)MEM_readLE32(memPtr);
+    else
+        return (size_t)MEM_readLE64(memPtr);
+}
+
+MEM_STATIC void MEM_writeLEST(void* memPtr, size_t val)
+{
+    if (MEM_32bits())
+        MEM_writeLE32(memPtr, (U32)val);
+    else
+        MEM_writeLE64(memPtr, (U64)val);
+}
+
+/*=== Big endian r/w ===*/
+
+MEM_STATIC U32 MEM_readBE32(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_swap32(MEM_read32(memPtr));
+    else
+        return MEM_read32(memPtr);
+}
+
+MEM_STATIC void MEM_writeBE32(void* memPtr, U32 val32)
+{
+    if (MEM_isLittleEndian())
+        MEM_write32(memPtr, MEM_swap32(val32));
+    else
+        MEM_write32(memPtr, val32);
+}
+
+MEM_STATIC U64 MEM_readBE64(const void* memPtr)
+{
+    if (MEM_isLittleEndian())
+        return MEM_swap64(MEM_read64(memPtr));
+    else
+        return MEM_read64(memPtr);
+}
+
+MEM_STATIC void MEM_writeBE64(void* memPtr, U64 val64)
+{
+    if (MEM_isLittleEndian())
+        MEM_write64(memPtr, MEM_swap64(val64));
+    else
+        MEM_write64(memPtr, val64);
+}
+
+MEM_STATIC size_t MEM_readBEST(const void* memPtr)
+{
+    if (MEM_32bits())
+        return (size_t)MEM_readBE32(memPtr);
+    else
+        return (size_t)MEM_readBE64(memPtr);
+}
+
+MEM_STATIC void MEM_writeBEST(void* memPtr, size_t val)
+{
+    if (MEM_32bits())
+        MEM_writeBE32(memPtr, (U32)val);
+    else
+        MEM_writeBE64(memPtr, (U64)val);
+}
+
+
+#if defined (__cplusplus)
+}
+#endif
+
+#endif /* MEM_H_MODULE */
\ No newline at end of file
diff --git a/win64/include/normalmap.h b/win64/include/normalmap.h
new file mode 100644
index 0000000..dd5b477
--- /dev/null
+++ b/win64/include/normalmap.h
@@ -0,0 +1,89 @@
+/*
+   Copyright (C) 2002-2008 Shawn Kirst <skirst@insightbb.com>
+
+   This program is free software; you can redistribute it and/or
+   modify it under the terms of the GNU General Public
+   License as published by the Free Software Foundation; either
+   version 2 of the License, or (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program; see the file COPYING.  If not, write to
+   the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.
+*/
+#ifndef NORMAL_MAP
+#define NORMAL_MAP
+
+#include <stdint.h>
+
+#define ENUM_EACH(PARAM) PARAM,
+#define STR_EACH(PARAM) #PARAM,
+#define STR_EACH_SEP_WS(PARAM) #PARAM " "
+
+#define EACH_FILTER_TYPE(DO) \
+    DO(FILTER_NONE) \
+    DO(FILTER_SOBEL_3x3) \
+    DO(FILTER_SOBEL_5x5) \
+    DO(FILTER_PREWITT_3x3) \
+    DO(FILTER_PREWITT_5x5) \
+    DO(FILTER_3x3) \
+    DO(FILTER_5x5) \
+    DO(FILTER_7x7) \
+    DO(FILTER_9x9) \
+    DO(MAX_FILTER_TYPE)
+#define FILTER_TYPE_COUNT (10)
+
+enum FILTER_TYPE { EACH_FILTER_TYPE(ENUM_EACH) };
+
+static char * FILTER_TYPE_NAMES[] = {
+        EACH_FILTER_TYPE(STR_EACH)
+};
+
+enum ALPHA_TYPE
+{
+    ALPHA_NONE, ALPHA_HEIGHT, ALPHA_INVERSE_HEIGHT, ALPHA_ZERO, ALPHA_ONE,
+    ALPHA_INVERT, ALPHA_MAP, MAX_ALPHA_TYPE
+};
+
+enum CONVERSION_TYPE
+{
+    CONVERT_NONE, CONVERT_BIASED_RGB, CONVERT_RED, CONVERT_GREEN,
+    CONVERT_BLUE, CONVERT_MAX_RGB, CONVERT_MIN_RGB, CONVERT_COLORSPACE,
+    CONVERT_NORMALIZE_ONLY, CONVERT_DUDV_TO_NORMAL, CONVERT_HEIGHTMAP,
+    CONVERT_KEY_RGB,
+    MAX_CONVERSION_TYPE
+};
+
+enum DUDV_TYPE
+{
+    DUDV_NONE, DUDV_8BIT_SIGNED, DUDV_8BIT_UNSIGNED, DUDV_16BIT_SIGNED,
+    DUDV_16BIT_UNSIGNED,
+    MAX_DUDV_TYPE
+};
+
+struct NormalmapVals
+{
+    int filter;
+    double minz;
+    double scale;
+    int wrap;
+    int height_source;
+    int alpha;
+    int conversion;
+    int dudv;
+    int xinvert;
+    int yinvert;
+    int swapRGB;
+    double contrast;
+    int32_t alphamap_id;
+} nmapvals;
+
+int32_t normalmap(uint8_t * img, uint8_t * nm_img, size_t dx, size_t dy, struct NormalmapVals nmapvals);
+
+
+#endif
\ No newline at end of file
diff --git a/win64/include/stb_image.h b/win64/include/stb_image.h
new file mode 100644
index 0000000..afcb49b
--- /dev/null
+++ b/win64/include/stb_image.h
@@ -0,0 +1,7099 @@
+/* stb_image - v2.13 - public domain image loader - http://nothings.org/stb_image.h
+                                     no warranty implied; use at your own risk
+
+   Do this:
+      #define STB_IMAGE_IMPLEMENTATION
+   before you include this file in *one* C or C++ file to create the implementation.
+
+   // i.e. it should look like this:
+   #include ...
+   #include ...
+   #include ...
+   #define STB_IMAGE_IMPLEMENTATION
+   #include "stb_image.h"
+
+   You can #define STBI_ASSERT(x) before the #include to avoid using assert.h.
+   And #define STBI_MALLOC, STBI_REALLOC, and STBI_FREE to avoid using malloc,realloc,free
+
+
+   QUICK NOTES:
+      Primarily of interest to game developers and other people who can
+          avoid problematic images and only need the trivial interface
+
+      JPEG baseline & progressive (12 bpc/arithmetic not supported, same as stock IJG lib)
+      PNG 1/2/4/8-bit-per-channel (16 bpc not supported)
+
+      TGA (not sure what subset, if a subset)
+      BMP non-1bpp, non-RLE
+      PSD (composited view only, no extra channels, 8/16 bit-per-channel)
+
+      GIF (*comp always reports as 4-channel)
+      HDR (radiance rgbE format)
+      PIC (Softimage PIC)
+      PNM (PPM and PGM binary only)
+
+      Animated GIF still needs a proper API, but here's one way to do it:
+          http://gist.github.com/urraka/685d9a6340b26b830d49
+
+      - decode from memory or through FILE (define STBI_NO_STDIO to remove code)
+      - decode from arbitrary I/O callbacks
+      - SIMD acceleration on x86/x64 (SSE2) and ARM (NEON)
+
+   Full documentation under "DOCUMENTATION" below.
+
+
+   Revision 2.00 release notes:
+
+      - Progressive JPEG is now supported.
+
+      - PPM and PGM binary formats are now supported, thanks to Ken Miller.
+
+      - x86 platforms now make use of SSE2 SIMD instructions for
+        JPEG decoding, and ARM platforms can use NEON SIMD if requested.
+        This work was done by Fabian "ryg" Giesen. SSE2 is used by
+        default, but NEON must be enabled explicitly; see docs.
+
+        With other JPEG optimizations included in this version, we see
+        2x speedup on a JPEG on an x86 machine, and a 1.5x speedup
+        on a JPEG on an ARM machine, relative to previous versions of this
+        library. The same results will not obtain for all JPGs and for all
+        x86/ARM machines. (Note that progressive JPEGs are significantly
+        slower to decode than regular JPEGs.) This doesn't mean that this
+        is the fastest JPEG decoder in the land; rather, it brings it
+        closer to parity with standard libraries. If you want the fastest
+        decode, look elsewhere. (See "Philosophy" section of docs below.)
+
+        See final bullet items below for more info on SIMD.
+
+      - Added STBI_MALLOC, STBI_REALLOC, and STBI_FREE macros for replacing
+        the memory allocator. Unlike other STBI libraries, these macros don't
+        support a context parameter, so if you need to pass a context in to
+        the allocator, you'll have to store it in a global or a thread-local
+        variable.
+
+      - Split existing STBI_NO_HDR flag into two flags, STBI_NO_HDR and
+        STBI_NO_LINEAR.
+            STBI_NO_HDR:     suppress implementation of .hdr reader format
+            STBI_NO_LINEAR:  suppress high-dynamic-range light-linear float API
+
+      - You can suppress implementation of any of the decoders to reduce
+        your code footprint by #defining one or more of the following
+        symbols before creating the implementation.
+
+            STBI_NO_JPEG
+            STBI_NO_PNG
+            STBI_NO_BMP
+            STBI_NO_PSD
+            STBI_NO_TGA
+            STBI_NO_GIF
+            STBI_NO_HDR
+            STBI_NO_PIC
+            STBI_NO_PNM   (.ppm and .pgm)
+
+      - You can request *only* certain decoders and suppress all other ones
+        (this will be more forward-compatible, as addition of new decoders
+        doesn't require you to disable them explicitly):
+
+            STBI_ONLY_JPEG
+            STBI_ONLY_PNG
+            STBI_ONLY_BMP
+            STBI_ONLY_PSD
+            STBI_ONLY_TGA
+            STBI_ONLY_GIF
+            STBI_ONLY_HDR
+            STBI_ONLY_PIC
+            STBI_ONLY_PNM   (.ppm and .pgm)
+
+         Note that you can define multiples of these, and you will get all
+         of them ("only x" and "only y" is interpreted to mean "only x&y").
+
+       - If you use STBI_NO_PNG (or _ONLY_ without PNG), and you still
+         want the zlib decoder to be available, #define STBI_SUPPORT_ZLIB
+
+      - Compilation of all SIMD code can be suppressed with
+            #define STBI_NO_SIMD
+        It should not be necessary to disable SIMD unless you have issues
+        compiling (e.g. using an x86 compiler which doesn't support SSE
+        intrinsics or that doesn't support the method used to detect
+        SSE2 support at run-time), and even those can be reported as
+        bugs so I can refine the built-in compile-time checking to be
+        smarter.
+
+      - The old STBI_SIMD system which allowed installing a user-defined
+        IDCT etc. has been removed. If you need this, don't upgrade. My
+        assumption is that almost nobody was doing this, and those who
+        were will find the built-in SIMD more satisfactory anyway.
+
+      - RGB values computed for JPEG images are slightly different from
+        previous versions of stb_image. (This is due to using less
+        integer precision in SIMD.) The C code has been adjusted so
+        that the same RGB values will be computed regardless of whether
+        SIMD support is available, so your app should always produce
+        consistent results. But these results are slightly different from
+        previous versions. (Specifically, about 3% of available YCbCr values
+        will compute different RGB results from pre-1.49 versions by +-1;
+        most of the deviating values are one smaller in the G channel.)
+
+      - If you must produce consistent results with previous versions of
+        stb_image, #define STBI_JPEG_OLD and you will get the same results
+        you used to; however, you will not get the SIMD speedups for
+        the YCbCr-to-RGB conversion step (although you should still see
+        significant JPEG speedup from the other changes).
+
+        Please note that STBI_JPEG_OLD is a temporary feature; it will be
+        removed in future versions of the library. It is only intended for
+        near-term back-compatibility use.
+
+
+   Latest revision history:
+      2.13  (2016-12-04) experimental 16-bit API, only for PNG so far; fixes
+      2.12  (2016-04-02) fix typo in 2.11 PSD fix that caused crashes
+      2.11  (2016-04-02) 16-bit PNGS; enable SSE2 in non-gcc x64
+                         RGB-format JPEG; remove white matting in PSD;
+                         allocate large structures on the stack;
+                         correct channel count for PNG & BMP
+      2.10  (2016-01-22) avoid warning introduced in 2.09
+      2.09  (2016-01-16) 16-bit TGA; comments in PNM files; STBI_REALLOC_SIZED
+      2.08  (2015-09-13) fix to 2.07 cleanup, reading RGB PSD as RGBA
+      2.07  (2015-09-13) partial animated GIF support
+                         limited 16-bit PSD support
+                         minor bugs, code cleanup, and compiler warnings
+
+   See end of file for full revision history.
+
+
+ ============================    Contributors    =========================
+
+ Image formats                          Extensions, features
+    Sean Barrett (jpeg, png, bmp)          Jetro Lauha (stbi_info)
+    Nicolas Schulz (hdr, psd)              Martin "SpartanJ" Golini (stbi_info)
+    Jonathan Dummer (tga)                  James "moose2000" Brown (iPhone PNG)
+    Jean-Marc Lienher (gif)                Ben "Disch" Wenger (io callbacks)
+    Tom Seddon (pic)                       Omar Cornut (1/2/4-bit PNG)
+    Thatcher Ulrich (psd)                  Nicolas Guillemot (vertical flip)
+    Ken Miller (pgm, ppm)                  Richard Mitton (16-bit PSD)
+    urraka@github (animated gif)           Junggon Kim (PNM comments)
+                                           Daniel Gibson (16-bit TGA)
+
+ Optimizations & bugfixes
+    Fabian "ryg" Giesen
+    Arseny Kapoulkine
+
+ Bug & warning fixes
+    Marc LeBlanc            David Woo          Guillaume George   Martins Mozeiko
+    Christpher Lloyd        Martin Golini      Jerry Jansson      Joseph Thomson
+    Dave Moore              Roy Eltham         Hayaki Saito       Phil Jordan
+    Won Chun                Luke Graham        Johan Duparc       Nathan Reed
+    the Horde3D community   Thomas Ruf         Ronny Chevalier    Nick Verigakis
+    Janez Zemva             John Bartholomew   Michal Cichon      svdijk@github
+    Jonathan Blow           Ken Hamada         Tero Hanninen      Baldur Karlsson
+    Laurent Gomila          Cort Stratton      Sergio Gonzalez    romigrou@github
+    Aruelien Pocheville     Thibault Reuille   Cass Everitt       Matthew Gregan
+    Ryamond Barbiero        Paul Du Bois       Engin Manap        snagar@github
+    Michaelangel007@github  Oriol Ferrer Mesia socks-the-fox      Zelex@github
+    Philipp Wiesemann       Josh Tobin         rlyeh@github       grim210@github
+    Blazej Dariusz Roszkowski
+
+
+LICENSE
+
+This software is dual-licensed to the public domain and under the following
+license: you are granted a perpetual, irrevocable license to copy, modify,
+publish, and distribute this file as you see fit.
+
+*/
+
+#ifndef STBI_INCLUDE_STB_IMAGE_H
+#define STBI_INCLUDE_STB_IMAGE_H
+
+// DOCUMENTATION
+//
+// Limitations:
+//    - no 16-bit-per-channel PNG
+//    - no 12-bit-per-channel JPEG
+//    - no JPEGs with arithmetic coding
+//    - no 1-bit BMP
+//    - GIF always returns *comp=4
+//
+// Basic usage (see HDR discussion below for HDR usage):
+//    int x,y,n;
+//    unsigned char *data = stbi_load(filename, &x, &y, &n, 0);
+//    // ... process data if not NULL ...
+//    // ... x = width, y = height, n = # 8-bit components per pixel ...
+//    // ... replace '0' with '1'..'4' to force that many components per pixel
+//    // ... but 'n' will always be the number that it would have been if you said 0
+//    stbi_image_free(data)
+//
+// Standard parameters:
+//    int *x                 -- outputs image width in pixels
+//    int *y                 -- outputs image height in pixels
+//    int *channels_in_file  -- outputs # of image components in image file
+//    int desired_channels   -- if non-zero, # of image components requested in result
+//
+// The return value from an image loader is an 'unsigned char *' which points
+// to the pixel data, or NULL on an allocation failure or if the image is
+// corrupt or invalid. The pixel data consists of *y scanlines of *x pixels,
+// with each pixel consisting of N interleaved 8-bit components; the first
+// pixel pointed to is top-left-most in the image. There is no padding between
+// image scanlines or between pixels, regardless of format. The number of
+// components N is 'req_comp' if req_comp is non-zero, or *comp otherwise.
+// If req_comp is non-zero, *comp has the number of components that _would_
+// have been output otherwise. E.g. if you set req_comp to 4, you will always
+// get RGBA output, but you can check *comp to see if it's trivially opaque
+// because e.g. there were only 3 channels in the source image.
+//
+// An output image with N components has the following components interleaved
+// in this order in each pixel:
+//
+//     N=#comp     components
+//       1           grey
+//       2           grey, alpha
+//       3           red, green, blue
+//       4           red, green, blue, alpha
+//
+// If image loading fails for any reason, the return value will be NULL,
+// and *x, *y, *comp will be unchanged. The function stbi_failure_reason()
+// can be queried for an extremely brief, end-user unfriendly explanation
+// of why the load failed. Define STBI_NO_FAILURE_STRINGS to avoid
+// compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly
+// more user-friendly ones.
+//
+// Paletted PNG, BMP, GIF, and PIC images are automatically depalettized.
+//
+// ===========================================================================
+//
+// Philosophy
+//
+// stb libraries are designed with the following priorities:
+//
+//    1. easy to use
+//    2. easy to maintain
+//    3. good performance
+//
+// Sometimes I let "good performance" creep up in priority over "easy to maintain",
+// and for best performance I may provide less-easy-to-use APIs that give higher
+// performance, in addition to the easy to use ones. Nevertheless, it's important
+// to keep in mind that from the standpoint of you, a client of this library,
+// all you care about is #1 and #3, and stb libraries do not emphasize #3 above all.
+//
+// Some secondary priorities arise directly from the first two, some of which
+// make more explicit reasons why performance can't be emphasized.
+//
+//    - Portable ("ease of use")
+//    - Small footprint ("easy to maintain")
+//    - No dependencies ("ease of use")
+//
+// ===========================================================================
+//
+// I/O callbacks
+//
+// I/O callbacks allow you to read from arbitrary sources, like packaged
+// files or some other source. Data read from callbacks are processed
+// through a small internal buffer (currently 128 bytes) to try to reduce
+// overhead.
+//
+// The three functions you must define are "read" (reads some bytes of data),
+// "skip" (skips some bytes of data), "eof" (reports if the stream is at the end).
+//
+// ===========================================================================
+//
+// SIMD support
+//
+// The JPEG decoder will try to automatically use SIMD kernels on x86 when
+// supported by the compiler. For ARM Neon support, you must explicitly
+// request it.
+//
+// (The old do-it-yourself SIMD API is no longer supported in the current
+// code.)
+//
+// On x86, SSE2 will automatically be used when available based on a run-time
+// test; if not, the generic C versions are used as a fall-back. On ARM targets,
+// the typical path is to have separate builds for NEON and non-NEON devices
+// (at least this is true for iOS and Android). Therefore, the NEON support is
+// toggled by a build flag: define STBI_NEON to get NEON loops.
+//
+// The output of the JPEG decoder is slightly different from versions where
+// SIMD support was introduced (that is, for versions before 1.49). The
+// difference is only +-1 in the 8-bit RGB channels, and only on a small
+// fraction of pixels. You can force the pre-1.49 behavior by defining
+// STBI_JPEG_OLD, but this will disable some of the SIMD decoding path
+// and hence cost some performance.
+//
+// If for some reason you do not want to use any of SIMD code, or if
+// you have issues compiling it, you can disable it entirely by
+// defining STBI_NO_SIMD.
+//
+// ===========================================================================
+//
+// HDR image support   (disable by defining STBI_NO_HDR)
+//
+// stb_image now supports loading HDR images in general, and currently
+// the Radiance .HDR file format, although the support is provided
+// generically. You can still load any file through the existing interface;
+// if you attempt to load an HDR file, it will be automatically remapped to
+// LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1;
+// both of these constants can be reconfigured through this interface:
+//
+//     stbi_hdr_to_ldr_gamma(2.2f);
+//     stbi_hdr_to_ldr_scale(1.0f);
+//
+// (note, do not use _inverse_ constants; stbi_image will invert them
+// appropriately).
+//
+// Additionally, there is a new, parallel interface for loading files as
+// (linear) floats to preserve the full dynamic range:
+//
+//    float *data = stbi_loadf(filename, &x, &y, &n, 0);
+//
+// If you load LDR images through this interface, those images will
+// be promoted to floating point values, run through the inverse of
+// constants corresponding to the above:
+//
+//     stbi_ldr_to_hdr_scale(1.0f);
+//     stbi_ldr_to_hdr_gamma(2.2f);
+//
+// Finally, given a filename (or an open file or memory block--see header
+// file for details) containing image data, you can query for the "most
+// appropriate" interface to use (that is, whether the image is HDR or
+// not), using:
+//
+//     stbi_is_hdr(char *filename);
+//
+// ===========================================================================
+//
+// iPhone PNG support:
+//
+// By default we convert iphone-formatted PNGs back to RGB, even though
+// they are internally encoded differently. You can disable this conversion
+// by by calling stbi_convert_iphone_png_to_rgb(0), in which case
+// you will always just get the native iphone "format" through (which
+// is BGR stored in RGB).
+//
+// Call stbi_set_unpremultiply_on_load(1) as well to force a divide per
+// pixel to remove any premultiplied alpha *only* if the image file explicitly
+// says there's premultiplied data (currently only happens in iPhone images,
+// and only if iPhone convert-to-rgb processing is on).
+//
+
+
+#ifndef STBI_NO_STDIO
+#include <stdio.h>
+#endif // STBI_NO_STDIO
+
+#define STBI_VERSION 1
+
+enum
+{
+    STBI_default = 0, // only used for req_comp
+
+    STBI_grey       = 1,
+    STBI_grey_alpha = 2,
+    STBI_rgb        = 3,
+    STBI_rgb_alpha  = 4
+};
+
+typedef unsigned char stbi_uc;
+typedef unsigned short stbi_us;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifdef STB_IMAGE_STATIC
+#define STBIDEF static
+#else
+#define STBIDEF extern
+#endif
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// PRIMARY API - works on images of any type
+//
+
+//
+// load image by filename, open file, or memory buffer
+//
+
+typedef struct
+{
+    int      (*read)  (void *user,char *data,int size);   // fill 'data' with 'size' bytes.  return number of bytes actually read
+    void     (*skip)  (void *user,int n);                 // skip the next 'n' bytes, or 'unget' the last -n bytes if negative
+    int      (*eof)   (void *user);                       // returns nonzero if we are at end of file/data
+} stbi_io_callbacks;
+
+////////////////////////////////////
+//
+// 8-bits-per-channel interface
+//
+
+STBIDEF stbi_uc *stbi_load               (char              const *filename,           int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_uc *stbi_load_from_memory   (stbi_uc           const *buffer, int len   , int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk  , void *user, int *x, int *y, int *channels_in_file, int desired_channels);
+
+#ifndef STBI_NO_STDIO
+STBIDEF stbi_uc *stbi_load_from_file   (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
+// for stbi_load_from_file, file pointer is left pointing immediately after image
+#endif
+
+////////////////////////////////////
+//
+// 16-bits-per-channel interface
+//
+
+STBIDEF stbi_us *stbi_load_16(char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
+#ifndef STBI_NO_STDIO
+STBIDEF stbi_us *stbi_load_from_file_16(FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
+#endif
+// @TODO the other variants
+
+////////////////////////////////////
+//
+// float-per-channel interface
+//
+#ifndef STBI_NO_LINEAR
+STBIDEF float *stbi_loadf                 (char const *filename,           int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF float *stbi_loadf_from_memory     (stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF float *stbi_loadf_from_callbacks  (stbi_io_callbacks const *clbk, void *user, int *x, int *y,  int *channels_in_file, int desired_channels);
+
+#ifndef STBI_NO_STDIO
+STBIDEF float *stbi_loadf_from_file  (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
+#endif
+#endif
+
+#ifndef STBI_NO_HDR
+STBIDEF void   stbi_hdr_to_ldr_gamma(float gamma);
+STBIDEF void   stbi_hdr_to_ldr_scale(float scale);
+#endif // STBI_NO_HDR
+
+#ifndef STBI_NO_LINEAR
+STBIDEF void   stbi_ldr_to_hdr_gamma(float gamma);
+STBIDEF void   stbi_ldr_to_hdr_scale(float scale);
+#endif // STBI_NO_LINEAR
+
+// stbi_is_hdr is always defined, but always returns false if STBI_NO_HDR
+STBIDEF int    stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user);
+STBIDEF int    stbi_is_hdr_from_memory(stbi_uc const *buffer, int len);
+#ifndef STBI_NO_STDIO
+STBIDEF int      stbi_is_hdr          (char const *filename);
+STBIDEF int      stbi_is_hdr_from_file(FILE *f);
+#endif // STBI_NO_STDIO
+
+
+// get a VERY brief reason for failure
+// NOT THREADSAFE
+STBIDEF const char *stbi_failure_reason  (void);
+
+// free the loaded image -- this is just free()
+STBIDEF void     stbi_image_free      (void *retval_from_stbi_load);
+
+// get image dimensions & components without fully decoding
+STBIDEF int      stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp);
+STBIDEF int      stbi_info_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp);
+
+#ifndef STBI_NO_STDIO
+STBIDEF int      stbi_info            (char const *filename,     int *x, int *y, int *comp);
+STBIDEF int      stbi_info_from_file  (FILE *f,                  int *x, int *y, int *comp);
+
+#endif
+
+
+
+// for image formats that explicitly notate that they have premultiplied alpha,
+// we just return the colors as stored in the file. set this flag to force
+// unpremultiplication. results are undefined if the unpremultiply overflow.
+STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply);
+
+// indicate whether we should process iphone images back to canonical format,
+// or just pass them through "as-is"
+STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert);
+
+// flip the image vertically, so the first pixel in the output array is the bottom left
+STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip);
+
+// ZLIB client - used by PNG, available for other purposes
+
+STBIDEF char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen);
+STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, int len, int initial_size, int *outlen, int parse_header);
+STBIDEF char *stbi_zlib_decode_malloc(const char *buffer, int len, int *outlen);
+STBIDEF int   stbi_zlib_decode_buffer(char *obuffer, int olen, const char *ibuffer, int ilen);
+
+STBIDEF char *stbi_zlib_decode_noheader_malloc(const char *buffer, int len, int *outlen);
+STBIDEF int   stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char *ibuffer, int ilen);
+
+
+#ifdef __cplusplus
+}
+#endif
+
+//
+//
+////   end header file   /////////////////////////////////////////////////////
+#endif // STBI_INCLUDE_STB_IMAGE_H
+
+#ifdef STB_IMAGE_IMPLEMENTATION
+
+#if defined(STBI_ONLY_JPEG) || defined(STBI_ONLY_PNG) || defined(STBI_ONLY_BMP) \
+  || defined(STBI_ONLY_TGA) || defined(STBI_ONLY_GIF) || defined(STBI_ONLY_PSD) \
+  || defined(STBI_ONLY_HDR) || defined(STBI_ONLY_PIC) || defined(STBI_ONLY_PNM) \
+  || defined(STBI_ONLY_ZLIB)
+   #ifndef STBI_ONLY_JPEG
+   #define STBI_NO_JPEG
+   #endif
+   #ifndef STBI_ONLY_PNG
+   #define STBI_NO_PNG
+   #endif
+   #ifndef STBI_ONLY_BMP
+   #define STBI_NO_BMP
+   #endif
+   #ifndef STBI_ONLY_PSD
+   #define STBI_NO_PSD
+   #endif
+   #ifndef STBI_ONLY_TGA
+   #define STBI_NO_TGA
+   #endif
+   #ifndef STBI_ONLY_GIF
+   #define STBI_NO_GIF
+   #endif
+   #ifndef STBI_ONLY_HDR
+   #define STBI_NO_HDR
+   #endif
+   #ifndef STBI_ONLY_PIC
+   #define STBI_NO_PIC
+   #endif
+   #ifndef STBI_ONLY_PNM
+   #define STBI_NO_PNM
+   #endif
+#endif
+
+#if defined(STBI_NO_PNG) && !defined(STBI_SUPPORT_ZLIB) && !defined(STBI_NO_ZLIB)
+#define STBI_NO_ZLIB
+#endif
+
+
+#include <stdarg.h>
+#include <stddef.h> // ptrdiff_t on osx
+#include <stdlib.h>
+#include <string.h>
+#include <limits.h>
+
+#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR)
+#include <math.h>  // ldexp
+#endif
+
+#ifndef STBI_NO_STDIO
+#include <stdio.h>
+#endif
+
+#ifndef STBI_ASSERT
+#include <assert.h>
+#define STBI_ASSERT(x) assert(x)
+#endif
+
+
+#ifndef _MSC_VER
+   #ifdef __cplusplus
+   #define stbi_inline inline
+   #else
+   #define stbi_inline
+   #endif
+#else
+   #define stbi_inline __forceinline
+#endif
+
+
+#ifdef _MSC_VER
+typedef unsigned short stbi__uint16;
+typedef   signed short stbi__int16;
+typedef unsigned int   stbi__uint32;
+typedef   signed int   stbi__int32;
+#else
+#include <stdint.h>
+typedef uint16_t stbi__uint16;
+typedef int16_t  stbi__int16;
+typedef uint32_t stbi__uint32;
+typedef int32_t  stbi__int32;
+#endif
+
+// should produce compiler error if size is wrong
+typedef unsigned char validate_uint32[sizeof(stbi__uint32)==4 ? 1 : -1];
+
+#ifdef _MSC_VER
+#define STBI_NOTUSED(v)  (void)(v)
+#else
+#define STBI_NOTUSED(v)  (void)sizeof(v)
+#endif
+
+#ifdef _MSC_VER
+#define STBI_HAS_LROTL
+#endif
+
+#ifdef STBI_HAS_LROTL
+   #define stbi_lrot(x,y)  _lrotl(x,y)
+#else
+   #define stbi_lrot(x,y)  (((x) << (y)) | ((x) >> (32 - (y))))
+#endif
+
+#if defined(STBI_MALLOC) && defined(STBI_FREE) && (defined(STBI_REALLOC) || defined(STBI_REALLOC_SIZED))
+// ok
+#elif !defined(STBI_MALLOC) && !defined(STBI_FREE) && !defined(STBI_REALLOC) && !defined(STBI_REALLOC_SIZED)
+// ok
+#else
+#error "Must define all or none of STBI_MALLOC, STBI_FREE, and STBI_REALLOC (or STBI_REALLOC_SIZED)."
+#endif
+
+#ifndef STBI_MALLOC
+#define STBI_MALLOC(sz)           malloc(sz)
+#define STBI_REALLOC(p,newsz)     realloc(p,newsz)
+#define STBI_FREE(p)              free(p)
+#endif
+
+#ifndef STBI_REALLOC_SIZED
+#define STBI_REALLOC_SIZED(p,oldsz,newsz) STBI_REALLOC(p,newsz)
+#endif
+
+// x86/x64 detection
+#if defined(__x86_64__) || defined(_M_X64)
+#define STBI__X64_TARGET
+#elif defined(__i386) || defined(_M_IX86)
+#define STBI__X86_TARGET
+#endif
+
+#if defined(__GNUC__) && (defined(STBI__X86_TARGET) || defined(STBI__X64_TARGET)) && !defined(__SSE2__) && !defined(STBI_NO_SIMD)
+// NOTE: not clear do we actually need this for the 64-bit path?
+// gcc doesn't support sse2 intrinsics unless you compile with -msse2,
+// (but compiling with -msse2 allows the compiler to use SSE2 everywhere;
+// this is just broken and gcc are jerks for not fixing it properly
+// http://www.virtualdub.org/blog/pivot/entry.php?id=363 )
+#define STBI_NO_SIMD
+#endif
+
+#if defined(__MINGW32__) && defined(STBI__X86_TARGET) && !defined(STBI_MINGW_ENABLE_SSE2) && !defined(STBI_NO_SIMD)
+// Note that __MINGW32__ doesn't actually mean 32-bit, so we have to avoid STBI__X64_TARGET
+//
+// 32-bit MinGW wants ESP to be 16-byte aligned, but this is not in the
+// Windows ABI and VC++ as well as Windows DLLs don't maintain that invariant.
+// As a result, enabling SSE2 on 32-bit MinGW is dangerous when not
+// simultaneously enabling "-mstackrealign".
+//
+// See https://github.com/nothings/stb/issues/81 for more information.
+//
+// So default to no SSE2 on 32-bit MinGW. If you've read this far and added
+// -mstackrealign to your build settings, feel free to #define STBI_MINGW_ENABLE_SSE2.
+#define STBI_NO_SIMD
+#endif
+
+#if !defined(STBI_NO_SIMD) && (defined(STBI__X86_TARGET) || defined(STBI__X64_TARGET))
+#define STBI_SSE2
+#include <emmintrin.h>
+
+#ifdef _MSC_VER
+
+#if _MSC_VER >= 1400  // not VC6
+#include <intrin.h> // __cpuid
+static int stbi__cpuid3(void)
+{
+   int info[4];
+   __cpuid(info,1);
+   return info[3];
+}
+#else
+static int stbi__cpuid3(void)
+{
+   int res;
+   __asm {
+      mov  eax,1
+      cpuid
+      mov  res,edx
+   }
+   return res;
+}
+#endif
+
+#define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name
+
+static int stbi__sse2_available()
+{
+   int info3 = stbi__cpuid3();
+   return ((info3 >> 26) & 1) != 0;
+}
+#else // assume GCC-style if not VC++
+#define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16)))
+
+static int stbi__sse2_available()
+{
+#if defined(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__) >= 408 // GCC 4.8 or later
+   // GCC 4.8+ has a nice way to do this
+   return __builtin_cpu_supports("sse2");
+#else
+   // portable way to do this, preferably without using GCC inline ASM?
+   // just bail for now.
+   return 0;
+#endif
+}
+#endif
+#endif
+
+// ARM NEON
+#if defined(STBI_NO_SIMD) && defined(STBI_NEON)
+#undef STBI_NEON
+#endif
+
+#ifdef STBI_NEON
+#include <arm_neon.h>
+// assume GCC or Clang on ARM targets
+#define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16)))
+#endif
+
+#ifndef STBI_SIMD_ALIGN
+#define STBI_SIMD_ALIGN(type, name) type name
+#endif
+
+///////////////////////////////////////////////
+//
+//  stbi__context struct and start_xxx functions
+
+// stbi__context structure is our basic context used by all images, so it
+// contains all the IO context, plus some basic image information
+typedef struct
+{
+   stbi__uint32 img_x, img_y;
+   int img_n, img_out_n;
+
+   stbi_io_callbacks io;
+   void *io_user_data;
+
+   int read_from_callbacks;
+   int buflen;
+   stbi_uc buffer_start[128];
+
+   stbi_uc *img_buffer, *img_buffer_end;
+   stbi_uc *img_buffer_original, *img_buffer_original_end;
+} stbi__context;
+
+
+static void stbi__refill_buffer(stbi__context *s);
+
+// initialize a memory-decode context
+static void stbi__start_mem(stbi__context *s, stbi_uc const *buffer, int len)
+{
+   s->io.read = NULL;
+   s->read_from_callbacks = 0;
+   s->img_buffer = s->img_buffer_original = (stbi_uc *) buffer;
+   s->img_buffer_end = s->img_buffer_original_end = (stbi_uc *) buffer+len;
+}
+
+// initialize a callback-based context
+static void stbi__start_callbacks(stbi__context *s, stbi_io_callbacks *c, void *user)
+{
+   s->io = *c;
+   s->io_user_data = user;
+   s->buflen = sizeof(s->buffer_start);
+   s->read_from_callbacks = 1;
+   s->img_buffer_original = s->buffer_start;
+   stbi__refill_buffer(s);
+   s->img_buffer_original_end = s->img_buffer_end;
+}
+
+#ifndef STBI_NO_STDIO
+
+static int stbi__stdio_read(void *user, char *data, int size)
+{
+   return (int) fread(data,1,size,(FILE*) user);
+}
+
+static void stbi__stdio_skip(void *user, int n)
+{
+   fseek((FILE*) user, n, SEEK_CUR);
+}
+
+static int stbi__stdio_eof(void *user)
+{
+   return feof((FILE*) user);
+}
+
+static stbi_io_callbacks stbi__stdio_callbacks =
+{
+   stbi__stdio_read,
+   stbi__stdio_skip,
+   stbi__stdio_eof,
+};
+
+static void stbi__start_file(stbi__context *s, FILE *f)
+{
+   stbi__start_callbacks(s, &stbi__stdio_callbacks, (void *) f);
+}
+
+//static void stop_file(stbi__context *s) { }
+
+#endif // !STBI_NO_STDIO
+
+static void stbi__rewind(stbi__context *s)
+{
+   // conceptually rewind SHOULD rewind to the beginning of the stream,
+   // but we just rewind to the beginning of the initial buffer, because
+   // we only use it after doing 'test', which only ever looks at at most 92 bytes
+   s->img_buffer = s->img_buffer_original;
+   s->img_buffer_end = s->img_buffer_original_end;
+}
+
+enum
+{
+   STBI_ORDER_RGB,
+   STBI_ORDER_BGR
+};
+
+typedef struct
+{
+   int bits_per_channel;
+   int num_channels;
+   int channel_order;
+} stbi__result_info;
+
+#ifndef STBI_NO_JPEG
+static int      stbi__jpeg_test(stbi__context *s);
+static void    *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static int      stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+#ifndef STBI_NO_PNG
+static int      stbi__png_test(stbi__context *s);
+static void    *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static int      stbi__png_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+#ifndef STBI_NO_BMP
+static int      stbi__bmp_test(stbi__context *s);
+static void    *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static int      stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+#ifndef STBI_NO_TGA
+static int      stbi__tga_test(stbi__context *s);
+static void    *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static int      stbi__tga_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+#ifndef STBI_NO_PSD
+static int      stbi__psd_test(stbi__context *s);
+static void    *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc);
+static int      stbi__psd_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+#ifndef STBI_NO_HDR
+static int      stbi__hdr_test(stbi__context *s);
+static float   *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static int      stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+#ifndef STBI_NO_PIC
+static int      stbi__pic_test(stbi__context *s);
+static void    *stbi__pic_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static int      stbi__pic_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+#ifndef STBI_NO_GIF
+static int      stbi__gif_test(stbi__context *s);
+static void    *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static int      stbi__gif_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+#ifndef STBI_NO_PNM
+static int      stbi__pnm_test(stbi__context *s);
+static void    *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static int      stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp);
+#endif
+
+// this is not threadsafe
+static const char *stbi__g_failure_reason;
+
+STBIDEF const char *stbi_failure_reason(void)
+{
+   return stbi__g_failure_reason;
+}
+
+static int stbi__err(const char *str)
+{
+   stbi__g_failure_reason = str;
+   return 0;
+}
+
+static void *stbi__malloc(size_t size)
+{
+    return STBI_MALLOC(size);
+}
+
+// stb_image uses ints pervasively, including for offset calculations.
+// therefore the largest decoded image size we can support with the
+// current code, even on 64-bit targets, is INT_MAX. this is not a
+// significant limitation for the intended use case.
+//
+// we do, however, need to make sure our size calculations don't
+// overflow. hence a few helper functions for size calculations that
+// multiply integers together, making sure that they're non-negative
+// and no overflow occurs.
+
+// return 1 if the sum is valid, 0 on overflow.
+// negative terms are considered invalid.
+static int stbi__addsizes_valid(int a, int b)
+{
+   if (b < 0) return 0;
+   // now 0 <= b <= INT_MAX, hence also
+   // 0 <= INT_MAX - b <= INTMAX.
+   // And "a + b <= INT_MAX" (which might overflow) is the
+   // same as a <= INT_MAX - b (no overflow)
+   return a <= INT_MAX - b;
+}
+
+// returns 1 if the product is valid, 0 on overflow.
+// negative factors are considered invalid.
+static int stbi__mul2sizes_valid(int a, int b)
+{
+   if (a < 0 || b < 0) return 0;
+   if (b == 0) return 1; // mul-by-0 is always safe
+   // portable way to check for no overflows in a*b
+   return a <= INT_MAX/b;
+}
+
+// returns 1 if "a*b + add" has no negative terms/factors and doesn't overflow
+static int stbi__mad2sizes_valid(int a, int b, int add)
+{
+   return stbi__mul2sizes_valid(a, b) && stbi__addsizes_valid(a*b, add);
+}
+
+// returns 1 if "a*b*c + add" has no negative terms/factors and doesn't overflow
+static int stbi__mad3sizes_valid(int a, int b, int c, int add)
+{
+   return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a*b, c) &&
+      stbi__addsizes_valid(a*b*c, add);
+}
+
+// returns 1 if "a*b*c*d + add" has no negative terms/factors and doesn't overflow
+static int stbi__mad4sizes_valid(int a, int b, int c, int d, int add)
+{
+   return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a*b, c) &&
+      stbi__mul2sizes_valid(a*b*c, d) && stbi__addsizes_valid(a*b*c*d, add);
+}
+
+// mallocs with size overflow checking
+static void *stbi__malloc_mad2(int a, int b, int add)
+{
+   if (!stbi__mad2sizes_valid(a, b, add)) return NULL;
+   return stbi__malloc(a*b + add);
+}
+
+static void *stbi__malloc_mad3(int a, int b, int c, int add)
+{
+   if (!stbi__mad3sizes_valid(a, b, c, add)) return NULL;
+   return stbi__malloc(a*b*c + add);
+}
+
+static void *stbi__malloc_mad4(int a, int b, int c, int d, int add)
+{
+   if (!stbi__mad4sizes_valid(a, b, c, d, add)) return NULL;
+   return stbi__malloc(a*b*c*d + add);
+}
+
+// stbi__err - error
+// stbi__errpf - error returning pointer to float
+// stbi__errpuc - error returning pointer to unsigned char
+
+#ifdef STBI_NO_FAILURE_STRINGS
+   #define stbi__err(x,y)  0
+#elif defined(STBI_FAILURE_USERMSG)
+   #define stbi__err(x,y)  stbi__err(y)
+#else
+   #define stbi__err(x,y)  stbi__err(x)
+#endif
+
+#define stbi__errpf(x,y)   ((float *)(size_t) (stbi__err(x,y)?NULL:NULL))
+#define stbi__errpuc(x,y)  ((unsigned char *)(size_t) (stbi__err(x,y)?NULL:NULL))
+
+STBIDEF void stbi_image_free(void *retval_from_stbi_load)
+{
+   STBI_FREE(retval_from_stbi_load);
+}
+
+#ifndef STBI_NO_LINEAR
+static float   *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp);
+#endif
+
+#ifndef STBI_NO_HDR
+static stbi_uc *stbi__hdr_to_ldr(float   *data, int x, int y, int comp);
+#endif
+
+static int stbi__vertically_flip_on_load = 0;
+
+STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip)
+{
+    stbi__vertically_flip_on_load = flag_true_if_should_flip;
+}
+
+static void *stbi__load_main(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc)
+{
+   memset(ri, 0, sizeof(*ri)); // make sure it's initialized if we add new fields
+   ri->bits_per_channel = 8; // default is 8 so most paths don't have to be changed
+   ri->channel_order = STBI_ORDER_RGB; // all current input & output are this, but this is here so we can add BGR order
+   ri->num_channels = 0;
+
+   #ifndef STBI_NO_JPEG
+   if (stbi__jpeg_test(s)) return stbi__jpeg_load(s,x,y,comp,req_comp, ri);
+   #endif
+   #ifndef STBI_NO_PNG
+   if (stbi__png_test(s))  return stbi__png_load(s,x,y,comp,req_comp, ri);
+   #endif
+   #ifndef STBI_NO_BMP
+   if (stbi__bmp_test(s))  return stbi__bmp_load(s,x,y,comp,req_comp, ri);
+   #endif
+   #ifndef STBI_NO_GIF
+   if (stbi__gif_test(s))  return stbi__gif_load(s,x,y,comp,req_comp, ri);
+   #endif
+   #ifndef STBI_NO_PSD
+   if (stbi__psd_test(s))  return stbi__psd_load(s,x,y,comp,req_comp, ri, bpc);
+   #endif
+   #ifndef STBI_NO_PIC
+   if (stbi__pic_test(s))  return stbi__pic_load(s,x,y,comp,req_comp, ri);
+   #endif
+   #ifndef STBI_NO_PNM
+   if (stbi__pnm_test(s))  return stbi__pnm_load(s,x,y,comp,req_comp, ri);
+   #endif
+
+   #ifndef STBI_NO_HDR
+   if (stbi__hdr_test(s)) {
+      float *hdr = stbi__hdr_load(s, x,y,comp,req_comp, ri);
+      return stbi__hdr_to_ldr(hdr, *x, *y, req_comp ? req_comp : *comp);
+   }
+   #endif
+
+   #ifndef STBI_NO_TGA
+   // test tga last because it's a crappy test!
+   if (stbi__tga_test(s))
+      return stbi__tga_load(s,x,y,comp,req_comp, ri);
+   #endif
+
+   return stbi__errpuc("unknown image type", "Image not of any known type, or corrupt");
+}
+
+static stbi_uc *stbi__convert_16_to_8(stbi__uint16 *orig, int w, int h, int channels)
+{
+   int i;
+   int img_len = w * h * channels;
+   stbi_uc *reduced;
+
+   reduced = (stbi_uc *) stbi__malloc(img_len);
+   if (reduced == NULL) return stbi__errpuc("outofmem", "Out of memory");
+
+   for (i = 0; i < img_len; ++i)
+      reduced[i] = (stbi_uc)((orig[i] >> 8) & 0xFF); // top half of each byte is sufficient approx of 16->8 bit scaling
+
+   STBI_FREE(orig);
+   return reduced;
+}
+
+static stbi__uint16 *stbi__convert_8_to_16(stbi_uc *orig, int w, int h, int channels)
+{
+   int i;
+   int img_len = w * h * channels;
+   stbi__uint16 *enlarged;
+
+   enlarged = (stbi__uint16 *) stbi__malloc(img_len*2);
+   if (enlarged == NULL) return (stbi__uint16 *) stbi__errpuc("outofmem", "Out of memory");
+
+   for (i = 0; i < img_len; ++i)
+      enlarged[i] = (stbi__uint16)((orig[i] << 8) + orig[i]); // replicate to high and low byte, maps 0->0, 255->0xffff
+
+   STBI_FREE(orig);
+   return enlarged;
+}
+
+static unsigned char *stbi__load_and_postprocess_8bit(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+{
+   stbi__result_info ri;
+   void *result = stbi__load_main(s, x, y, comp, req_comp, &ri, 8);
+
+   if (result == NULL)
+      return NULL;
+
+   if (ri.bits_per_channel != 8) {
+      STBI_ASSERT(ri.bits_per_channel == 16);
+      result = stbi__convert_16_to_8((stbi__uint16 *) result, *x, *y, req_comp == 0 ? *comp : req_comp);
+      ri.bits_per_channel = 8;
+   }
+
+   // @TODO: move stbi__convert_format to here
+
+   if (stbi__vertically_flip_on_load) {
+      int w = *x, h = *y;
+      int channels = req_comp ? req_comp : *comp;
+      int row,col,z;
+      stbi_uc *image = (stbi_uc *) result;
+
+      // @OPTIMIZE: use a bigger temp buffer and memcpy multiple pixels at once
+      for (row = 0; row < (h>>1); row++) {
+         for (col = 0; col < w; col++) {
+            for (z = 0; z < channels; z++) {
+               stbi_uc temp = image[(row * w + col) * channels + z];
+               image[(row * w + col) * channels + z] = image[((h - row - 1) * w + col) * channels + z];
+               image[((h - row - 1) * w + col) * channels + z] = temp;
+            }
+         }
+      }
+   }
+
+   return (unsigned char *) result;
+}
+
+static stbi__uint16 *stbi__load_and_postprocess_16bit(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+{
+   stbi__result_info ri;
+   void *result = stbi__load_main(s, x, y, comp, req_comp, &ri, 16);
+
+   if (result == NULL)
+      return NULL;
+
+   if (ri.bits_per_channel != 16) {
+      STBI_ASSERT(ri.bits_per_channel == 8);
+      result = stbi__convert_8_to_16((stbi_uc *) result, *x, *y, req_comp == 0 ? *comp : req_comp);
+      ri.bits_per_channel = 16;
+   }
+
+   // @TODO: move stbi__convert_format16 to here
+   // @TODO: special case RGB-to-Y (and RGBA-to-YA) for 8-bit-to-16-bit case to keep more precision
+
+   if (stbi__vertically_flip_on_load) {
+      int w = *x, h = *y;
+      int channels = req_comp ? req_comp : *comp;
+      int row,col,z;
+      stbi__uint16 *image = (stbi__uint16 *) result;
+
+      // @OPTIMIZE: use a bigger temp buffer and memcpy multiple pixels at once
+      for (row = 0; row < (h>>1); row++) {
+         for (col = 0; col < w; col++) {
+            for (z = 0; z < channels; z++) {
+               stbi__uint16 temp = image[(row * w + col) * channels + z];
+               image[(row * w + col) * channels + z] = image[((h - row - 1) * w + col) * channels + z];
+               image[((h - row - 1) * w + col) * channels + z] = temp;
+            }
+         }
+      }
+   }
+
+   return (stbi__uint16 *) result;
+}
+
+#ifndef STBI_NO_HDR
+static void stbi__float_postprocess(float *result, int *x, int *y, int *comp, int req_comp)
+{
+   if (stbi__vertically_flip_on_load && result != NULL) {
+      int w = *x, h = *y;
+      int depth = req_comp ? req_comp : *comp;
+      int row,col,z;
+      float temp;
+
+      // @OPTIMIZE: use a bigger temp buffer and memcpy multiple pixels at once
+      for (row = 0; row < (h>>1); row++) {
+         for (col = 0; col < w; col++) {
+            for (z = 0; z < depth; z++) {
+               temp = result[(row * w + col) * depth + z];
+               result[(row * w + col) * depth + z] = result[((h - row - 1) * w + col) * depth + z];
+               result[((h - row - 1) * w + col) * depth + z] = temp;
+            }
+         }
+      }
+   }
+}
+#endif
+
+#ifndef STBI_NO_STDIO
+
+static FILE *stbi__fopen(char const *filename, char const *mode)
+{
+   FILE *f;
+#if defined(_MSC_VER) && _MSC_VER >= 1400
+   if (0 != fopen_s(&f, filename, mode))
+      f=0;
+#else
+   f = fopen(filename, mode);
+#endif
+   return f;
+}
+
+
+STBIDEF stbi_uc *stbi_load(char const *filename, int *x, int *y, int *comp, int req_comp)
+{
+   FILE *f = stbi__fopen(filename, "rb");
+   unsigned char *result;
+   if (!f) return stbi__errpuc("can't fopen", "Unable to open file");
+   result = stbi_load_from_file(f,x,y,comp,req_comp);
+   fclose(f);
+   return result;
+}
+
+STBIDEF stbi_uc *stbi_load_from_file(FILE *f, int *x, int *y, int *comp, int req_comp)
+{
+   unsigned char *result;
+   stbi__context s;
+   stbi__start_file(&s,f);
+   result = stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp);
+   if (result) {
+      // need to 'unget' all the characters in the IO buffer
+      fseek(f, - (int) (s.img_buffer_end - s.img_buffer), SEEK_CUR);
+   }
+   return result;
+}
+
+STBIDEF stbi__uint16 *stbi_load_from_file_16(FILE *f, int *x, int *y, int *comp, int req_comp)
+{
+   stbi__uint16 *result;
+   stbi__context s;
+   stbi__start_file(&s,f);
+   result = stbi__load_and_postprocess_16bit(&s,x,y,comp,req_comp);
+   if (result) {
+      // need to 'unget' all the characters in the IO buffer
+      fseek(f, - (int) (s.img_buffer_end - s.img_buffer), SEEK_CUR);
+   }
+   return result;
+}
+
+STBIDEF stbi_us *stbi_load_16(char const *filename, int *x, int *y, int *comp, int req_comp)
+{
+   FILE *f = stbi__fopen(filename, "rb");
+   stbi__uint16 *result;
+   if (!f) return (stbi_us *) stbi__errpuc("can't fopen", "Unable to open file");
+   result = stbi_load_from_file_16(f,x,y,comp,req_comp);
+   fclose(f);
+   return result;
+}
+
+
+#endif //!STBI_NO_STDIO
+
+STBIDEF stbi_uc *stbi_load_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp)
+{
+   stbi__context s;
+   stbi__start_mem(&s,buffer,len);
+   return stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp);
+}
+
+STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp)
+{
+   stbi__context s;
+   stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user);
+   return stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp);
+}
+
+#ifndef STBI_NO_LINEAR
+static float *stbi__loadf_main(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+{
+   unsigned char *data;
+   #ifndef STBI_NO_HDR
+   if (stbi__hdr_test(s)) {
+      stbi__result_info ri;
+      float *hdr_data = stbi__hdr_load(s,x,y,comp,req_comp, &ri);
+      if (hdr_data)
+         stbi__float_postprocess(hdr_data,x,y,comp,req_comp);
+      return hdr_data;
+   }
+   #endif
+   data = stbi__load_and_postprocess_8bit(s, x, y, comp, req_comp);
+   if (data)
+      return stbi__ldr_to_hdr(data, *x, *y, req_comp ? req_comp : *comp);
+   return stbi__errpf("unknown image type", "Image not of any known type, or corrupt");
+}
+
+STBIDEF float *stbi_loadf_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp)
+{
+   stbi__context s;
+   stbi__start_mem(&s,buffer,len);
+   return stbi__loadf_main(&s,x,y,comp,req_comp);
+}
+
+STBIDEF float *stbi_loadf_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp)
+{
+   stbi__context s;
+   stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user);
+   return stbi__loadf_main(&s,x,y,comp,req_comp);
+}
+
+#ifndef STBI_NO_STDIO
+STBIDEF float *stbi_loadf(char const *filename, int *x, int *y, int *comp, int req_comp)
+{
+   float *result;
+   FILE *f = stbi__fopen(filename, "rb");
+   if (!f) return stbi__errpf("can't fopen", "Unable to open file");
+   result = stbi_loadf_from_file(f,x,y,comp,req_comp);
+   fclose(f);
+   return result;
+}
+
+STBIDEF float *stbi_loadf_from_file(FILE *f, int *x, int *y, int *comp, int req_comp)
+{
+   stbi__context s;
+   stbi__start_file(&s,f);
+   return stbi__loadf_main(&s,x,y,comp,req_comp);
+}
+#endif // !STBI_NO_STDIO
+
+#endif // !STBI_NO_LINEAR
+
+// these is-hdr-or-not is defined independent of whether STBI_NO_LINEAR is
+// defined, for API simplicity; if STBI_NO_LINEAR is defined, it always
+// reports false!
+
+STBIDEF int stbi_is_hdr_from_memory(stbi_uc const *buffer, int len)
+{
+   #ifndef STBI_NO_HDR
+   stbi__context s;
+   stbi__start_mem(&s,buffer,len);
+   return stbi__hdr_test(&s);
+   #else
+   STBI_NOTUSED(buffer);
+   STBI_NOTUSED(len);
+   return 0;
+   #endif
+}
+
+#ifndef STBI_NO_STDIO
+STBIDEF int      stbi_is_hdr          (char const *filename)
+{
+   FILE *f = stbi__fopen(filename, "rb");
+   int result=0;
+   if (f) {
+      result = stbi_is_hdr_from_file(f);
+      fclose(f);
+   }
+   return result;
+}
+
+STBIDEF int      stbi_is_hdr_from_file(FILE *f)
+{
+   #ifndef STBI_NO_HDR
+   stbi__context s;
+   stbi__start_file(&s,f);
+   return stbi__hdr_test(&s);
+   #else
+   STBI_NOTUSED(f);
+   return 0;
+   #endif
+}
+#endif // !STBI_NO_STDIO
+
+STBIDEF int      stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user)
+{
+   #ifndef STBI_NO_HDR
+   stbi__context s;
+   stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user);
+   return stbi__hdr_test(&s);
+   #else
+   STBI_NOTUSED(clbk);
+   STBI_NOTUSED(user);
+   return 0;
+   #endif
+}
+
+#ifndef STBI_NO_LINEAR
+static float stbi__l2h_gamma=2.2f, stbi__l2h_scale=1.0f;
+
+STBIDEF void   stbi_ldr_to_hdr_gamma(float gamma) { stbi__l2h_gamma = gamma; }
+STBIDEF void   stbi_ldr_to_hdr_scale(float scale) { stbi__l2h_scale = scale; }
+#endif
+
+static float stbi__h2l_gamma_i=1.0f/2.2f, stbi__h2l_scale_i=1.0f;
+
+STBIDEF void   stbi_hdr_to_ldr_gamma(float gamma) { stbi__h2l_gamma_i = 1/gamma; }
+STBIDEF void   stbi_hdr_to_ldr_scale(float scale) { stbi__h2l_scale_i = 1/scale; }
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// Common code used by all image loaders
+//
+
+enum
+{
+   STBI__SCAN_load=0,
+   STBI__SCAN_type,
+   STBI__SCAN_header
+};
+
+static void stbi__refill_buffer(stbi__context *s)
+{
+   int n = (s->io.read)(s->io_user_data,(char*)s->buffer_start,s->buflen);
+   if (n == 0) {
+      // at end of file, treat same as if from memory, but need to handle case
+      // where s->img_buffer isn't pointing to safe memory, e.g. 0-byte file
+      s->read_from_callbacks = 0;
+      s->img_buffer = s->buffer_start;
+      s->img_buffer_end = s->buffer_start+1;
+      *s->img_buffer = 0;
+   } else {
+      s->img_buffer = s->buffer_start;
+      s->img_buffer_end = s->buffer_start + n;
+   }
+}
+
+stbi_inline static stbi_uc stbi__get8(stbi__context *s)
+{
+   if (s->img_buffer < s->img_buffer_end)
+      return *s->img_buffer++;
+   if (s->read_from_callbacks) {
+      stbi__refill_buffer(s);
+      return *s->img_buffer++;
+   }
+   return 0;
+}
+
+stbi_inline static int stbi__at_eof(stbi__context *s)
+{
+   if (s->io.read) {
+      if (!(s->io.eof)(s->io_user_data)) return 0;
+      // if feof() is true, check if buffer = end
+      // special case: we've only got the special 0 character at the end
+      if (s->read_from_callbacks == 0) return 1;
+   }
+
+   return s->img_buffer >= s->img_buffer_end;
+}
+
+static void stbi__skip(stbi__context *s, int n)
+{
+   if (n < 0) {
+      s->img_buffer = s->img_buffer_end;
+      return;
+   }
+   if (s->io.read) {
+      int blen = (int) (s->img_buffer_end - s->img_buffer);
+      if (blen < n) {
+         s->img_buffer = s->img_buffer_end;
+         (s->io.skip)(s->io_user_data, n - blen);
+         return;
+      }
+   }
+   s->img_buffer += n;
+}
+
+static int stbi__getn(stbi__context *s, stbi_uc *buffer, int n)
+{
+   if (s->io.read) {
+      int blen = (int) (s->img_buffer_end - s->img_buffer);
+      if (blen < n) {
+         int res, count;
+
+         memcpy(buffer, s->img_buffer, blen);
+
+         count = (s->io.read)(s->io_user_data, (char*) buffer + blen, n - blen);
+         res = (count == (n-blen));
+         s->img_buffer = s->img_buffer_end;
+         return res;
+      }
+   }
+
+   if (s->img_buffer+n <= s->img_buffer_end) {
+      memcpy(buffer, s->img_buffer, n);
+      s->img_buffer += n;
+      return 1;
+   } else
+      return 0;
+}
+
+static int stbi__get16be(stbi__context *s)
+{
+   int z = stbi__get8(s);
+   return (z << 8) + stbi__get8(s);
+}
+
+static stbi__uint32 stbi__get32be(stbi__context *s)
+{
+   stbi__uint32 z = stbi__get16be(s);
+   return (z << 16) + stbi__get16be(s);
+}
+
+#if defined(STBI_NO_BMP) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF)
+// nothing
+#else
+static int stbi__get16le(stbi__context *s)
+{
+   int z = stbi__get8(s);
+   return z + (stbi__get8(s) << 8);
+}
+#endif
+
+#ifndef STBI_NO_BMP
+static stbi__uint32 stbi__get32le(stbi__context *s)
+{
+   stbi__uint32 z = stbi__get16le(s);
+   return z + (stbi__get16le(s) << 16);
+}
+#endif
+
+#define STBI__BYTECAST(x)  ((stbi_uc) ((x) & 255))  // truncate int to byte without warnings
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+//  generic converter from built-in img_n to req_comp
+//    individual types do this automatically as much as possible (e.g. jpeg
+//    does all cases internally since it needs to colorspace convert anyway,
+//    and it never has alpha, so very few cases ). png can automatically
+//    interleave an alpha=255 channel, but falls back to this for other cases
+//
+//  assume data buffer is malloced, so malloc a new one and free that one
+//  only failure mode is malloc failing
+
+static stbi_uc stbi__compute_y(int r, int g, int b)
+{
+   return (stbi_uc) (((r*77) + (g*150) +  (29*b)) >> 8);
+}
+
+static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int req_comp, unsigned int x, unsigned int y)
+{
+   int i,j;
+   unsigned char *good;
+
+   if (req_comp == img_n) return data;
+   STBI_ASSERT(req_comp >= 1 && req_comp <= 4);
+
+   good = (unsigned char *) stbi__malloc_mad3(req_comp, x, y, 0);
+   if (good == NULL) {
+      STBI_FREE(data);
+      return stbi__errpuc("outofmem", "Out of memory");
+   }
+
+   for (j=0; j < (int) y; ++j) {
+      unsigned char *src  = data + j * x * img_n   ;
+      unsigned char *dest = good + j * x * req_comp;
+
+      #define STBI__COMBO(a,b)  ((a)*8+(b))
+      #define STBI__CASE(a,b)   case STBI__COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b)
+      // convert source image with img_n components to one with req_comp components;
+      // avoid switch per pixel, so use switch per scanline and massive macros
+      switch (STBI__COMBO(img_n, req_comp)) {
+         STBI__CASE(1,2) { dest[0]=src[0], dest[1]=255;                                     } break;
+         STBI__CASE(1,3) { dest[0]=dest[1]=dest[2]=src[0];                                  } break;
+         STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0], dest[3]=255;                     } break;
+         STBI__CASE(2,1) { dest[0]=src[0];                                                  } break;
+         STBI__CASE(2,3) { dest[0]=dest[1]=dest[2]=src[0];                                  } break;
+         STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0], dest[3]=src[1];                  } break;
+         STBI__CASE(3,4) { dest[0]=src[0],dest[1]=src[1],dest[2]=src[2],dest[3]=255;        } break;
+         STBI__CASE(3,1) { dest[0]=stbi__compute_y(src[0],src[1],src[2]);                   } break;
+         STBI__CASE(3,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]), dest[1] = 255;    } break;
+         STBI__CASE(4,1) { dest[0]=stbi__compute_y(src[0],src[1],src[2]);                   } break;
+         STBI__CASE(4,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]), dest[1] = src[3]; } break;
+         STBI__CASE(4,3) { dest[0]=src[0],dest[1]=src[1],dest[2]=src[2];                    } break;
+         default: STBI_ASSERT(0);
+      }
+      #undef STBI__CASE
+   }
+
+   STBI_FREE(data);
+   return good;
+}
+
+static stbi__uint16 stbi__compute_y_16(int r, int g, int b)
+{
+   return (stbi__uint16) (((r*77) + (g*150) +  (29*b)) >> 8);
+}
+
+static stbi__uint16 *stbi__convert_format16(stbi__uint16 *data, int img_n, int req_comp, unsigned int x, unsigned int y)
+{
+   int i,j;
+   stbi__uint16 *good;
+
+   if (req_comp == img_n) return data;
+   STBI_ASSERT(req_comp >= 1 && req_comp <= 4);
+
+   good = (stbi__uint16 *) stbi__malloc(req_comp * x * y * 2);
+   if (good == NULL) {
+      STBI_FREE(data);
+      return (stbi__uint16 *) stbi__errpuc("outofmem", "Out of memory");
+   }
+
+   for (j=0; j < (int) y; ++j) {
+      stbi__uint16 *src  = data + j * x * img_n   ;
+      stbi__uint16 *dest = good + j * x * req_comp;
+
+      #define STBI__COMBO(a,b)  ((a)*8+(b))
+      #define STBI__CASE(a,b)   case STBI__COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b)
+      // convert source image with img_n components to one with req_comp components;
+      // avoid switch per pixel, so use switch per scanline and massive macros
+      switch (STBI__COMBO(img_n, req_comp)) {
+         STBI__CASE(1,2) { dest[0]=src[0], dest[1]=0xffff;                                     } break;
+         STBI__CASE(1,3) { dest[0]=dest[1]=dest[2]=src[0];                                     } break;
+         STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0], dest[3]=0xffff;                     } break;
+         STBI__CASE(2,1) { dest[0]=src[0];                                                     } break;
+         STBI__CASE(2,3) { dest[0]=dest[1]=dest[2]=src[0];                                     } break;
+         STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0], dest[3]=src[1];                     } break;
+         STBI__CASE(3,4) { dest[0]=src[0],dest[1]=src[1],dest[2]=src[2],dest[3]=0xffff;        } break;
+         STBI__CASE(3,1) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]);                   } break;
+         STBI__CASE(3,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]), dest[1] = 0xffff; } break;
+         STBI__CASE(4,1) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]);                   } break;
+         STBI__CASE(4,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]), dest[1] = src[3]; } break;
+         STBI__CASE(4,3) { dest[0]=src[0],dest[1]=src[1],dest[2]=src[2];                       } break;
+         default: STBI_ASSERT(0);
+      }
+      #undef STBI__CASE
+   }
+
+   STBI_FREE(data);
+   return good;
+}
+
+#ifndef STBI_NO_LINEAR
+static float   *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp)
+{
+   int i,k,n;
+   float *output;
+   if (!data) return NULL;
+   output = (float *) stbi__malloc_mad4(x, y, comp, sizeof(float), 0);
+   if (output == NULL) { STBI_FREE(data); return stbi__errpf("outofmem", "Out of memory"); }
+   // compute number of non-alpha components
+   if (comp & 1) n = comp; else n = comp-1;
+   for (i=0; i < x*y; ++i) {
+      for (k=0; k < n; ++k) {
+         output[i*comp + k] = (float) (pow(data[i*comp+k]/255.0f, stbi__l2h_gamma) * stbi__l2h_scale);
+      }
+      if (k < comp) output[i*comp + k] = data[i*comp+k]/255.0f;
+   }
+   STBI_FREE(data);
+   return output;
+}
+#endif
+
+#ifndef STBI_NO_HDR
+#define stbi__float2int(x)   ((int) (x))
+static stbi_uc *stbi__hdr_to_ldr(float   *data, int x, int y, int comp)
+{
+   int i,k,n;
+   stbi_uc *output;
+   if (!data) return NULL;
+   output = (stbi_uc *) stbi__malloc_mad3(x, y, comp, 0);
+   if (output == NULL) { STBI_FREE(data); return stbi__errpuc("outofmem", "Out of memory"); }
+   // compute number of non-alpha components
+   if (comp & 1) n = comp; else n = comp-1;
+   for (i=0; i < x*y; ++i) {
+      for (k=0; k < n; ++k) {
+         float z = (float) pow(data[i*comp+k]*stbi__h2l_scale_i, stbi__h2l_gamma_i) * 255 + 0.5f;
+         if (z < 0) z = 0;
+         if (z > 255) z = 255;
+         output[i*comp + k] = (stbi_uc) stbi__float2int(z);
+      }
+      if (k < comp) {
+         float z = data[i*comp+k] * 255 + 0.5f;
+         if (z < 0) z = 0;
+         if (z > 255) z = 255;
+         output[i*comp + k] = (stbi_uc) stbi__float2int(z);
+      }
+   }
+   STBI_FREE(data);
+   return output;
+}
+#endif
+
+//////////////////////////////////////////////////////////////////////////////
+//
+//  "baseline" JPEG/JFIF decoder
+//
+//    simple implementation
+//      - doesn't support delayed output of y-dimension
+//      - simple interface (only one output format: 8-bit interleaved RGB)
+//      - doesn't try to recover corrupt jpegs
+//      - doesn't allow partial loading, loading multiple at once
+//      - still fast on x86 (copying globals into locals doesn't help x86)
+//      - allocates lots of intermediate memory (full size of all components)
+//        - non-interleaved case requires this anyway
+//        - allows good upsampling (see next)
+//    high-quality
+//      - upsampled channels are bilinearly interpolated, even across blocks
+//      - quality integer IDCT derived from IJG's 'slow'
+//    performance
+//      - fast huffman; reasonable integer IDCT
+//      - some SIMD kernels for common paths on targets with SSE2/NEON
+//      - uses a lot of intermediate memory, could cache poorly
+
+#ifndef STBI_NO_JPEG
+
+// huffman decoding acceleration
+#define FAST_BITS   9  // larger handles more cases; smaller stomps less cache
+
+typedef struct
+{
+   stbi_uc  fast[1 << FAST_BITS];
+   // weirdly, repacking this into AoS is a 10% speed loss, instead of a win
+   stbi__uint16 code[256];
+   stbi_uc  values[256];
+   stbi_uc  size[257];
+   unsigned int maxcode[18];
+   int    delta[17];   // old 'firstsymbol' - old 'firstcode'
+} stbi__huffman;
+
+typedef struct
+{
+   stbi__context *s;
+   stbi__huffman huff_dc[4];
+   stbi__huffman huff_ac[4];
+   stbi_uc dequant[4][64];
+   stbi__int16 fast_ac[4][1 << FAST_BITS];
+
+// sizes for components, interleaved MCUs
+   int img_h_max, img_v_max;
+   int img_mcu_x, img_mcu_y;
+   int img_mcu_w, img_mcu_h;
+
+// definition of jpeg image component
+   struct
+   {
+      int id;
+      int h,v;
+      int tq;
+      int hd,ha;
+      int dc_pred;
+
+      int x,y,w2,h2;
+      stbi_uc *data;
+      void *raw_data, *raw_coeff;
+      stbi_uc *linebuf;
+      short   *coeff;   // progressive only
+      int      coeff_w, coeff_h; // number of 8x8 coefficient blocks
+   } img_comp[4];
+
+   stbi__uint32   code_buffer; // jpeg entropy-coded buffer
+   int            code_bits;   // number of valid bits
+   unsigned char  marker;      // marker seen while filling entropy buffer
+   int            nomore;      // flag if we saw a marker so must stop
+
+   int            progressive;
+   int            spec_start;
+   int            spec_end;
+   int            succ_high;
+   int            succ_low;
+   int            eob_run;
+   int            rgb;
+
+   int scan_n, order[4];
+   int restart_interval, todo;
+
+// kernels
+   void (*idct_block_kernel)(stbi_uc *out, int out_stride, short data[64]);
+   void (*YCbCr_to_RGB_kernel)(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step);
+   stbi_uc *(*resample_row_hv_2_kernel)(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs);
+} stbi__jpeg;
+
+static int stbi__build_huffman(stbi__huffman *h, int *count)
+{
+   int i,j,k=0,code;
+   // build size list for each symbol (from JPEG spec)
+   for (i=0; i < 16; ++i)
+      for (j=0; j < count[i]; ++j)
+         h->size[k++] = (stbi_uc) (i+1);
+   h->size[k] = 0;
+
+   // compute actual symbols (from jpeg spec)
+   code = 0;
+   k = 0;
+   for(j=1; j <= 16; ++j) {
+      // compute delta to add to code to compute symbol id
+      h->delta[j] = k - code;
+      if (h->size[k] == j) {
+         while (h->size[k] == j)
+            h->code[k++] = (stbi__uint16) (code++);
+         if (code-1 >= (1 << j)) return stbi__err("bad code lengths","Corrupt JPEG");
+      }
+      // compute largest code + 1 for this size, preshifted as needed later
+      h->maxcode[j] = code << (16-j);
+      code <<= 1;
+   }
+   h->maxcode[j] = 0xffffffff;
+
+   // build non-spec acceleration table; 255 is flag for not-accelerated
+   memset(h->fast, 255, 1 << FAST_BITS);
+   for (i=0; i < k; ++i) {
+      int s = h->size[i];
+      if (s <= FAST_BITS) {
+         int c = h->code[i] << (FAST_BITS-s);
+         int m = 1 << (FAST_BITS-s);
+         for (j=0; j < m; ++j) {
+            h->fast[c+j] = (stbi_uc) i;
+         }
+      }
+   }
+   return 1;
+}
+
+// build a table that decodes both magnitude and value of small ACs in
+// one go.
+static void stbi__build_fast_ac(stbi__int16 *fast_ac, stbi__huffman *h)
+{
+   int i;
+   for (i=0; i < (1 << FAST_BITS); ++i) {
+      stbi_uc fast = h->fast[i];
+      fast_ac[i] = 0;
+      if (fast < 255) {
+         int rs = h->values[fast];
+         int run = (rs >> 4) & 15;
+         int magbits = rs & 15;
+         int len = h->size[fast];
+
+         if (magbits && len + magbits <= FAST_BITS) {
+            // magnitude code followed by receive_extend code
+            int k = ((i << len) & ((1 << FAST_BITS) - 1)) >> (FAST_BITS - magbits);
+            int m = 1 << (magbits - 1);
+            if (k < m) k += (-1 << magbits) + 1;
+            // if the result is small enough, we can fit it in fast_ac table
+            if (k >= -128 && k <= 127)
+               fast_ac[i] = (stbi__int16) ((k << 8) + (run << 4) + (len + magbits));
+         }
+      }
+   }
+}
+
+static void stbi__grow_buffer_unsafe(stbi__jpeg *j)
+{
+   do {
+      int b = j->nomore ? 0 : stbi__get8(j->s);
+      if (b == 0xff) {
+         int c = stbi__get8(j->s);
+         if (c != 0) {
+            j->marker = (unsigned char) c;
+            j->nomore = 1;
+            return;
+         }
+      }
+      j->code_buffer |= b << (24 - j->code_bits);
+      j->code_bits += 8;
+   } while (j->code_bits <= 24);
+}
+
+// (1 << n) - 1
+static stbi__uint32 stbi__bmask[17]={0,1,3,7,15,31,63,127,255,511,1023,2047,4095,8191,16383,32767,65535};
+
+// decode a jpeg huffman value from the bitstream
+stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h)
+{
+   unsigned int temp;
+   int c,k;
+
+   if (j->code_bits < 16) stbi__grow_buffer_unsafe(j);
+
+   // look at the top FAST_BITS and determine what symbol ID it is,
+   // if the code is <= FAST_BITS
+   c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1);
+   k = h->fast[c];
+   if (k < 255) {
+      int s = h->size[k];
+      if (s > j->code_bits)
+         return -1;
+      j->code_buffer <<= s;
+      j->code_bits -= s;
+      return h->values[k];
+   }
+
+   // naive test is to shift the code_buffer down so k bits are
+   // valid, then test against maxcode. To speed this up, we've
+   // preshifted maxcode left so that it has (16-k) 0s at the
+   // end; in other words, regardless of the number of bits, it
+   // wants to be compared against something shifted to have 16;
+   // that way we don't need to shift inside the loop.
+   temp = j->code_buffer >> 16;
+   for (k=FAST_BITS+1 ; ; ++k)
+      if (temp < h->maxcode[k])
+         break;
+   if (k == 17) {
+      // error! code not found
+      j->code_bits -= 16;
+      return -1;
+   }
+
+   if (k > j->code_bits)
+      return -1;
+
+   // convert the huffman code to the symbol id
+   c = ((j->code_buffer >> (32 - k)) & stbi__bmask[k]) + h->delta[k];
+   STBI_ASSERT((((j->code_buffer) >> (32 - h->size[c])) & stbi__bmask[h->size[c]]) == h->code[c]);
+
+   // convert the id to a symbol
+   j->code_bits -= k;
+   j->code_buffer <<= k;
+   return h->values[c];
+}
+
+// bias[n] = (-1<<n) + 1
+static int const stbi__jbias[16] = {0,-1,-3,-7,-15,-31,-63,-127,-255,-511,-1023,-2047,-4095,-8191,-16383,-32767};
+
+// combined JPEG 'receive' and JPEG 'extend', since baseline
+// always extends everything it receives.
+stbi_inline static int stbi__extend_receive(stbi__jpeg *j, int n)
+{
+   unsigned int k;
+   int sgn;
+   if (j->code_bits < n) stbi__grow_buffer_unsafe(j);
+
+   sgn = (stbi__int32)j->code_buffer >> 31; // sign bit is always in MSB
+   k = stbi_lrot(j->code_buffer, n);
+   STBI_ASSERT(n >= 0 && n < (int) (sizeof(stbi__bmask)/sizeof(*stbi__bmask)));
+   j->code_buffer = k & ~stbi__bmask[n];
+   k &= stbi__bmask[n];
+   j->code_bits -= n;
+   return k + (stbi__jbias[n] & ~sgn);
+}
+
+// get some unsigned bits
+stbi_inline static int stbi__jpeg_get_bits(stbi__jpeg *j, int n)
+{
+   unsigned int k;
+   if (j->code_bits < n) stbi__grow_buffer_unsafe(j);
+   k = stbi_lrot(j->code_buffer, n);
+   j->code_buffer = k & ~stbi__bmask[n];
+   k &= stbi__bmask[n];
+   j->code_bits -= n;
+   return k;
+}
+
+stbi_inline static int stbi__jpeg_get_bit(stbi__jpeg *j)
+{
+   unsigned int k;
+   if (j->code_bits < 1) stbi__grow_buffer_unsafe(j);
+   k = j->code_buffer;
+   j->code_buffer <<= 1;
+   --j->code_bits;
+   return k & 0x80000000;
+}
+
+// given a value that's at position X in the zigzag stream,
+// where does it appear in the 8x8 matrix coded as row-major?
+static stbi_uc stbi__jpeg_dezigzag[64+15] =
+{
+    0,  1,  8, 16,  9,  2,  3, 10,
+   17, 24, 32, 25, 18, 11,  4,  5,
+   12, 19, 26, 33, 40, 48, 41, 34,
+   27, 20, 13,  6,  7, 14, 21, 28,
+   35, 42, 49, 56, 57, 50, 43, 36,
+   29, 22, 15, 23, 30, 37, 44, 51,
+   58, 59, 52, 45, 38, 31, 39, 46,
+   53, 60, 61, 54, 47, 55, 62, 63,
+   // let corrupt input sample past end
+   63, 63, 63, 63, 63, 63, 63, 63,
+   63, 63, 63, 63, 63, 63, 63
+};
+
+// decode one 64-entry block--
+static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman *hdc, stbi__huffman *hac, stbi__int16 *fac, int b, stbi_uc *dequant)
+{
+   int diff,dc,k;
+   int t;
+
+   if (j->code_bits < 16) stbi__grow_buffer_unsafe(j);
+   t = stbi__jpeg_huff_decode(j, hdc);
+   if (t < 0) return stbi__err("bad huffman code","Corrupt JPEG");
+
+   // 0 all the ac values now so we can do it 32-bits at a time
+   memset(data,0,64*sizeof(data[0]));
+
+   diff = t ? stbi__extend_receive(j, t) : 0;
+   dc = j->img_comp[b].dc_pred + diff;
+   j->img_comp[b].dc_pred = dc;
+   data[0] = (short) (dc * dequant[0]);
+
+   // decode AC components, see JPEG spec
+   k = 1;
+   do {
+      unsigned int zig;
+      int c,r,s;
+      if (j->code_bits < 16) stbi__grow_buffer_unsafe(j);
+      c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1);
+      r = fac[c];
+      if (r) { // fast-AC path
+         k += (r >> 4) & 15; // run
+         s = r & 15; // combined length
+         j->code_buffer <<= s;
+         j->code_bits -= s;
+         // decode into unzigzag'd location
+         zig = stbi__jpeg_dezigzag[k++];
+         data[zig] = (short) ((r >> 8) * dequant[zig]);
+      } else {
+         int rs = stbi__jpeg_huff_decode(j, hac);
+         if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG");
+         s = rs & 15;
+         r = rs >> 4;
+         if (s == 0) {
+            if (rs != 0xf0) break; // end block
+            k += 16;
+         } else {
+            k += r;
+            // decode into unzigzag'd location
+            zig = stbi__jpeg_dezigzag[k++];
+            data[zig] = (short) (stbi__extend_receive(j,s) * dequant[zig]);
+         }
+      }
+   } while (k < 64);
+   return 1;
+}
+
+static int stbi__jpeg_decode_block_prog_dc(stbi__jpeg *j, short data[64], stbi__huffman *hdc, int b)
+{
+   int diff,dc;
+   int t;
+   if (j->spec_end != 0) return stbi__err("can't merge dc and ac", "Corrupt JPEG");
+
+   if (j->code_bits < 16) stbi__grow_buffer_unsafe(j);
+
+   if (j->succ_high == 0) {
+      // first scan for DC coefficient, must be first
+      memset(data,0,64*sizeof(data[0])); // 0 all the ac values now
+      t = stbi__jpeg_huff_decode(j, hdc);
+      diff = t ? stbi__extend_receive(j, t) : 0;
+
+      dc = j->img_comp[b].dc_pred + diff;
+      j->img_comp[b].dc_pred = dc;
+      data[0] = (short) (dc << j->succ_low);
+   } else {
+      // refinement scan for DC coefficient
+      if (stbi__jpeg_get_bit(j))
+         data[0] += (short) (1 << j->succ_low);
+   }
+   return 1;
+}
+
+// @OPTIMIZE: store non-zigzagged during the decode passes,
+// and only de-zigzag when dequantizing
+static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__huffman *hac, stbi__int16 *fac)
+{
+   int k;
+   if (j->spec_start == 0) return stbi__err("can't merge dc and ac", "Corrupt JPEG");
+
+   if (j->succ_high == 0) {
+      int shift = j->succ_low;
+
+      if (j->eob_run) {
+         --j->eob_run;
+         return 1;
+      }
+
+      k = j->spec_start;
+      do {
+         unsigned int zig;
+         int c,r,s;
+         if (j->code_bits < 16) stbi__grow_buffer_unsafe(j);
+         c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1);
+         r = fac[c];
+         if (r) { // fast-AC path
+            k += (r >> 4) & 15; // run
+            s = r & 15; // combined length
+            j->code_buffer <<= s;
+            j->code_bits -= s;
+            zig = stbi__jpeg_dezigzag[k++];
+            data[zig] = (short) ((r >> 8) << shift);
+         } else {
+            int rs = stbi__jpeg_huff_decode(j, hac);
+            if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG");
+            s = rs & 15;
+            r = rs >> 4;
+            if (s == 0) {
+               if (r < 15) {
+                  j->eob_run = (1 << r);
+                  if (r)
+                     j->eob_run += stbi__jpeg_get_bits(j, r);
+                  --j->eob_run;
+                  break;
+               }
+               k += 16;
+            } else {
+               k += r;
+               zig = stbi__jpeg_dezigzag[k++];
+               data[zig] = (short) (stbi__extend_receive(j,s) << shift);
+            }
+         }
+      } while (k <= j->spec_end);
+   } else {
+      // refinement scan for these AC coefficients
+
+      short bit = (short) (1 << j->succ_low);
+
+      if (j->eob_run) {
+         --j->eob_run;
+         for (k = j->spec_start; k <= j->spec_end; ++k) {
+            short *p = &data[stbi__jpeg_dezigzag[k]];
+            if (*p != 0)
+               if (stbi__jpeg_get_bit(j))
+                  if ((*p & bit)==0) {
+                     if (*p > 0)
+                        *p += bit;
+                     else
+                        *p -= bit;
+                  }
+         }
+      } else {
+         k = j->spec_start;
+         do {
+            int r,s;
+            int rs = stbi__jpeg_huff_decode(j, hac); // @OPTIMIZE see if we can use the fast path here, advance-by-r is so slow, eh
+            if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG");
+            s = rs & 15;
+            r = rs >> 4;
+            if (s == 0) {
+               if (r < 15) {
+                  j->eob_run = (1 << r) - 1;
+                  if (r)
+                     j->eob_run += stbi__jpeg_get_bits(j, r);
+                  r = 64; // force end of block
+               } else {
+                  // r=15 s=0 should write 16 0s, so we just do
+                  // a run of 15 0s and then write s (which is 0),
+                  // so we don't have to do anything special here
+               }
+            } else {
+               if (s != 1) return stbi__err("bad huffman code", "Corrupt JPEG");
+               // sign bit
+               if (stbi__jpeg_get_bit(j))
+                  s = bit;
+               else
+                  s = -bit;
+            }
+
+            // advance by r
+            while (k <= j->spec_end) {
+               short *p = &data[stbi__jpeg_dezigzag[k++]];
+               if (*p != 0) {
+                  if (stbi__jpeg_get_bit(j))
+                     if ((*p & bit)==0) {
+                        if (*p > 0)
+                           *p += bit;
+                        else
+                           *p -= bit;
+                     }
+               } else {
+                  if (r == 0) {
+                     *p = (short) s;
+                     break;
+                  }
+                  --r;
+               }
+            }
+         } while (k <= j->spec_end);
+      }
+   }
+   return 1;
+}
+
+// take a -128..127 value and stbi__clamp it and convert to 0..255
+stbi_inline static stbi_uc stbi__clamp(int x)
+{
+   // trick to use a single test to catch both cases
+   if ((unsigned int) x > 255) {
+      if (x < 0) return 0;
+      if (x > 255) return 255;
+   }
+   return (stbi_uc) x;
+}
+
+#define stbi__f2f(x)  ((int) (((x) * 4096 + 0.5)))
+#define stbi__fsh(x)  ((x) << 12)
+
+// derived from jidctint -- DCT_ISLOW
+#define STBI__IDCT_1D(s0,s1,s2,s3,s4,s5,s6,s7) \
+   int t0,t1,t2,t3,p1,p2,p3,p4,p5,x0,x1,x2,x3; \
+   p2 = s2;                                    \
+   p3 = s6;                                    \
+   p1 = (p2+p3) * stbi__f2f(0.5411961f);       \
+   t2 = p1 + p3*stbi__f2f(-1.847759065f);      \
+   t3 = p1 + p2*stbi__f2f( 0.765366865f);      \
+   p2 = s0;                                    \
+   p3 = s4;                                    \
+   t0 = stbi__fsh(p2+p3);                      \
+   t1 = stbi__fsh(p2-p3);                      \
+   x0 = t0+t3;                                 \
+   x3 = t0-t3;                                 \
+   x1 = t1+t2;                                 \
+   x2 = t1-t2;                                 \
+   t0 = s7;                                    \
+   t1 = s5;                                    \
+   t2 = s3;                                    \
+   t3 = s1;                                    \
+   p3 = t0+t2;                                 \
+   p4 = t1+t3;                                 \
+   p1 = t0+t3;                                 \
+   p2 = t1+t2;                                 \
+   p5 = (p3+p4)*stbi__f2f( 1.175875602f);      \
+   t0 = t0*stbi__f2f( 0.298631336f);           \
+   t1 = t1*stbi__f2f( 2.053119869f);           \
+   t2 = t2*stbi__f2f( 3.072711026f);           \
+   t3 = t3*stbi__f2f( 1.501321110f);           \
+   p1 = p5 + p1*stbi__f2f(-0.899976223f);      \
+   p2 = p5 + p2*stbi__f2f(-2.562915447f);      \
+   p3 = p3*stbi__f2f(-1.961570560f);           \
+   p4 = p4*stbi__f2f(-0.390180644f);           \
+   t3 += p1+p4;                                \
+   t2 += p2+p3;                                \
+   t1 += p2+p4;                                \
+   t0 += p1+p3;
+
+static void stbi__idct_block(stbi_uc *out, int out_stride, short data[64])
+{
+   int i,val[64],*v=val;
+   stbi_uc *o;
+   short *d = data;
+
+   // columns
+   for (i=0; i < 8; ++i,++d, ++v) {
+      // if all zeroes, shortcut -- this avoids dequantizing 0s and IDCTing
+      if (d[ 8]==0 && d[16]==0 && d[24]==0 && d[32]==0
+           && d[40]==0 && d[48]==0 && d[56]==0) {
+         //    no shortcut                 0     seconds
+         //    (1|2|3|4|5|6|7)==0          0     seconds
+         //    all separate               -0.047 seconds
+         //    1 && 2|3 && 4|5 && 6|7:    -0.047 seconds
+         int dcterm = d[0] << 2;
+         v[0] = v[8] = v[16] = v[24] = v[32] = v[40] = v[48] = v[56] = dcterm;
+      } else {
+         STBI__IDCT_1D(d[ 0],d[ 8],d[16],d[24],d[32],d[40],d[48],d[56])
+         // constants scaled things up by 1<<12; let's bring them back
+         // down, but keep 2 extra bits of precision
+         x0 += 512; x1 += 512; x2 += 512; x3 += 512;
+         v[ 0] = (x0+t3) >> 10;
+         v[56] = (x0-t3) >> 10;
+         v[ 8] = (x1+t2) >> 10;
+         v[48] = (x1-t2) >> 10;
+         v[16] = (x2+t1) >> 10;
+         v[40] = (x2-t1) >> 10;
+         v[24] = (x3+t0) >> 10;
+         v[32] = (x3-t0) >> 10;
+      }
+   }
+
+   for (i=0, v=val, o=out; i < 8; ++i,v+=8,o+=out_stride) {
+      // no fast case since the first 1D IDCT spread components out
+      STBI__IDCT_1D(v[0],v[1],v[2],v[3],v[4],v[5],v[6],v[7])
+      // constants scaled things up by 1<<12, plus we had 1<<2 from first
+      // loop, plus horizontal and vertical each scale by sqrt(8) so together
+      // we've got an extra 1<<3, so 1<<17 total we need to remove.
+      // so we want to round that, which means adding 0.5 * 1<<17,
+      // aka 65536. Also, we'll end up with -128 to 127 that we want
+      // to encode as 0..255 by adding 128, so we'll add that before the shift
+      x0 += 65536 + (128<<17);
+      x1 += 65536 + (128<<17);
+      x2 += 65536 + (128<<17);
+      x3 += 65536 + (128<<17);
+      // tried computing the shifts into temps, or'ing the temps to see
+      // if any were out of range, but that was slower
+      o[0] = stbi__clamp((x0+t3) >> 17);
+      o[7] = stbi__clamp((x0-t3) >> 17);
+      o[1] = stbi__clamp((x1+t2) >> 17);
+      o[6] = stbi__clamp((x1-t2) >> 17);
+      o[2] = stbi__clamp((x2+t1) >> 17);
+      o[5] = stbi__clamp((x2-t1) >> 17);
+      o[3] = stbi__clamp((x3+t0) >> 17);
+      o[4] = stbi__clamp((x3-t0) >> 17);
+   }
+}
+
+#ifdef STBI_SSE2
+// sse2 integer IDCT. not the fastest possible implementation but it
+// produces bit-identical results to the generic C version so it's
+// fully "transparent".
+static void stbi__idct_simd(stbi_uc *out, int out_stride, short data[64])
+{
+   // This is constructed to match our regular (generic) integer IDCT exactly.
+   __m128i row0, row1, row2, row3, row4, row5, row6, row7;
+   __m128i tmp;
+
+   // dot product constant: even elems=x, odd elems=y
+   #define dct_const(x,y)  _mm_setr_epi16((x),(y),(x),(y),(x),(y),(x),(y))
+
+   // out(0) = c0[even]*x + c0[odd]*y   (c0, x, y 16-bit, out 32-bit)
+   // out(1) = c1[even]*x + c1[odd]*y
+   #define dct_rot(out0,out1, x,y,c0,c1) \
+      __m128i c0##lo = _mm_unpacklo_epi16((x),(y)); \
+      __m128i c0##hi = _mm_unpackhi_epi16((x),(y)); \
+      __m128i out0##_l = _mm_madd_epi16(c0##lo, c0); \
+      __m128i out0##_h = _mm_madd_epi16(c0##hi, c0); \
+      __m128i out1##_l = _mm_madd_epi16(c0##lo, c1); \
+      __m128i out1##_h = _mm_madd_epi16(c0##hi, c1)
+
+   // out = in << 12  (in 16-bit, out 32-bit)
+   #define dct_widen(out, in) \
+      __m128i out##_l = _mm_srai_epi32(_mm_unpacklo_epi16(_mm_setzero_si128(), (in)), 4); \
+      __m128i out##_h = _mm_srai_epi32(_mm_unpackhi_epi16(_mm_setzero_si128(), (in)), 4)
+
+   // wide add
+   #define dct_wadd(out, a, b) \
+      __m128i out##_l = _mm_add_epi32(a##_l, b##_l); \
+      __m128i out##_h = _mm_add_epi32(a##_h, b##_h)
+
+   // wide sub
+   #define dct_wsub(out, a, b) \
+      __m128i out##_l = _mm_sub_epi32(a##_l, b##_l); \
+      __m128i out##_h = _mm_sub_epi32(a##_h, b##_h)
+
+   // butterfly a/b, add bias, then shift by "s" and pack
+   #define dct_bfly32o(out0, out1, a,b,bias,s) \
+      { \
+         __m128i abiased_l = _mm_add_epi32(a##_l, bias); \
+         __m128i abiased_h = _mm_add_epi32(a##_h, bias); \
+         dct_wadd(sum, abiased, b); \
+         dct_wsub(dif, abiased, b); \
+         out0 = _mm_packs_epi32(_mm_srai_epi32(sum_l, s), _mm_srai_epi32(sum_h, s)); \
+         out1 = _mm_packs_epi32(_mm_srai_epi32(dif_l, s), _mm_srai_epi32(dif_h, s)); \
+      }
+
+   // 8-bit interleave step (for transposes)
+   #define dct_interleave8(a, b) \
+      tmp = a; \
+      a = _mm_unpacklo_epi8(a, b); \
+      b = _mm_unpackhi_epi8(tmp, b)
+
+   // 16-bit interleave step (for transposes)
+   #define dct_interleave16(a, b) \
+      tmp = a; \
+      a = _mm_unpacklo_epi16(a, b); \
+      b = _mm_unpackhi_epi16(tmp, b)
+
+   #define dct_pass(bias,shift) \
+      { \
+         /* even part */ \
+         dct_rot(t2e,t3e, row2,row6, rot0_0,rot0_1); \
+         __m128i sum04 = _mm_add_epi16(row0, row4); \
+         __m128i dif04 = _mm_sub_epi16(row0, row4); \
+         dct_widen(t0e, sum04); \
+         dct_widen(t1e, dif04); \
+         dct_wadd(x0, t0e, t3e); \
+         dct_wsub(x3, t0e, t3e); \
+         dct_wadd(x1, t1e, t2e); \
+         dct_wsub(x2, t1e, t2e); \
+         /* odd part */ \
+         dct_rot(y0o,y2o, row7,row3, rot2_0,rot2_1); \
+         dct_rot(y1o,y3o, row5,row1, rot3_0,rot3_1); \
+         __m128i sum17 = _mm_add_epi16(row1, row7); \
+         __m128i sum35 = _mm_add_epi16(row3, row5); \
+         dct_rot(y4o,y5o, sum17,sum35, rot1_0,rot1_1); \
+         dct_wadd(x4, y0o, y4o); \
+         dct_wadd(x5, y1o, y5o); \
+         dct_wadd(x6, y2o, y5o); \
+         dct_wadd(x7, y3o, y4o); \
+         dct_bfly32o(row0,row7, x0,x7,bias,shift); \
+         dct_bfly32o(row1,row6, x1,x6,bias,shift); \
+         dct_bfly32o(row2,row5, x2,x5,bias,shift); \
+         dct_bfly32o(row3,row4, x3,x4,bias,shift); \
+      }
+
+   __m128i rot0_0 = dct_const(stbi__f2f(0.5411961f), stbi__f2f(0.5411961f) + stbi__f2f(-1.847759065f));
+   __m128i rot0_1 = dct_const(stbi__f2f(0.5411961f) + stbi__f2f( 0.765366865f), stbi__f2f(0.5411961f));
+   __m128i rot1_0 = dct_const(stbi__f2f(1.175875602f) + stbi__f2f(-0.899976223f), stbi__f2f(1.175875602f));
+   __m128i rot1_1 = dct_const(stbi__f2f(1.175875602f), stbi__f2f(1.175875602f) + stbi__f2f(-2.562915447f));
+   __m128i rot2_0 = dct_const(stbi__f2f(-1.961570560f) + stbi__f2f( 0.298631336f), stbi__f2f(-1.961570560f));
+   __m128i rot2_1 = dct_const(stbi__f2f(-1.961570560f), stbi__f2f(-1.961570560f) + stbi__f2f( 3.072711026f));
+   __m128i rot3_0 = dct_const(stbi__f2f(-0.390180644f) + stbi__f2f( 2.053119869f), stbi__f2f(-0.390180644f));
+   __m128i rot3_1 = dct_const(stbi__f2f(-0.390180644f), stbi__f2f(-0.390180644f) + stbi__f2f( 1.501321110f));
+
+   // rounding biases in column/row passes, see stbi__idct_block for explanation.
+   __m128i bias_0 = _mm_set1_epi32(512);
+   __m128i bias_1 = _mm_set1_epi32(65536 + (128<<17));
+
+   // load
+   row0 = _mm_load_si128((const __m128i *) (data + 0*8));
+   row1 = _mm_load_si128((const __m128i *) (data + 1*8));
+   row2 = _mm_load_si128((const __m128i *) (data + 2*8));
+   row3 = _mm_load_si128((const __m128i *) (data + 3*8));
+   row4 = _mm_load_si128((const __m128i *) (data + 4*8));
+   row5 = _mm_load_si128((const __m128i *) (data + 5*8));
+   row6 = _mm_load_si128((const __m128i *) (data + 6*8));
+   row7 = _mm_load_si128((const __m128i *) (data + 7*8));
+
+   // column pass
+   dct_pass(bias_0, 10);
+
+   {
+      // 16bit 8x8 transpose pass 1
+      dct_interleave16(row0, row4);
+      dct_interleave16(row1, row5);
+      dct_interleave16(row2, row6);
+      dct_interleave16(row3, row7);
+
+      // transpose pass 2
+      dct_interleave16(row0, row2);
+      dct_interleave16(row1, row3);
+      dct_interleave16(row4, row6);
+      dct_interleave16(row5, row7);
+
+      // transpose pass 3
+      dct_interleave16(row0, row1);
+      dct_interleave16(row2, row3);
+      dct_interleave16(row4, row5);
+      dct_interleave16(row6, row7);
+   }
+
+   // row pass
+   dct_pass(bias_1, 17);
+
+   {
+      // pack
+      __m128i p0 = _mm_packus_epi16(row0, row1); // a0a1a2a3...a7b0b1b2b3...b7
+      __m128i p1 = _mm_packus_epi16(row2, row3);
+      __m128i p2 = _mm_packus_epi16(row4, row5);
+      __m128i p3 = _mm_packus_epi16(row6, row7);
+
+      // 8bit 8x8 transpose pass 1
+      dct_interleave8(p0, p2); // a0e0a1e1...
+      dct_interleave8(p1, p3); // c0g0c1g1...
+
+      // transpose pass 2
+      dct_interleave8(p0, p1); // a0c0e0g0...
+      dct_interleave8(p2, p3); // b0d0f0h0...
+
+      // transpose pass 3
+      dct_interleave8(p0, p2); // a0b0c0d0...
+      dct_interleave8(p1, p3); // a4b4c4d4...
+
+      // store
+      _mm_storel_epi64((__m128i *) out, p0); out += out_stride;
+      _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p0, 0x4e)); out += out_stride;
+      _mm_storel_epi64((__m128i *) out, p2); out += out_stride;
+      _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p2, 0x4e)); out += out_stride;
+      _mm_storel_epi64((__m128i *) out, p1); out += out_stride;
+      _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p1, 0x4e)); out += out_stride;
+      _mm_storel_epi64((__m128i *) out, p3); out += out_stride;
+      _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p3, 0x4e));
+   }
+
+#undef dct_const
+#undef dct_rot
+#undef dct_widen
+#undef dct_wadd
+#undef dct_wsub
+#undef dct_bfly32o
+#undef dct_interleave8
+#undef dct_interleave16
+#undef dct_pass
+}
+
+#endif // STBI_SSE2
+
+#ifdef STBI_NEON
+
+// NEON integer IDCT. should produce bit-identical
+// results to the generic C version.
+static void stbi__idct_simd(stbi_uc *out, int out_stride, short data[64])
+{
+   int16x8_t row0, row1, row2, row3, row4, row5, row6, row7;
+
+   int16x4_t rot0_0 = vdup_n_s16(stbi__f2f(0.5411961f));
+   int16x4_t rot0_1 = vdup_n_s16(stbi__f2f(-1.847759065f));
+   int16x4_t rot0_2 = vdup_n_s16(stbi__f2f( 0.765366865f));
+   int16x4_t rot1_0 = vdup_n_s16(stbi__f2f( 1.175875602f));
+   int16x4_t rot1_1 = vdup_n_s16(stbi__f2f(-0.899976223f));
+   int16x4_t rot1_2 = vdup_n_s16(stbi__f2f(-2.562915447f));
+   int16x4_t rot2_0 = vdup_n_s16(stbi__f2f(-1.961570560f));
+   int16x4_t rot2_1 = vdup_n_s16(stbi__f2f(-0.390180644f));
+   int16x4_t rot3_0 = vdup_n_s16(stbi__f2f( 0.298631336f));
+   int16x4_t rot3_1 = vdup_n_s16(stbi__f2f( 2.053119869f));
+   int16x4_t rot3_2 = vdup_n_s16(stbi__f2f( 3.072711026f));
+   int16x4_t rot3_3 = vdup_n_s16(stbi__f2f( 1.501321110f));
+
+#define dct_long_mul(out, inq, coeff) \
+   int32x4_t out##_l = vmull_s16(vget_low_s16(inq), coeff); \
+   int32x4_t out##_h = vmull_s16(vget_high_s16(inq), coeff)
+
+#define dct_long_mac(out, acc, inq, coeff) \
+   int32x4_t out##_l = vmlal_s16(acc##_l, vget_low_s16(inq), coeff); \
+   int32x4_t out##_h = vmlal_s16(acc##_h, vget_high_s16(inq), coeff)
+
+#define dct_widen(out, inq) \
+   int32x4_t out##_l = vshll_n_s16(vget_low_s16(inq), 12); \
+   int32x4_t out##_h = vshll_n_s16(vget_high_s16(inq), 12)
+
+// wide add
+#define dct_wadd(out, a, b) \
+   int32x4_t out##_l = vaddq_s32(a##_l, b##_l); \
+   int32x4_t out##_h = vaddq_s32(a##_h, b##_h)
+
+// wide sub
+#define dct_wsub(out, a, b) \
+   int32x4_t out##_l = vsubq_s32(a##_l, b##_l); \
+   int32x4_t out##_h = vsubq_s32(a##_h, b##_h)
+
+// butterfly a/b, then shift using "shiftop" by "s" and pack
+#define dct_bfly32o(out0,out1, a,b,shiftop,s) \
+   { \
+      dct_wadd(sum, a, b); \
+      dct_wsub(dif, a, b); \
+      out0 = vcombine_s16(shiftop(sum_l, s), shiftop(sum_h, s)); \
+      out1 = vcombine_s16(shiftop(dif_l, s), shiftop(dif_h, s)); \
+   }
+
+#define dct_pass(shiftop, shift) \
+   { \
+      /* even part */ \
+      int16x8_t sum26 = vaddq_s16(row2, row6); \
+      dct_long_mul(p1e, sum26, rot0_0); \
+      dct_long_mac(t2e, p1e, row6, rot0_1); \
+      dct_long_mac(t3e, p1e, row2, rot0_2); \
+      int16x8_t sum04 = vaddq_s16(row0, row4); \
+      int16x8_t dif04 = vsubq_s16(row0, row4); \
+      dct_widen(t0e, sum04); \
+      dct_widen(t1e, dif04); \
+      dct_wadd(x0, t0e, t3e); \
+      dct_wsub(x3, t0e, t3e); \
+      dct_wadd(x1, t1e, t2e); \
+      dct_wsub(x2, t1e, t2e); \
+      /* odd part */ \
+      int16x8_t sum15 = vaddq_s16(row1, row5); \
+      int16x8_t sum17 = vaddq_s16(row1, row7); \
+      int16x8_t sum35 = vaddq_s16(row3, row5); \
+      int16x8_t sum37 = vaddq_s16(row3, row7); \
+      int16x8_t sumodd = vaddq_s16(sum17, sum35); \
+      dct_long_mul(p5o, sumodd, rot1_0); \
+      dct_long_mac(p1o, p5o, sum17, rot1_1); \
+      dct_long_mac(p2o, p5o, sum35, rot1_2); \
+      dct_long_mul(p3o, sum37, rot2_0); \
+      dct_long_mul(p4o, sum15, rot2_1); \
+      dct_wadd(sump13o, p1o, p3o); \
+      dct_wadd(sump24o, p2o, p4o); \
+      dct_wadd(sump23o, p2o, p3o); \
+      dct_wadd(sump14o, p1o, p4o); \
+      dct_long_mac(x4, sump13o, row7, rot3_0); \
+      dct_long_mac(x5, sump24o, row5, rot3_1); \
+      dct_long_mac(x6, sump23o, row3, rot3_2); \
+      dct_long_mac(x7, sump14o, row1, rot3_3); \
+      dct_bfly32o(row0,row7, x0,x7,shiftop,shift); \
+      dct_bfly32o(row1,row6, x1,x6,shiftop,shift); \
+      dct_bfly32o(row2,row5, x2,x5,shiftop,shift); \
+      dct_bfly32o(row3,row4, x3,x4,shiftop,shift); \
+   }
+
+   // load
+   row0 = vld1q_s16(data + 0*8);
+   row1 = vld1q_s16(data + 1*8);
+   row2 = vld1q_s16(data + 2*8);
+   row3 = vld1q_s16(data + 3*8);
+   row4 = vld1q_s16(data + 4*8);
+   row5 = vld1q_s16(data + 5*8);
+   row6 = vld1q_s16(data + 6*8);
+   row7 = vld1q_s16(data + 7*8);
+
+   // add DC bias
+   row0 = vaddq_s16(row0, vsetq_lane_s16(1024, vdupq_n_s16(0), 0));
+
+   // column pass
+   dct_pass(vrshrn_n_s32, 10);
+
+   // 16bit 8x8 transpose
+   {
+// these three map to a single VTRN.16, VTRN.32, and VSWP, respectively.
+// whether compilers actually get this is another story, sadly.
+#define dct_trn16(x, y) { int16x8x2_t t = vtrnq_s16(x, y); x = t.val[0]; y = t.val[1]; }
+#define dct_trn32(x, y) { int32x4x2_t t = vtrnq_s32(vreinterpretq_s32_s16(x), vreinterpretq_s32_s16(y)); x = vreinterpretq_s16_s32(t.val[0]); y = vreinterpretq_s16_s32(t.val[1]); }
+#define dct_trn64(x, y) { int16x8_t x0 = x; int16x8_t y0 = y; x = vcombine_s16(vget_low_s16(x0), vget_low_s16(y0)); y = vcombine_s16(vget_high_s16(x0), vget_high_s16(y0)); }
+
+      // pass 1
+      dct_trn16(row0, row1); // a0b0a2b2a4b4a6b6
+      dct_trn16(row2, row3);
+      dct_trn16(row4, row5);
+      dct_trn16(row6, row7);
+
+      // pass 2
+      dct_trn32(row0, row2); // a0b0c0d0a4b4c4d4
+      dct_trn32(row1, row3);
+      dct_trn32(row4, row6);
+      dct_trn32(row5, row7);
+
+      // pass 3
+      dct_trn64(row0, row4); // a0b0c0d0e0f0g0h0
+      dct_trn64(row1, row5);
+      dct_trn64(row2, row6);
+      dct_trn64(row3, row7);
+
+#undef dct_trn16
+#undef dct_trn32
+#undef dct_trn64
+   }
+
+   // row pass
+   // vrshrn_n_s32 only supports shifts up to 16, we need
+   // 17. so do a non-rounding shift of 16 first then follow
+   // up with a rounding shift by 1.
+   dct_pass(vshrn_n_s32, 16);
+
+   {
+      // pack and round
+      uint8x8_t p0 = vqrshrun_n_s16(row0, 1);
+      uint8x8_t p1 = vqrshrun_n_s16(row1, 1);
+      uint8x8_t p2 = vqrshrun_n_s16(row2, 1);
+      uint8x8_t p3 = vqrshrun_n_s16(row3, 1);
+      uint8x8_t p4 = vqrshrun_n_s16(row4, 1);
+      uint8x8_t p5 = vqrshrun_n_s16(row5, 1);
+      uint8x8_t p6 = vqrshrun_n_s16(row6, 1);
+      uint8x8_t p7 = vqrshrun_n_s16(row7, 1);
+
+      // again, these can translate into one instruction, but often don't.
+#define dct_trn8_8(x, y) { uint8x8x2_t t = vtrn_u8(x, y); x = t.val[0]; y = t.val[1]; }
+#define dct_trn8_16(x, y) { uint16x4x2_t t = vtrn_u16(vreinterpret_u16_u8(x), vreinterpret_u16_u8(y)); x = vreinterpret_u8_u16(t.val[0]); y = vreinterpret_u8_u16(t.val[1]); }
+#define dct_trn8_32(x, y) { uint32x2x2_t t = vtrn_u32(vreinterpret_u32_u8(x), vreinterpret_u32_u8(y)); x = vreinterpret_u8_u32(t.val[0]); y = vreinterpret_u8_u32(t.val[1]); }
+
+      // sadly can't use interleaved stores here since we only write
+      // 8 bytes to each scan line!
+
+      // 8x8 8-bit transpose pass 1
+      dct_trn8_8(p0, p1);
+      dct_trn8_8(p2, p3);
+      dct_trn8_8(p4, p5);
+      dct_trn8_8(p6, p7);
+
+      // pass 2
+      dct_trn8_16(p0, p2);
+      dct_trn8_16(p1, p3);
+      dct_trn8_16(p4, p6);
+      dct_trn8_16(p5, p7);
+
+      // pass 3
+      dct_trn8_32(p0, p4);
+      dct_trn8_32(p1, p5);
+      dct_trn8_32(p2, p6);
+      dct_trn8_32(p3, p7);
+
+      // store
+      vst1_u8(out, p0); out += out_stride;
+      vst1_u8(out, p1); out += out_stride;
+      vst1_u8(out, p2); out += out_stride;
+      vst1_u8(out, p3); out += out_stride;
+      vst1_u8(out, p4); out += out_stride;
+      vst1_u8(out, p5); out += out_stride;
+      vst1_u8(out, p6); out += out_stride;
+      vst1_u8(out, p7);
+
+#undef dct_trn8_8
+#undef dct_trn8_16
+#undef dct_trn8_32
+   }
+
+#undef dct_long_mul
+#undef dct_long_mac
+#undef dct_widen
+#undef dct_wadd
+#undef dct_wsub
+#undef dct_bfly32o
+#undef dct_pass
+}
+
+#endif // STBI_NEON
+
+#define STBI__MARKER_none  0xff
+// if there's a pending marker from the entropy stream, return that
+// otherwise, fetch from the stream and get a marker. if there's no
+// marker, return 0xff, which is never a valid marker value
+static stbi_uc stbi__get_marker(stbi__jpeg *j)
+{
+   stbi_uc x;
+   if (j->marker != STBI__MARKER_none) { x = j->marker; j->marker = STBI__MARKER_none; return x; }
+   x = stbi__get8(j->s);
+   if (x != 0xff) return STBI__MARKER_none;
+   while (x == 0xff)
+      x = stbi__get8(j->s);
+   return x;
+}
+
+// in each scan, we'll have scan_n components, and the order
+// of the components is specified by order[]
+#define STBI__RESTART(x)     ((x) >= 0xd0 && (x) <= 0xd7)
+
+// after a restart interval, stbi__jpeg_reset the entropy decoder and
+// the dc prediction
+static void stbi__jpeg_reset(stbi__jpeg *j)
+{
+   j->code_bits = 0;
+   j->code_buffer = 0;
+   j->nomore = 0;
+   j->img_comp[0].dc_pred = j->img_comp[1].dc_pred = j->img_comp[2].dc_pred = 0;
+   j->marker = STBI__MARKER_none;
+   j->todo = j->restart_interval ? j->restart_interval : 0x7fffffff;
+   j->eob_run = 0;
+   // no more than 1<<31 MCUs if no restart_interal? that's plenty safe,
+   // since we don't even allow 1<<30 pixels
+}
+
+static int stbi__parse_entropy_coded_data(stbi__jpeg *z)
+{
+   stbi__jpeg_reset(z);
+   if (!z->progressive) {
+      if (z->scan_n == 1) {
+         int i,j;
+         STBI_SIMD_ALIGN(short, data[64]);
+         int n = z->order[0];
+         // non-interleaved data, we just need to process one block at a time,
+         // in trivial scanline order
+         // number of blocks to do just depends on how many actual "pixels" this
+         // component has, independent of interleaved MCU blocking and such
+         int w = (z->img_comp[n].x+7) >> 3;
+         int h = (z->img_comp[n].y+7) >> 3;
+         for (j=0; j < h; ++j) {
+            for (i=0; i < w; ++i) {
+               int ha = z->img_comp[n].ha;
+               if (!stbi__jpeg_decode_block(z, data, z->huff_dc+z->img_comp[n].hd, z->huff_ac+ha, z->fast_ac[ha], n, z->dequant[z->img_comp[n].tq])) return 0;
+               z->idct_block_kernel(z->img_comp[n].data+z->img_comp[n].w2*j*8+i*8, z->img_comp[n].w2, data);
+               // every data block is an MCU, so countdown the restart interval
+               if (--z->todo <= 0) {
+                  if (z->code_bits < 24) stbi__grow_buffer_unsafe(z);
+                  // if it's NOT a restart, then just bail, so we get corrupt data
+                  // rather than no data
+                  if (!STBI__RESTART(z->marker)) return 1;
+                  stbi__jpeg_reset(z);
+               }
+            }
+         }
+         return 1;
+      } else { // interleaved
+         int i,j,k,x,y;
+         STBI_SIMD_ALIGN(short, data[64]);
+         for (j=0; j < z->img_mcu_y; ++j) {
+            for (i=0; i < z->img_mcu_x; ++i) {
+               // scan an interleaved mcu... process scan_n components in order
+               for (k=0; k < z->scan_n; ++k) {
+                  int n = z->order[k];
+                  // scan out an mcu's worth of this component; that's just determined
+                  // by the basic H and V specified for the component
+                  for (y=0; y < z->img_comp[n].v; ++y) {
+                     for (x=0; x < z->img_comp[n].h; ++x) {
+                        int x2 = (i*z->img_comp[n].h + x)*8;
+                        int y2 = (j*z->img_comp[n].v + y)*8;
+                        int ha = z->img_comp[n].ha;
+                        if (!stbi__jpeg_decode_block(z, data, z->huff_dc+z->img_comp[n].hd, z->huff_ac+ha, z->fast_ac[ha], n, z->dequant[z->img_comp[n].tq])) return 0;
+                        z->idct_block_kernel(z->img_comp[n].data+z->img_comp[n].w2*y2+x2, z->img_comp[n].w2, data);
+                     }
+                  }
+               }
+               // after all interleaved components, that's an interleaved MCU,
+               // so now count down the restart interval
+               if (--z->todo <= 0) {
+                  if (z->code_bits < 24) stbi__grow_buffer_unsafe(z);
+                  if (!STBI__RESTART(z->marker)) return 1;
+                  stbi__jpeg_reset(z);
+               }
+            }
+         }
+         return 1;
+      }
+   } else {
+      if (z->scan_n == 1) {
+         int i,j;
+         int n = z->order[0];
+         // non-interleaved data, we just need to process one block at a time,
+         // in trivial scanline order
+         // number of blocks to do just depends on how many actual "pixels" this
+         // component has, independent of interleaved MCU blocking and such
+         int w = (z->img_comp[n].x+7) >> 3;
+         int h = (z->img_comp[n].y+7) >> 3;
+         for (j=0; j < h; ++j) {
+            for (i=0; i < w; ++i) {
+               short *data = z->img_comp[n].coeff + 64 * (i + j * z->img_comp[n].coeff_w);
+               if (z->spec_start == 0) {
+                  if (!stbi__jpeg_decode_block_prog_dc(z, data, &z->huff_dc[z->img_comp[n].hd], n))
+                     return 0;
+               } else {
+                  int ha = z->img_comp[n].ha;
+                  if (!stbi__jpeg_decode_block_prog_ac(z, data, &z->huff_ac[ha], z->fast_ac[ha]))
+                     return 0;
+               }
+               // every data block is an MCU, so countdown the restart interval
+               if (--z->todo <= 0) {
+                  if (z->code_bits < 24) stbi__grow_buffer_unsafe(z);
+                  if (!STBI__RESTART(z->marker)) return 1;
+                  stbi__jpeg_reset(z);
+               }
+            }
+         }
+         return 1;
+      } else { // interleaved
+         int i,j,k,x,y;
+         for (j=0; j < z->img_mcu_y; ++j) {
+            for (i=0; i < z->img_mcu_x; ++i) {
+               // scan an interleaved mcu... process scan_n components in order
+               for (k=0; k < z->scan_n; ++k) {
+                  int n = z->order[k];
+                  // scan out an mcu's worth of this component; that's just determined
+                  // by the basic H and V specified for the component
+                  for (y=0; y < z->img_comp[n].v; ++y) {
+                     for (x=0; x < z->img_comp[n].h; ++x) {
+                        int x2 = (i*z->img_comp[n].h + x);
+                        int y2 = (j*z->img_comp[n].v + y);
+                        short *data = z->img_comp[n].coeff + 64 * (x2 + y2 * z->img_comp[n].coeff_w);
+                        if (!stbi__jpeg_decode_block_prog_dc(z, data, &z->huff_dc[z->img_comp[n].hd], n))
+                           return 0;
+                     }
+                  }
+               }
+               // after all interleaved components, that's an interleaved MCU,
+               // so now count down the restart interval
+               if (--z->todo <= 0) {
+                  if (z->code_bits < 24) stbi__grow_buffer_unsafe(z);
+                  if (!STBI__RESTART(z->marker)) return 1;
+                  stbi__jpeg_reset(z);
+               }
+            }
+         }
+         return 1;
+      }
+   }
+}
+
+static void stbi__jpeg_dequantize(short *data, stbi_uc *dequant)
+{
+   int i;
+   for (i=0; i < 64; ++i)
+      data[i] *= dequant[i];
+}
+
+static void stbi__jpeg_finish(stbi__jpeg *z)
+{
+   if (z->progressive) {
+      // dequantize and idct the data
+      int i,j,n;
+      for (n=0; n < z->s->img_n; ++n) {
+         int w = (z->img_comp[n].x+7) >> 3;
+         int h = (z->img_comp[n].y+7) >> 3;
+         for (j=0; j < h; ++j) {
+            for (i=0; i < w; ++i) {
+               short *data = z->img_comp[n].coeff + 64 * (i + j * z->img_comp[n].coeff_w);
+               stbi__jpeg_dequantize(data, z->dequant[z->img_comp[n].tq]);
+               z->idct_block_kernel(z->img_comp[n].data+z->img_comp[n].w2*j*8+i*8, z->img_comp[n].w2, data);
+            }
+         }
+      }
+   }
+}
+
+static int stbi__process_marker(stbi__jpeg *z, int m)
+{
+   int L;
+   switch (m) {
+      case STBI__MARKER_none: // no marker found
+         return stbi__err("expected marker","Corrupt JPEG");
+
+      case 0xDD: // DRI - specify restart interval
+         if (stbi__get16be(z->s) != 4) return stbi__err("bad DRI len","Corrupt JPEG");
+         z->restart_interval = stbi__get16be(z->s);
+         return 1;
+
+      case 0xDB: // DQT - define quantization table
+         L = stbi__get16be(z->s)-2;
+         while (L > 0) {
+            int q = stbi__get8(z->s);
+            int p = q >> 4;
+            int t = q & 15,i;
+            if (p != 0) return stbi__err("bad DQT type","Corrupt JPEG");
+            if (t > 3) return stbi__err("bad DQT table","Corrupt JPEG");
+            for (i=0; i < 64; ++i)
+               z->dequant[t][stbi__jpeg_dezigzag[i]] = stbi__get8(z->s);
+            L -= 65;
+         }
+         return L==0;
+
+      case 0xC4: // DHT - define huffman table
+         L = stbi__get16be(z->s)-2;
+         while (L > 0) {
+            stbi_uc *v;
+            int sizes[16],i,n=0;
+            int q = stbi__get8(z->s);
+            int tc = q >> 4;
+            int th = q & 15;
+            if (tc > 1 || th > 3) return stbi__err("bad DHT header","Corrupt JPEG");
+            for (i=0; i < 16; ++i) {
+               sizes[i] = stbi__get8(z->s);
+               n += sizes[i];
+            }
+            L -= 17;
+            if (tc == 0) {
+               if (!stbi__build_huffman(z->huff_dc+th, sizes)) return 0;
+               v = z->huff_dc[th].values;
+            } else {
+               if (!stbi__build_huffman(z->huff_ac+th, sizes)) return 0;
+               v = z->huff_ac[th].values;
+            }
+            for (i=0; i < n; ++i)
+               v[i] = stbi__get8(z->s);
+            if (tc != 0)
+               stbi__build_fast_ac(z->fast_ac[th], z->huff_ac + th);
+            L -= n;
+         }
+         return L==0;
+   }
+   // check for comment block or APP blocks
+   if ((m >= 0xE0 && m <= 0xEF) || m == 0xFE) {
+      stbi__skip(z->s, stbi__get16be(z->s)-2);
+      return 1;
+   }
+   return 0;
+}
+
+// after we see SOS
+static int stbi__process_scan_header(stbi__jpeg *z)
+{
+   int i;
+   int Ls = stbi__get16be(z->s);
+   z->scan_n = stbi__get8(z->s);
+   if (z->scan_n < 1 || z->scan_n > 4 || z->scan_n > (int) z->s->img_n) return stbi__err("bad SOS component count","Corrupt JPEG");
+   if (Ls != 6+2*z->scan_n) return stbi__err("bad SOS len","Corrupt JPEG");
+   for (i=0; i < z->scan_n; ++i) {
+      int id = stbi__get8(z->s), which;
+      int q = stbi__get8(z->s);
+      for (which = 0; which < z->s->img_n; ++which)
+         if (z->img_comp[which].id == id)
+            break;
+      if (which == z->s->img_n) return 0; // no match
+      z->img_comp[which].hd = q >> 4;   if (z->img_comp[which].hd > 3) return stbi__err("bad DC huff","Corrupt JPEG");
+      z->img_comp[which].ha = q & 15;   if (z->img_comp[which].ha > 3) return stbi__err("bad AC huff","Corrupt JPEG");
+      z->order[i] = which;
+   }
+
+   {
+      int aa;
+      z->spec_start = stbi__get8(z->s);
+      z->spec_end   = stbi__get8(z->s); // should be 63, but might be 0
+      aa = stbi__get8(z->s);
+      z->succ_high = (aa >> 4);
+      z->succ_low  = (aa & 15);
+      if (z->progressive) {
+         if (z->spec_start > 63 || z->spec_end > 63  || z->spec_start > z->spec_end || z->succ_high > 13 || z->succ_low > 13)
+            return stbi__err("bad SOS", "Corrupt JPEG");
+      } else {
+         if (z->spec_start != 0) return stbi__err("bad SOS","Corrupt JPEG");
+         if (z->succ_high != 0 || z->succ_low != 0) return stbi__err("bad SOS","Corrupt JPEG");
+         z->spec_end = 63;
+      }
+   }
+
+   return 1;
+}
+
+static int stbi__free_jpeg_components(stbi__jpeg *z, int ncomp, int why)
+{
+   int i;
+   for (i=0; i < ncomp; ++i) {
+      if (z->img_comp[i].raw_data) {
+         STBI_FREE(z->img_comp[i].raw_data);
+         z->img_comp[i].raw_data = NULL;
+         z->img_comp[i].data = NULL;
+      }
+      if (z->img_comp[i].raw_coeff) {
+         STBI_FREE(z->img_comp[i].raw_coeff);
+         z->img_comp[i].raw_coeff = 0;
+         z->img_comp[i].coeff = 0;
+      }
+      if (z->img_comp[i].linebuf) {
+         STBI_FREE(z->img_comp[i].linebuf);
+         z->img_comp[i].linebuf = NULL;
+      }
+   }
+   return why;
+}
+
+static int stbi__process_frame_header(stbi__jpeg *z, int scan)
+{
+   stbi__context *s = z->s;
+   int Lf,p,i,q, h_max=1,v_max=1,c;
+   Lf = stbi__get16be(s);         if (Lf < 11) return stbi__err("bad SOF len","Corrupt JPEG"); // JPEG
+   p  = stbi__get8(s);            if (p != 8) return stbi__err("only 8-bit","JPEG format not supported: 8-bit only"); // JPEG baseline
+   s->img_y = stbi__get16be(s);   if (s->img_y == 0) return stbi__err("no header height", "JPEG format not supported: delayed height"); // Legal, but we don't handle it--but neither does IJG
+   s->img_x = stbi__get16be(s);   if (s->img_x == 0) return stbi__err("0 width","Corrupt JPEG"); // JPEG requires
+   c = stbi__get8(s);
+   if (c != 3 && c != 1) return stbi__err("bad component count","Corrupt JPEG");    // JFIF requires
+   s->img_n = c;
+   for (i=0; i < c; ++i) {
+      z->img_comp[i].data = NULL;
+      z->img_comp[i].linebuf = NULL;
+   }
+
+   if (Lf != 8+3*s->img_n) return stbi__err("bad SOF len","Corrupt JPEG");
+
+   z->rgb = 0;
+   for (i=0; i < s->img_n; ++i) {
+      static unsigned char rgb[3] = { 'R', 'G', 'B' };
+      z->img_comp[i].id = stbi__get8(s);
+      if (z->img_comp[i].id != i+1)   // JFIF requires
+         if (z->img_comp[i].id != i) {  // some version of jpegtran outputs non-JFIF-compliant files!
+            // somethings output this (see http://fileformats.archiveteam.org/wiki/JPEG#Color_format)
+            if (z->img_comp[i].id != rgb[i])
+               return stbi__err("bad component ID","Corrupt JPEG");
+            ++z->rgb;
+         }
+      q = stbi__get8(s);
+      z->img_comp[i].h = (q >> 4);  if (!z->img_comp[i].h || z->img_comp[i].h > 4) return stbi__err("bad H","Corrupt JPEG");
+      z->img_comp[i].v = q & 15;    if (!z->img_comp[i].v || z->img_comp[i].v > 4) return stbi__err("bad V","Corrupt JPEG");
+      z->img_comp[i].tq = stbi__get8(s);  if (z->img_comp[i].tq > 3) return stbi__err("bad TQ","Corrupt JPEG");
+   }
+
+   if (scan != STBI__SCAN_load) return 1;
+
+   if (!stbi__mad3sizes_valid(s->img_x, s->img_y, s->img_n, 0)) return stbi__err("too large", "Image too large to decode");
+
+   for (i=0; i < s->img_n; ++i) {
+      if (z->img_comp[i].h > h_max) h_max = z->img_comp[i].h;
+      if (z->img_comp[i].v > v_max) v_max = z->img_comp[i].v;
+   }
+
+   // compute interleaved mcu info
+   z->img_h_max = h_max;
+   z->img_v_max = v_max;
+   z->img_mcu_w = h_max * 8;
+   z->img_mcu_h = v_max * 8;
+   // these sizes can't be more than 17 bits
+   z->img_mcu_x = (s->img_x + z->img_mcu_w-1) / z->img_mcu_w;
+   z->img_mcu_y = (s->img_y + z->img_mcu_h-1) / z->img_mcu_h;
+
+   for (i=0; i < s->img_n; ++i) {
+      // number of effective pixels (e.g. for non-interleaved MCU)
+      z->img_comp[i].x = (s->img_x * z->img_comp[i].h + h_max-1) / h_max;
+      z->img_comp[i].y = (s->img_y * z->img_comp[i].v + v_max-1) / v_max;
+      // to simplify generation, we'll allocate enough memory to decode
+      // the bogus oversized data from using interleaved MCUs and their
+      // big blocks (e.g. a 16x16 iMCU on an image of width 33); we won't
+      // discard the extra data until colorspace conversion
+      //
+      // img_mcu_x, img_mcu_y: <=17 bits; comp[i].h and .v are <=4 (checked earlier)
+      // so these muls can't overflow with 32-bit ints (which we require)
+      z->img_comp[i].w2 = z->img_mcu_x * z->img_comp[i].h * 8;
+      z->img_comp[i].h2 = z->img_mcu_y * z->img_comp[i].v * 8;
+      z->img_comp[i].coeff = 0;
+      z->img_comp[i].raw_coeff = 0;
+      z->img_comp[i].linebuf = NULL;
+      z->img_comp[i].raw_data = stbi__malloc_mad2(z->img_comp[i].w2, z->img_comp[i].h2, 15);
+      if (z->img_comp[i].raw_data == NULL)
+         return stbi__free_jpeg_components(z, i+1, stbi__err("outofmem", "Out of memory"));
+      // align blocks for idct using mmx/sse
+      z->img_comp[i].data = (stbi_uc*) (((size_t) z->img_comp[i].raw_data + 15) & ~15);
+      if (z->progressive) {
+         // w2, h2 are multiples of 8 (see above)
+         z->img_comp[i].coeff_w = z->img_comp[i].w2 / 8;
+         z->img_comp[i].coeff_h = z->img_comp[i].h2 / 8;
+         z->img_comp[i].raw_coeff = stbi__malloc_mad3(z->img_comp[i].w2, z->img_comp[i].h2, sizeof(short), 15);
+         if (z->img_comp[i].raw_coeff == NULL)
+            return stbi__free_jpeg_components(z, i+1, stbi__err("outofmem", "Out of memory"));
+         z->img_comp[i].coeff = (short*) (((size_t) z->img_comp[i].raw_coeff + 15) & ~15);
+      }
+   }
+
+   return 1;
+}
+
+// use comparisons since in some cases we handle more than one case (e.g. SOF)
+#define stbi__DNL(x)         ((x) == 0xdc)
+#define stbi__SOI(x)         ((x) == 0xd8)
+#define stbi__EOI(x)         ((x) == 0xd9)
+#define stbi__SOF(x)         ((x) == 0xc0 || (x) == 0xc1 || (x) == 0xc2)
+#define stbi__SOS(x)         ((x) == 0xda)
+
+#define stbi__SOF_progressive(x)   ((x) == 0xc2)
+
+static int stbi__decode_jpeg_header(stbi__jpeg *z, int scan)
+{
+   int m;
+   z->marker = STBI__MARKER_none; // initialize cached marker to empty
+   m = stbi__get_marker(z);
+   if (!stbi__SOI(m)) return stbi__err("no SOI","Corrupt JPEG");
+   if (scan == STBI__SCAN_type) return 1;
+   m = stbi__get_marker(z);
+   while (!stbi__SOF(m)) {
+      if (!stbi__process_marker(z,m)) return 0;
+      m = stbi__get_marker(z);
+      while (m == STBI__MARKER_none) {
+         // some files have extra padding after their blocks, so ok, we'll scan
+         if (stbi__at_eof(z->s)) return stbi__err("no SOF", "Corrupt JPEG");
+         m = stbi__get_marker(z);
+      }
+   }
+   z->progressive = stbi__SOF_progressive(m);
+   if (!stbi__process_frame_header(z, scan)) return 0;
+   return 1;
+}
+
+// decode image to YCbCr format
+static int stbi__decode_jpeg_image(stbi__jpeg *j)
+{
+   int m;
+   for (m = 0; m < 4; m++) {
+      j->img_comp[m].raw_data = NULL;
+      j->img_comp[m].raw_coeff = NULL;
+   }
+   j->restart_interval = 0;
+   if (!stbi__decode_jpeg_header(j, STBI__SCAN_load)) return 0;
+   m = stbi__get_marker(j);
+   while (!stbi__EOI(m)) {
+      if (stbi__SOS(m)) {
+         if (!stbi__process_scan_header(j)) return 0;
+         if (!stbi__parse_entropy_coded_data(j)) return 0;
+         if (j->marker == STBI__MARKER_none ) {
+            // handle 0s at the end of image data from IP Kamera 9060
+            while (!stbi__at_eof(j->s)) {
+               int x = stbi__get8(j->s);
+               if (x == 255) {
+                  j->marker = stbi__get8(j->s);
+                  break;
+               } else if (x != 0) {
+                  return stbi__err("junk before marker", "Corrupt JPEG");
+               }
+            }
+            // if we reach eof without hitting a marker, stbi__get_marker() below will fail and we'll eventually return 0
+         }
+      } else {
+         if (!stbi__process_marker(j, m)) return 0;
+      }
+      m = stbi__get_marker(j);
+   }
+   if (j->progressive)
+      stbi__jpeg_finish(j);
+   return 1;
+}
+
+// static jfif-centered resampling (across block boundaries)
+
+typedef stbi_uc *(*resample_row_func)(stbi_uc *out, stbi_uc *in0, stbi_uc *in1,
+                                    int w, int hs);
+
+#define stbi__div4(x) ((stbi_uc) ((x) >> 2))
+
+static stbi_uc *resample_row_1(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs)
+{
+   STBI_NOTUSED(out);
+   STBI_NOTUSED(in_far);
+   STBI_NOTUSED(w);
+   STBI_NOTUSED(hs);
+   return in_near;
+}
+
+static stbi_uc* stbi__resample_row_v_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs)
+{
+   // need to generate two samples vertically for every one in input
+   int i;
+   STBI_NOTUSED(hs);
+   for (i=0; i < w; ++i)
+      out[i] = stbi__div4(3*in_near[i] + in_far[i] + 2);
+   return out;
+}
+
+static stbi_uc*  stbi__resample_row_h_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs)
+{
+   // need to generate two samples horizontally for every one in input
+   int i;
+   stbi_uc *input = in_near;
+
+   if (w == 1) {
+      // if only one sample, can't do any interpolation
+      out[0] = out[1] = input[0];
+      return out;
+   }
+
+   out[0] = input[0];
+   out[1] = stbi__div4(input[0]*3 + input[1] + 2);
+   for (i=1; i < w-1; ++i) {
+      int n = 3*input[i]+2;
+      out[i*2+0] = stbi__div4(n+input[i-1]);
+      out[i*2+1] = stbi__div4(n+input[i+1]);
+   }
+   out[i*2+0] = stbi__div4(input[w-2]*3 + input[w-1] + 2);
+   out[i*2+1] = input[w-1];
+
+   STBI_NOTUSED(in_far);
+   STBI_NOTUSED(hs);
+
+   return out;
+}
+
+#define stbi__div16(x) ((stbi_uc) ((x) >> 4))
+
+static stbi_uc *stbi__resample_row_hv_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs)
+{
+   // need to generate 2x2 samples for every one in input
+   int i,t0,t1;
+   if (w == 1) {
+      out[0] = out[1] = stbi__div4(3*in_near[0] + in_far[0] + 2);
+      return out;
+   }
+
+   t1 = 3*in_near[0] + in_far[0];
+   out[0] = stbi__div4(t1+2);
+   for (i=1; i < w; ++i) {
+      t0 = t1;
+      t1 = 3*in_near[i]+in_far[i];
+      out[i*2-1] = stbi__div16(3*t0 + t1 + 8);
+      out[i*2  ] = stbi__div16(3*t1 + t0 + 8);
+   }
+   out[w*2-1] = stbi__div4(t1+2);
+
+   STBI_NOTUSED(hs);
+
+   return out;
+}
+
+#if defined(STBI_SSE2) || defined(STBI_NEON)
+static stbi_uc *stbi__resample_row_hv_2_simd(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs)
+{
+   // need to generate 2x2 samples for every one in input
+   int i=0,t0,t1;
+
+   if (w == 1) {
+      out[0] = out[1] = stbi__div4(3*in_near[0] + in_far[0] + 2);
+      return out;
+   }
+
+   t1 = 3*in_near[0] + in_far[0];
+   // process groups of 8 pixels for as long as we can.
+   // note we can't handle the last pixel in a row in this loop
+   // because we need to handle the filter boundary conditions.
+   for (; i < ((w-1) & ~7); i += 8) {
+#if defined(STBI_SSE2)
+      // load and perform the vertical filtering pass
+      // this uses 3*x + y = 4*x + (y - x)
+      __m128i zero  = _mm_setzero_si128();
+      __m128i farb  = _mm_loadl_epi64((__m128i *) (in_far + i));
+      __m128i nearb = _mm_loadl_epi64((__m128i *) (in_near + i));
+      __m128i farw  = _mm_unpacklo_epi8(farb, zero);
+      __m128i nearw = _mm_unpacklo_epi8(nearb, zero);
+      __m128i diff  = _mm_sub_epi16(farw, nearw);
+      __m128i nears = _mm_slli_epi16(nearw, 2);
+      __m128i curr  = _mm_add_epi16(nears, diff); // current row
+
+      // horizontal filter works the same based on shifted vers of current
+      // row. "prev" is current row shifted right by 1 pixel; we need to
+      // insert the previous pixel value (from t1).
+      // "next" is current row shifted left by 1 pixel, with first pixel
+      // of next block of 8 pixels added in.
+      __m128i prv0 = _mm_slli_si128(curr, 2);
+      __m128i nxt0 = _mm_srli_si128(curr, 2);
+      __m128i prev = _mm_insert_epi16(prv0, t1, 0);
+      __m128i next = _mm_insert_epi16(nxt0, 3*in_near[i+8] + in_far[i+8], 7);
+
+      // horizontal filter, polyphase implementation since it's convenient:
+      // even pixels = 3*cur + prev = cur*4 + (prev - cur)
+      // odd  pixels = 3*cur + next = cur*4 + (next - cur)
+      // note the shared term.
+      __m128i bias  = _mm_set1_epi16(8);
+      __m128i curs = _mm_slli_epi16(curr, 2);
+      __m128i prvd = _mm_sub_epi16(prev, curr);
+      __m128i nxtd = _mm_sub_epi16(next, curr);
+      __m128i curb = _mm_add_epi16(curs, bias);
+      __m128i even = _mm_add_epi16(prvd, curb);
+      __m128i odd  = _mm_add_epi16(nxtd, curb);
+
+      // interleave even and odd pixels, then undo scaling.
+      __m128i int0 = _mm_unpacklo_epi16(even, odd);
+      __m128i int1 = _mm_unpackhi_epi16(even, odd);
+      __m128i de0  = _mm_srli_epi16(int0, 4);
+      __m128i de1  = _mm_srli_epi16(int1, 4);
+
+      // pack and write output
+      __m128i outv = _mm_packus_epi16(de0, de1);
+      _mm_storeu_si128((__m128i *) (out + i*2), outv);
+#elif defined(STBI_NEON)
+      // load and perform the vertical filtering pass
+      // this uses 3*x + y = 4*x + (y - x)
+      uint8x8_t farb  = vld1_u8(in_far + i);
+      uint8x8_t nearb = vld1_u8(in_near + i);
+      int16x8_t diff  = vreinterpretq_s16_u16(vsubl_u8(farb, nearb));
+      int16x8_t nears = vreinterpretq_s16_u16(vshll_n_u8(nearb, 2));
+      int16x8_t curr  = vaddq_s16(nears, diff); // current row
+
+      // horizontal filter works the same based on shifted vers of current
+      // row. "prev" is current row shifted right by 1 pixel; we need to
+      // insert the previous pixel value (from t1).
+      // "next" is current row shifted left by 1 pixel, with first pixel
+      // of next block of 8 pixels added in.
+      int16x8_t prv0 = vextq_s16(curr, curr, 7);
+      int16x8_t nxt0 = vextq_s16(curr, curr, 1);
+      int16x8_t prev = vsetq_lane_s16(t1, prv0, 0);
+      int16x8_t next = vsetq_lane_s16(3*in_near[i+8] + in_far[i+8], nxt0, 7);
+
+      // horizontal filter, polyphase implementation since it's convenient:
+      // even pixels = 3*cur + prev = cur*4 + (prev - cur)
+      // odd  pixels = 3*cur + next = cur*4 + (next - cur)
+      // note the shared term.
+      int16x8_t curs = vshlq_n_s16(curr, 2);
+      int16x8_t prvd = vsubq_s16(prev, curr);
+      int16x8_t nxtd = vsubq_s16(next, curr);
+      int16x8_t even = vaddq_s16(curs, prvd);
+      int16x8_t odd  = vaddq_s16(curs, nxtd);
+
+      // undo scaling and round, then store with even/odd phases interleaved
+      uint8x8x2_t o;
+      o.val[0] = vqrshrun_n_s16(even, 4);
+      o.val[1] = vqrshrun_n_s16(odd,  4);
+      vst2_u8(out + i*2, o);
+#endif
+
+      // "previous" value for next iter
+      t1 = 3*in_near[i+7] + in_far[i+7];
+   }
+
+   t0 = t1;
+   t1 = 3*in_near[i] + in_far[i];
+   out[i*2] = stbi__div16(3*t1 + t0 + 8);
+
+   for (++i; i < w; ++i) {
+      t0 = t1;
+      t1 = 3*in_near[i]+in_far[i];
+      out[i*2-1] = stbi__div16(3*t0 + t1 + 8);
+      out[i*2  ] = stbi__div16(3*t1 + t0 + 8);
+   }
+   out[w*2-1] = stbi__div4(t1+2);
+
+   STBI_NOTUSED(hs);
+
+   return out;
+}
+#endif
+
+static stbi_uc *stbi__resample_row_generic(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs)
+{
+   // resample with nearest-neighbor
+   int i,j;
+   STBI_NOTUSED(in_far);
+   for (i=0; i < w; ++i)
+      for (j=0; j < hs; ++j)
+         out[i*hs+j] = in_near[i];
+   return out;
+}
+
+#ifdef STBI_JPEG_OLD
+// this is the same YCbCr-to-RGB calculation that stb_image has used
+// historically before the algorithm changes in 1.49
+#define float2fixed(x)  ((int) ((x) * 65536 + 0.5))
+static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step)
+{
+   int i;
+   for (i=0; i < count; ++i) {
+      int y_fixed = (y[i] << 16) + 32768; // rounding
+      int r,g,b;
+      int cr = pcr[i] - 128;
+      int cb = pcb[i] - 128;
+      r = y_fixed + cr*float2fixed(1.40200f);
+      g = y_fixed - cr*float2fixed(0.71414f) - cb*float2fixed(0.34414f);
+      b = y_fixed                            + cb*float2fixed(1.77200f);
+      r >>= 16;
+      g >>= 16;
+      b >>= 16;
+      if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; }
+      if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; }
+      if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; }
+      out[0] = (stbi_uc)r;
+      out[1] = (stbi_uc)g;
+      out[2] = (stbi_uc)b;
+      out[3] = 255;
+      out += step;
+   }
+}
+#else
+// this is a reduced-precision calculation of YCbCr-to-RGB introduced
+// to make sure the code produces the same results in both SIMD and scalar
+#define float2fixed(x)  (((int) ((x) * 4096.0f + 0.5f)) << 8)
+static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step)
+{
+   int i;
+   for (i=0; i < count; ++i) {
+      int y_fixed = (y[i] << 20) + (1<<19); // rounding
+      int r,g,b;
+      int cr = pcr[i] - 128;
+      int cb = pcb[i] - 128;
+      r = y_fixed +  cr* float2fixed(1.40200f);
+      g = y_fixed + (cr*-float2fixed(0.71414f)) + ((cb*-float2fixed(0.34414f)) & 0xffff0000);
+      b = y_fixed                               +   cb* float2fixed(1.77200f);
+      r >>= 20;
+      g >>= 20;
+      b >>= 20;
+      if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; }
+      if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; }
+      if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; }
+      out[0] = (stbi_uc)r;
+      out[1] = (stbi_uc)g;
+      out[2] = (stbi_uc)b;
+      out[3] = 255;
+      out += step;
+   }
+}
+#endif
+
+#if defined(STBI_SSE2) || defined(STBI_NEON)
+static void stbi__YCbCr_to_RGB_simd(stbi_uc *out, stbi_uc const *y, stbi_uc const *pcb, stbi_uc const *pcr, int count, int step)
+{
+   int i = 0;
+
+#ifdef STBI_SSE2
+   // step == 3 is pretty ugly on the final interleave, and i'm not convinced
+   // it's useful in practice (you wouldn't use it for textures, for example).
+   // so just accelerate step == 4 case.
+   if (step == 4) {
+      // this is a fairly straightforward implementation and not super-optimized.
+      __m128i signflip  = _mm_set1_epi8(-0x80);
+      __m128i cr_const0 = _mm_set1_epi16(   (short) ( 1.40200f*4096.0f+0.5f));
+      __m128i cr_const1 = _mm_set1_epi16( - (short) ( 0.71414f*4096.0f+0.5f));
+      __m128i cb_const0 = _mm_set1_epi16( - (short) ( 0.34414f*4096.0f+0.5f));
+      __m128i cb_const1 = _mm_set1_epi16(   (short) ( 1.77200f*4096.0f+0.5f));
+      __m128i y_bias = _mm_set1_epi8((char) (unsigned char) 128);
+      __m128i xw = _mm_set1_epi16(255); // alpha channel
+
+      for (; i+7 < count; i += 8) {
+         // load
+         __m128i y_bytes = _mm_loadl_epi64((__m128i *) (y+i));
+         __m128i cr_bytes = _mm_loadl_epi64((__m128i *) (pcr+i));
+         __m128i cb_bytes = _mm_loadl_epi64((__m128i *) (pcb+i));
+         __m128i cr_biased = _mm_xor_si128(cr_bytes, signflip); // -128
+         __m128i cb_biased = _mm_xor_si128(cb_bytes, signflip); // -128
+
+         // unpack to short (and left-shift cr, cb by 8)
+         __m128i yw  = _mm_unpacklo_epi8(y_bias, y_bytes);
+         __m128i crw = _mm_unpacklo_epi8(_mm_setzero_si128(), cr_biased);
+         __m128i cbw = _mm_unpacklo_epi8(_mm_setzero_si128(), cb_biased);
+
+         // color transform
+         __m128i yws = _mm_srli_epi16(yw, 4);
+         __m128i cr0 = _mm_mulhi_epi16(cr_const0, crw);
+         __m128i cb0 = _mm_mulhi_epi16(cb_const0, cbw);
+         __m128i cb1 = _mm_mulhi_epi16(cbw, cb_const1);
+         __m128i cr1 = _mm_mulhi_epi16(crw, cr_const1);
+         __m128i rws = _mm_add_epi16(cr0, yws);
+         __m128i gwt = _mm_add_epi16(cb0, yws);
+         __m128i bws = _mm_add_epi16(yws, cb1);
+         __m128i gws = _mm_add_epi16(gwt, cr1);
+
+         // descale
+         __m128i rw = _mm_srai_epi16(rws, 4);
+         __m128i bw = _mm_srai_epi16(bws, 4);
+         __m128i gw = _mm_srai_epi16(gws, 4);
+
+         // back to byte, set up for transpose
+         __m128i brb = _mm_packus_epi16(rw, bw);
+         __m128i gxb = _mm_packus_epi16(gw, xw);
+
+         // transpose to interleave channels
+         __m128i t0 = _mm_unpacklo_epi8(brb, gxb);
+         __m128i t1 = _mm_unpackhi_epi8(brb, gxb);
+         __m128i o0 = _mm_unpacklo_epi16(t0, t1);
+         __m128i o1 = _mm_unpackhi_epi16(t0, t1);
+
+         // store
+         _mm_storeu_si128((__m128i *) (out + 0), o0);
+         _mm_storeu_si128((__m128i *) (out + 16), o1);
+         out += 32;
+      }
+   }
+#endif
+
+#ifdef STBI_NEON
+   // in this version, step=3 support would be easy to add. but is there demand?
+   if (step == 4) {
+      // this is a fairly straightforward implementation and not super-optimized.
+      uint8x8_t signflip = vdup_n_u8(0x80);
+      int16x8_t cr_const0 = vdupq_n_s16(   (short) ( 1.40200f*4096.0f+0.5f));
+      int16x8_t cr_const1 = vdupq_n_s16( - (short) ( 0.71414f*4096.0f+0.5f));
+      int16x8_t cb_const0 = vdupq_n_s16( - (short) ( 0.34414f*4096.0f+0.5f));
+      int16x8_t cb_const1 = vdupq_n_s16(   (short) ( 1.77200f*4096.0f+0.5f));
+
+      for (; i+7 < count; i += 8) {
+         // load
+         uint8x8_t y_bytes  = vld1_u8(y + i);
+         uint8x8_t cr_bytes = vld1_u8(pcr + i);
+         uint8x8_t cb_bytes = vld1_u8(pcb + i);
+         int8x8_t cr_biased = vreinterpret_s8_u8(vsub_u8(cr_bytes, signflip));
+         int8x8_t cb_biased = vreinterpret_s8_u8(vsub_u8(cb_bytes, signflip));
+
+         // expand to s16
+         int16x8_t yws = vreinterpretq_s16_u16(vshll_n_u8(y_bytes, 4));
+         int16x8_t crw = vshll_n_s8(cr_biased, 7);
+         int16x8_t cbw = vshll_n_s8(cb_biased, 7);
+
+         // color transform
+         int16x8_t cr0 = vqdmulhq_s16(crw, cr_const0);
+         int16x8_t cb0 = vqdmulhq_s16(cbw, cb_const0);
+         int16x8_t cr1 = vqdmulhq_s16(crw, cr_const1);
+         int16x8_t cb1 = vqdmulhq_s16(cbw, cb_const1);
+         int16x8_t rws = vaddq_s16(yws, cr0);
+         int16x8_t gws = vaddq_s16(vaddq_s16(yws, cb0), cr1);
+         int16x8_t bws = vaddq_s16(yws, cb1);
+
+         // undo scaling, round, convert to byte
+         uint8x8x4_t o;
+         o.val[0] = vqrshrun_n_s16(rws, 4);
+         o.val[1] = vqrshrun_n_s16(gws, 4);
+         o.val[2] = vqrshrun_n_s16(bws, 4);
+         o.val[3] = vdup_n_u8(255);
+
+         // store, interleaving r/g/b/a
+         vst4_u8(out, o);
+         out += 8*4;
+      }
+   }
+#endif
+
+   for (; i < count; ++i) {
+      int y_fixed = (y[i] << 20) + (1<<19); // rounding
+      int r,g,b;
+      int cr = pcr[i] - 128;
+      int cb = pcb[i] - 128;
+      r = y_fixed + cr* float2fixed(1.40200f);
+      g = y_fixed + cr*-float2fixed(0.71414f) + ((cb*-float2fixed(0.34414f)) & 0xffff0000);
+      b = y_fixed                             +   cb* float2fixed(1.77200f);
+      r >>= 20;
+      g >>= 20;
+      b >>= 20;
+      if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; }
+      if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; }
+      if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; }
+      out[0] = (stbi_uc)r;
+      out[1] = (stbi_uc)g;
+      out[2] = (stbi_uc)b;
+      out[3] = 255;
+      out += step;
+   }
+}
+#endif
+
+// set up the kernels
+static void stbi__setup_jpeg(stbi__jpeg *j)
+{
+   j->idct_block_kernel = stbi__idct_block;
+   j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_row;
+   j->resample_row_hv_2_kernel = stbi__resample_row_hv_2;
+
+#ifdef STBI_SSE2
+   if (stbi__sse2_available()) {
+      j->idct_block_kernel = stbi__idct_simd;
+      #ifndef STBI_JPEG_OLD
+      j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd;
+      #endif
+      j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd;
+   }
+#endif
+
+#ifdef STBI_NEON
+   j->idct_block_kernel = stbi__idct_simd;
+   #ifndef STBI_JPEG_OLD
+   j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd;
+   #endif
+   j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd;
+#endif
+}
+
+// clean up the temporary component buffers
+static void stbi__cleanup_jpeg(stbi__jpeg *j)
+{
+   stbi__free_jpeg_components(j, j->s->img_n, 0);
+}
+
+typedef struct
+{
+   resample_row_func resample;
+   stbi_uc *line0,*line1;
+   int hs,vs;   // expansion factor in each axis
+   int w_lores; // horizontal pixels pre-expansion
+   int ystep;   // how far through vertical expansion we are
+   int ypos;    // which pre-expansion row we're on
+} stbi__resample;
+
+static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp, int req_comp)
+{
+   int n, decode_n;
+   z->s->img_n = 0; // make stbi__cleanup_jpeg safe
+
+   // validate req_comp
+   if (req_comp < 0 || req_comp > 4) return stbi__errpuc("bad req_comp", "Internal error");
+
+   // load a jpeg image from whichever source, but leave in YCbCr format
+   if (!stbi__decode_jpeg_image(z)) { stbi__cleanup_jpeg(z); return NULL; }
+
+   // determine actual number of components to generate
+   n = req_comp ? req_comp : z->s->img_n;
+
+   if (z->s->img_n == 3 && n < 3)
+      decode_n = 1;
+   else
+      decode_n = z->s->img_n;
+
+   // resample and color-convert
+   {
+      int k;
+      unsigned int i,j;
+      stbi_uc *output;
+      stbi_uc *coutput[4];
+
+      stbi__resample res_comp[4];
+
+      for (k=0; k < decode_n; ++k) {
+         stbi__resample *r = &res_comp[k];
+
+         // allocate line buffer big enough for upsampling off the edges
+         // with upsample factor of 4
+         z->img_comp[k].linebuf = (stbi_uc *) stbi__malloc(z->s->img_x + 3);
+         if (!z->img_comp[k].linebuf) { stbi__cleanup_jpeg(z); return stbi__errpuc("outofmem", "Out of memory"); }
+
+         r->hs      = z->img_h_max / z->img_comp[k].h;
+         r->vs      = z->img_v_max / z->img_comp[k].v;
+         r->ystep   = r->vs >> 1;
+         r->w_lores = (z->s->img_x + r->hs-1) / r->hs;
+         r->ypos    = 0;
+         r->line0   = r->line1 = z->img_comp[k].data;
+
+         if      (r->hs == 1 && r->vs == 1) r->resample = resample_row_1;
+         else if (r->hs == 1 && r->vs == 2) r->resample = stbi__resample_row_v_2;
+         else if (r->hs == 2 && r->vs == 1) r->resample = stbi__resample_row_h_2;
+         else if (r->hs == 2 && r->vs == 2) r->resample = z->resample_row_hv_2_kernel;
+         else                               r->resample = stbi__resample_row_generic;
+      }
+
+      // can't error after this so, this is safe
+      output = (stbi_uc *) stbi__malloc_mad3(n, z->s->img_x, z->s->img_y, 1);
+      if (!output) { stbi__cleanup_jpeg(z); return stbi__errpuc("outofmem", "Out of memory"); }
+
+      // now go ahead and resample
+      for (j=0; j < z->s->img_y; ++j) {
+         stbi_uc *out = output + n * z->s->img_x * j;
+         for (k=0; k < decode_n; ++k) {
+            stbi__resample *r = &res_comp[k];
+            int y_bot = r->ystep >= (r->vs >> 1);
+            coutput[k] = r->resample(z->img_comp[k].linebuf,
+                                     y_bot ? r->line1 : r->line0,
+                                     y_bot ? r->line0 : r->line1,
+                                     r->w_lores, r->hs);
+            if (++r->ystep >= r->vs) {
+               r->ystep = 0;
+               r->line0 = r->line1;
+               if (++r->ypos < z->img_comp[k].y)
+                  r->line1 += z->img_comp[k].w2;
+            }
+         }
+         if (n >= 3) {
+            stbi_uc *y = coutput[0];
+            if (z->s->img_n == 3) {
+               if (z->rgb == 3) {
+                  for (i=0; i < z->s->img_x; ++i) {
+                     out[0] = y[i];
+                     out[1] = coutput[1][i];
+                     out[2] = coutput[2][i];
+                     out[3] = 255;
+                     out += n;
+                  }
+               } else {
+                  z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n);
+               }
+            } else
+               for (i=0; i < z->s->img_x; ++i) {
+                  out[0] = out[1] = out[2] = y[i];
+                  out[3] = 255; // not used if n==3
+                  out += n;
+               }
+         } else {
+            stbi_uc *y = coutput[0];
+            if (n == 1)
+               for (i=0; i < z->s->img_x; ++i) out[i] = y[i];
+            else
+               for (i=0; i < z->s->img_x; ++i) *out++ = y[i], *out++ = 255;
+         }
+      }
+      stbi__cleanup_jpeg(z);
+      *out_x = z->s->img_x;
+      *out_y = z->s->img_y;
+      if (comp) *comp  = z->s->img_n; // report original components, not output
+      return output;
+   }
+}
+
+static void *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
+{
+   unsigned char* result;
+   stbi__jpeg* j = (stbi__jpeg*) stbi__malloc(sizeof(stbi__jpeg));
+   j->s = s;
+   stbi__setup_jpeg(j);
+   result = load_jpeg_image(j, x,y,comp,req_comp);
+   STBI_FREE(j);
+   return result;
+}
+
+static int stbi__jpeg_test(stbi__context *s)
+{
+   int r;
+   stbi__jpeg j;
+   j.s = s;
+   stbi__setup_jpeg(&j);
+   r = stbi__decode_jpeg_header(&j, STBI__SCAN_type);
+   stbi__rewind(s);
+   return r;
+}
+
+static int stbi__jpeg_info_raw(stbi__jpeg *j, int *x, int *y, int *comp)
+{
+   if (!stbi__decode_jpeg_header(j, STBI__SCAN_header)) {
+      stbi__rewind( j->s );
+      return 0;
+   }
+   if (x) *x = j->s->img_x;
+   if (y) *y = j->s->img_y;
+   if (comp) *comp = j->s->img_n;
+   return 1;
+}
+
+static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp)
+{
+   int result;
+   stbi__jpeg* j = (stbi__jpeg*) (stbi__malloc(sizeof(stbi__jpeg)));
+   j->s = s;
+   result = stbi__jpeg_info_raw(j, x, y, comp);
+   STBI_FREE(j);
+   return result;
+}
+#endif
+
+// public domain zlib decode    v0.2  Sean Barrett 2006-11-18
+//    simple implementation
+//      - all input must be provided in an upfront buffer
+//      - all output is written to a single output buffer (can malloc/realloc)
+//    performance
+//      - fast huffman
+
+#ifndef STBI_NO_ZLIB
+
+// fast-way is faster to check than jpeg huffman, but slow way is slower
+#define STBI__ZFAST_BITS  9 // accelerate all cases in default tables
+#define STBI__ZFAST_MASK  ((1 << STBI__ZFAST_BITS) - 1)
+
+// zlib-style huffman encoding
+// (jpegs packs from left, zlib from right, so can't share code)
+typedef struct
+{
+   stbi__uint16 fast[1 << STBI__ZFAST_BITS];
+   stbi__uint16 firstcode[16];
+   int maxcode[17];
+   stbi__uint16 firstsymbol[16];
+   stbi_uc  size[288];
+   stbi__uint16 value[288];
+} stbi__zhuffman;
+
+stbi_inline static int stbi__bitreverse16(int n)
+{
+  n = ((n & 0xAAAA) >>  1) | ((n & 0x5555) << 1);
+  n = ((n & 0xCCCC) >>  2) | ((n & 0x3333) << 2);
+  n = ((n & 0xF0F0) >>  4) | ((n & 0x0F0F) << 4);
+  n = ((n & 0xFF00) >>  8) | ((n & 0x00FF) << 8);
+  return n;
+}
+
+stbi_inline static int stbi__bit_reverse(int v, int bits)
+{
+   STBI_ASSERT(bits <= 16);
+   // to bit reverse n bits, reverse 16 and shift
+   // e.g. 11 bits, bit reverse and shift away 5
+   return stbi__bitreverse16(v) >> (16-bits);
+}
+
+static int stbi__zbuild_huffman(stbi__zhuffman *z, stbi_uc *sizelist, int num)
+{
+   int i,k=0;
+   int code, next_code[16], sizes[17];
+
+   // DEFLATE spec for generating codes
+   memset(sizes, 0, sizeof(sizes));
+   memset(z->fast, 0, sizeof(z->fast));
+   for (i=0; i < num; ++i)
+      ++sizes[sizelist[i]];
+   sizes[0] = 0;
+   for (i=1; i < 16; ++i)
+      if (sizes[i] > (1 << i))
+         return stbi__err("bad sizes", "Corrupt PNG");
+   code = 0;
+   for (i=1; i < 16; ++i) {
+      next_code[i] = code;
+      z->firstcode[i] = (stbi__uint16) code;
+      z->firstsymbol[i] = (stbi__uint16) k;
+      code = (code + sizes[i]);
+      if (sizes[i])
+         if (code-1 >= (1 << i)) return stbi__err("bad codelengths","Corrupt PNG");
+      z->maxcode[i] = code << (16-i); // preshift for inner loop
+      code <<= 1;
+      k += sizes[i];
+   }
+   z->maxcode[16] = 0x10000; // sentinel
+   for (i=0; i < num; ++i) {
+      int s = sizelist[i];
+      if (s) {
+         int c = next_code[s] - z->firstcode[s] + z->firstsymbol[s];
+         stbi__uint16 fastv = (stbi__uint16) ((s << 9) | i);
+         z->size [c] = (stbi_uc     ) s;
+         z->value[c] = (stbi__uint16) i;
+         if (s <= STBI__ZFAST_BITS) {
+            int j = stbi__bit_reverse(next_code[s],s);
+            while (j < (1 << STBI__ZFAST_BITS)) {
+               z->fast[j] = fastv;
+               j += (1 << s);
+            }
+         }
+         ++next_code[s];
+      }
+   }
+   return 1;
+}
+
+// zlib-from-memory implementation for PNG reading
+//    because PNG allows splitting the zlib stream arbitrarily,
+//    and it's annoying structurally to have PNG call ZLIB call PNG,
+//    we require PNG read all the IDATs and combine them into a single
+//    memory buffer
+
+typedef struct
+{
+   stbi_uc *zbuffer, *zbuffer_end;
+   int num_bits;
+   stbi__uint32 code_buffer;
+
+   char *zout;
+   char *zout_start;
+   char *zout_end;
+   int   z_expandable;
+
+   stbi__zhuffman z_length, z_distance;
+} stbi__zbuf;
+
+stbi_inline static stbi_uc stbi__zget8(stbi__zbuf *z)
+{
+   if (z->zbuffer >= z->zbuffer_end) return 0;
+   return *z->zbuffer++;
+}
+
+static void stbi__fill_bits(stbi__zbuf *z)
+{
+   do {
+      STBI_ASSERT(z->code_buffer < (1U << z->num_bits));
+      z->code_buffer |= (unsigned int) stbi__zget8(z) << z->num_bits;
+      z->num_bits += 8;
+   } while (z->num_bits <= 24);
+}
+
+stbi_inline static unsigned int stbi__zreceive(stbi__zbuf *z, int n)
+{
+   unsigned int k;
+   if (z->num_bits < n) stbi__fill_bits(z);
+   k = z->code_buffer & ((1 << n) - 1);
+   z->code_buffer >>= n;
+   z->num_bits -= n;
+   return k;
+}
+
+static int stbi__zhuffman_decode_slowpath(stbi__zbuf *a, stbi__zhuffman *z)
+{
+   int b,s,k;
+   // not resolved by fast table, so compute it the slow way
+   // use jpeg approach, which requires MSbits at top
+   k = stbi__bit_reverse(a->code_buffer, 16);
+   for (s=STBI__ZFAST_BITS+1; ; ++s)
+      if (k < z->maxcode[s])
+         break;
+   if (s == 16) return -1; // invalid code!
+   // code size is s, so:
+   b = (k >> (16-s)) - z->firstcode[s] + z->firstsymbol[s];
+   STBI_ASSERT(z->size[b] == s);
+   a->code_buffer >>= s;
+   a->num_bits -= s;
+   return z->value[b];
+}
+
+stbi_inline static int stbi__zhuffman_decode(stbi__zbuf *a, stbi__zhuffman *z)
+{
+   int b,s;
+   if (a->num_bits < 16) stbi__fill_bits(a);
+   b = z->fast[a->code_buffer & STBI__ZFAST_MASK];
+   if (b) {
+      s = b >> 9;
+      a->code_buffer >>= s;
+      a->num_bits -= s;
+      return b & 511;
+   }
+   return stbi__zhuffman_decode_slowpath(a, z);
+}
+
+static int stbi__zexpand(stbi__zbuf *z, char *zout, int n)  // need to make room for n bytes
+{
+   char *q;
+   int cur, limit, old_limit;
+   z->zout = zout;
+   if (!z->z_expandable) return stbi__err("output buffer limit","Corrupt PNG");
+   cur   = (int) (z->zout     - z->zout_start);
+   limit = old_limit = (int) (z->zout_end - z->zout_start);
+   while (cur + n > limit)
+      limit *= 2;
+   q = (char *) STBI_REALLOC_SIZED(z->zout_start, old_limit, limit);
+   STBI_NOTUSED(old_limit);
+   if (q == NULL) return stbi__err("outofmem", "Out of memory");
+   z->zout_start = q;
+   z->zout       = q + cur;
+   z->zout_end   = q + limit;
+   return 1;
+}
+
+static int stbi__zlength_base[31] = {
+   3,4,5,6,7,8,9,10,11,13,
+   15,17,19,23,27,31,35,43,51,59,
+   67,83,99,115,131,163,195,227,258,0,0 };
+
+static int stbi__zlength_extra[31]=
+{ 0,0,0,0,0,0,0,0,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,5,5,5,5,0,0,0 };
+
+static int stbi__zdist_base[32] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,
+257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577,0,0};
+
+static int stbi__zdist_extra[32] =
+{ 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13};
+
+static int stbi__parse_huffman_block(stbi__zbuf *a)
+{
+   char *zout = a->zout;
+   for(;;) {
+      int z = stbi__zhuffman_decode(a, &a->z_length);
+      if (z < 256) {
+         if (z < 0) return stbi__err("bad huffman code","Corrupt PNG"); // error in huffman codes
+         if (zout >= a->zout_end) {
+            if (!stbi__zexpand(a, zout, 1)) return 0;
+            zout = a->zout;
+         }
+         *zout++ = (char) z;
+      } else {
+         stbi_uc *p;
+         int len,dist;
+         if (z == 256) {
+            a->zout = zout;
+            return 1;
+         }
+         z -= 257;
+         len = stbi__zlength_base[z];
+         if (stbi__zlength_extra[z]) len += stbi__zreceive(a, stbi__zlength_extra[z]);
+         z = stbi__zhuffman_decode(a, &a->z_distance);
+         if (z < 0) return stbi__err("bad huffman code","Corrupt PNG");
+         dist = stbi__zdist_base[z];
+         if (stbi__zdist_extra[z]) dist += stbi__zreceive(a, stbi__zdist_extra[z]);
+         if (zout - a->zout_start < dist) return stbi__err("bad dist","Corrupt PNG");
+         if (zout + len > a->zout_end) {
+            if (!stbi__zexpand(a, zout, len)) return 0;
+            zout = a->zout;
+         }
+         p = (stbi_uc *) (zout - dist);
+         if (dist == 1) { // run of one byte; common in images.
+            stbi_uc v = *p;
+            if (len) { do *zout++ = v; while (--len); }
+         } else {
+            if (len) { do *zout++ = *p++; while (--len); }
+         }
+      }
+   }
+}
+
+static int stbi__compute_huffman_codes(stbi__zbuf *a)
+{
+   static stbi_uc length_dezigzag[19] = { 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 };
+   stbi__zhuffman z_codelength;
+   stbi_uc lencodes[286+32+137];//padding for maximum single op
+   stbi_uc codelength_sizes[19];
+   int i,n;
+
+   int hlit  = stbi__zreceive(a,5) + 257;
+   int hdist = stbi__zreceive(a,5) + 1;
+   int hclen = stbi__zreceive(a,4) + 4;
+   int ntot  = hlit + hdist;
+
+   memset(codelength_sizes, 0, sizeof(codelength_sizes));
+   for (i=0; i < hclen; ++i) {
+      int s = stbi__zreceive(a,3);
+      codelength_sizes[length_dezigzag[i]] = (stbi_uc) s;
+   }
+   if (!stbi__zbuild_huffman(&z_codelength, codelength_sizes, 19)) return 0;
+
+   n = 0;
+   while (n < ntot) {
+      int c = stbi__zhuffman_decode(a, &z_codelength);
+      if (c < 0 || c >= 19) return stbi__err("bad codelengths", "Corrupt PNG");
+      if (c < 16)
+         lencodes[n++] = (stbi_uc) c;
+      else {
+         stbi_uc fill = 0;
+         if (c == 16) {
+            c = stbi__zreceive(a,2)+3;
+            if (n == 0) return stbi__err("bad codelengths", "Corrupt PNG");
+            fill = lencodes[n-1];
+         } else if (c == 17)
+            c = stbi__zreceive(a,3)+3;
+         else {
+            STBI_ASSERT(c == 18);
+            c = stbi__zreceive(a,7)+11;
+         }
+         if (ntot - n < c) return stbi__err("bad codelengths", "Corrupt PNG");
+         memset(lencodes+n, fill, c);
+         n += c;
+      }
+   }
+   if (n != ntot) return stbi__err("bad codelengths","Corrupt PNG");
+   if (!stbi__zbuild_huffman(&a->z_length, lencodes, hlit)) return 0;
+   if (!stbi__zbuild_huffman(&a->z_distance, lencodes+hlit, hdist)) return 0;
+   return 1;
+}
+
+static int stbi__parse_uncompressed_block(stbi__zbuf *a)
+{
+   stbi_uc header[4];
+   int len,nlen,k;
+   if (a->num_bits & 7)
+      stbi__zreceive(a, a->num_bits & 7); // discard
+   // drain the bit-packed data into header
+   k = 0;
+   while (a->num_bits > 0) {
+      header[k++] = (stbi_uc) (a->code_buffer & 255); // suppress MSVC run-time check
+      a->code_buffer >>= 8;
+      a->num_bits -= 8;
+   }
+   STBI_ASSERT(a->num_bits == 0);
+   // now fill header the normal way
+   while (k < 4)
+      header[k++] = stbi__zget8(a);
+   len  = header[1] * 256 + header[0];
+   nlen = header[3] * 256 + header[2];
+   if (nlen != (len ^ 0xffff)) return stbi__err("zlib corrupt","Corrupt PNG");
+   if (a->zbuffer + len > a->zbuffer_end) return stbi__err("read past buffer","Corrupt PNG");
+   if (a->zout + len > a->zout_end)
+      if (!stbi__zexpand(a, a->zout, len)) return 0;
+   memcpy(a->zout, a->zbuffer, len);
+   a->zbuffer += len;
+   a->zout += len;
+   return 1;
+}
+
+static int stbi__parse_zlib_header(stbi__zbuf *a)
+{
+   int cmf   = stbi__zget8(a);
+   int cm    = cmf & 15;
+   /* int cinfo = cmf >> 4; */
+   int flg   = stbi__zget8(a);
+   if ((cmf*256+flg) % 31 != 0) return stbi__err("bad zlib header","Corrupt PNG"); // zlib spec
+   if (flg & 32) return stbi__err("no preset dict","Corrupt PNG"); // preset dictionary not allowed in png
+   if (cm != 8) return stbi__err("bad compression","Corrupt PNG"); // DEFLATE required for png
+   // window = 1 << (8 + cinfo)... but who cares, we fully buffer output
+   return 1;
+}
+
+// @TODO: should statically initialize these for optimal thread safety
+static stbi_uc stbi__zdefault_length[288], stbi__zdefault_distance[32];
+static void stbi__init_zdefaults(void)
+{
+   int i;   // use <= to match clearly with spec
+   for (i=0; i <= 143; ++i)     stbi__zdefault_length[i]   = 8;
+   for (   ; i <= 255; ++i)     stbi__zdefault_length[i]   = 9;
+   for (   ; i <= 279; ++i)     stbi__zdefault_length[i]   = 7;
+   for (   ; i <= 287; ++i)     stbi__zdefault_length[i]   = 8;
+
+   for (i=0; i <=  31; ++i)     stbi__zdefault_distance[i] = 5;
+}
+
+static int stbi__parse_zlib(stbi__zbuf *a, int parse_header)
+{
+   int final, type;
+   if (parse_header)
+      if (!stbi__parse_zlib_header(a)) return 0;
+   a->num_bits = 0;
+   a->code_buffer = 0;
+   do {
+      final = stbi__zreceive(a,1);
+      type = stbi__zreceive(a,2);
+      if (type == 0) {
+         if (!stbi__parse_uncompressed_block(a)) return 0;
+      } else if (type == 3) {
+         return 0;
+      } else {
+         if (type == 1) {
+            // use fixed code lengths
+            if (!stbi__zdefault_distance[31]) stbi__init_zdefaults();
+            if (!stbi__zbuild_huffman(&a->z_length  , stbi__zdefault_length  , 288)) return 0;
+            if (!stbi__zbuild_huffman(&a->z_distance, stbi__zdefault_distance,  32)) return 0;
+         } else {
+            if (!stbi__compute_huffman_codes(a)) return 0;
+         }
+         if (!stbi__parse_huffman_block(a)) return 0;
+      }
+   } while (!final);
+   return 1;
+}
+
+static int stbi__do_zlib(stbi__zbuf *a, char *obuf, int olen, int exp, int parse_header)
+{
+   a->zout_start = obuf;
+   a->zout       = obuf;
+   a->zout_end   = obuf + olen;
+   a->z_expandable = exp;
+
+   return stbi__parse_zlib(a, parse_header);
+}
+
+STBIDEF char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen)
+{
+   stbi__zbuf a;
+   char *p = (char *) stbi__malloc(initial_size);
+   if (p == NULL) return NULL;
+   a.zbuffer = (stbi_uc *) buffer;
+   a.zbuffer_end = (stbi_uc *) buffer + len;
+   if (stbi__do_zlib(&a, p, initial_size, 1, 1)) {
+      if (outlen) *outlen = (int) (a.zout - a.zout_start);
+      return a.zout_start;
+   } else {
+      STBI_FREE(a.zout_start);
+      return NULL;
+   }
+}
+
+STBIDEF char *stbi_zlib_decode_malloc(char const *buffer, int len, int *outlen)
+{
+   return stbi_zlib_decode_malloc_guesssize(buffer, len, 16384, outlen);
+}
+
+STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, int len, int initial_size, int *outlen, int parse_header)
+{
+   stbi__zbuf a;
+   char *p = (char *) stbi__malloc(initial_size);
+   if (p == NULL) return NULL;
+   a.zbuffer = (stbi_uc *) buffer;
+   a.zbuffer_end = (stbi_uc *) buffer + len;
+   if (stbi__do_zlib(&a, p, initial_size, 1, parse_header)) {
+      if (outlen) *outlen = (int) (a.zout - a.zout_start);
+      return a.zout_start;
+   } else {
+      STBI_FREE(a.zout_start);
+      return NULL;
+   }
+}
+
+STBIDEF int stbi_zlib_decode_buffer(char *obuffer, int olen, char const *ibuffer, int ilen)
+{
+   stbi__zbuf a;
+   a.zbuffer = (stbi_uc *) ibuffer;
+   a.zbuffer_end = (stbi_uc *) ibuffer + ilen;
+   if (stbi__do_zlib(&a, obuffer, olen, 0, 1))
+      return (int) (a.zout - a.zout_start);
+   else
+      return -1;
+}
+
+STBIDEF char *stbi_zlib_decode_noheader_malloc(char const *buffer, int len, int *outlen)
+{
+   stbi__zbuf a;
+   char *p = (char *) stbi__malloc(16384);
+   if (p == NULL) return NULL;
+   a.zbuffer = (stbi_uc *) buffer;
+   a.zbuffer_end = (stbi_uc *) buffer+len;
+   if (stbi__do_zlib(&a, p, 16384, 1, 0)) {
+      if (outlen) *outlen = (int) (a.zout - a.zout_start);
+      return a.zout_start;
+   } else {
+      STBI_FREE(a.zout_start);
+      return NULL;
+   }
+}
+
+STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char *ibuffer, int ilen)
+{
+   stbi__zbuf a;
+   a.zbuffer = (stbi_uc *) ibuffer;
+   a.zbuffer_end = (stbi_uc *) ibuffer + ilen;
+   if (stbi__do_zlib(&a, obuffer, olen, 0, 0))
+      return (int) (a.zout - a.zout_start);
+   else
+      return -1;
+}
+#endif
+
+// public domain "baseline" PNG decoder   v0.10  Sean Barrett 2006-11-18
+//    simple implementation
+//      - only 8-bit samples
+//      - no CRC checking
+//      - allocates lots of intermediate memory
+//        - avoids problem of streaming data between subsystems
+//        - avoids explicit window management
+//    performance
+//      - uses stb_zlib, a PD zlib implementation with fast huffman decoding
+
+#ifndef STBI_NO_PNG
+typedef struct
+{
+   stbi__uint32 length;
+   stbi__uint32 type;
+} stbi__pngchunk;
+
+static stbi__pngchunk stbi__get_chunk_header(stbi__context *s)
+{
+   stbi__pngchunk c;
+   c.length = stbi__get32be(s);
+   c.type   = stbi__get32be(s);
+   return c;
+}
+
+static int stbi__check_png_header(stbi__context *s)
+{
+   static stbi_uc png_sig[8] = { 137,80,78,71,13,10,26,10 };
+   int i;
+   for (i=0; i < 8; ++i)
+      if (stbi__get8(s) != png_sig[i]) return stbi__err("bad png sig","Not a PNG");
+   return 1;
+}
+
+typedef struct
+{
+   stbi__context *s;
+   stbi_uc *idata, *expanded, *out;
+   int depth;
+} stbi__png;
+
+
+enum {
+   STBI__F_none=0,
+   STBI__F_sub=1,
+   STBI__F_up=2,
+   STBI__F_avg=3,
+   STBI__F_paeth=4,
+   // synthetic filters used for first scanline to avoid needing a dummy row of 0s
+   STBI__F_avg_first,
+   STBI__F_paeth_first
+};
+
+static stbi_uc first_row_filter[5] =
+{
+   STBI__F_none,
+   STBI__F_sub,
+   STBI__F_none,
+   STBI__F_avg_first,
+   STBI__F_paeth_first
+};
+
+static int stbi__paeth(int a, int b, int c)
+{
+   int p = a + b - c;
+   int pa = abs(p-a);
+   int pb = abs(p-b);
+   int pc = abs(p-c);
+   if (pa <= pb && pa <= pc) return a;
+   if (pb <= pc) return b;
+   return c;
+}
+
+static stbi_uc stbi__depth_scale_table[9] = { 0, 0xff, 0x55, 0, 0x11, 0,0,0, 0x01 };
+
+// create the png data from post-deflated data
+static int stbi__create_png_image_raw(stbi__png *a, stbi_uc *raw, stbi__uint32 raw_len, int out_n, stbi__uint32 x, stbi__uint32 y, int depth, int color)
+{
+   int bytes = (depth == 16? 2 : 1);
+   stbi__context *s = a->s;
+   stbi__uint32 i,j,stride = x*out_n*bytes;
+   stbi__uint32 img_len, img_width_bytes;
+   int k;
+   int img_n = s->img_n; // copy it into a local for later
+
+   int output_bytes = out_n*bytes;
+   int filter_bytes = img_n*bytes;
+   int width = x;
+
+   STBI_ASSERT(out_n == s->img_n || out_n == s->img_n+1);
+   a->out = (stbi_uc *) stbi__malloc_mad3(x, y, output_bytes, 0); // extra bytes to write off the end into
+   if (!a->out) return stbi__err("outofmem", "Out of memory");
+
+   img_width_bytes = (((img_n * x * depth) + 7) >> 3);
+   img_len = (img_width_bytes + 1) * y;
+   if (s->img_x == x && s->img_y == y) {
+      if (raw_len != img_len) return stbi__err("not enough pixels","Corrupt PNG");
+   } else { // interlaced:
+      if (raw_len < img_len) return stbi__err("not enough pixels","Corrupt PNG");
+   }
+
+   for (j=0; j < y; ++j) {
+      stbi_uc *cur = a->out + stride*j;
+      stbi_uc *prior = cur - stride;
+      int filter = *raw++;
+
+      if (filter > 4)
+         return stbi__err("invalid filter","Corrupt PNG");
+
+      if (depth < 8) {
+         STBI_ASSERT(img_width_bytes <= x);
+         cur += x*out_n - img_width_bytes; // store output to the rightmost img_len bytes, so we can decode in place
+         filter_bytes = 1;
+         width = img_width_bytes;
+      }
+
+      // if first row, use special filter that doesn't sample previous row
+      if (j == 0) filter = first_row_filter[filter];
+
+      // handle first byte explicitly
+      for (k=0; k < filter_bytes; ++k) {
+         switch (filter) {
+            case STBI__F_none       : cur[k] = raw[k]; break;
+            case STBI__F_sub        : cur[k] = raw[k]; break;
+            case STBI__F_up         : cur[k] = STBI__BYTECAST(raw[k] + prior[k]); break;
+            case STBI__F_avg        : cur[k] = STBI__BYTECAST(raw[k] + (prior[k]>>1)); break;
+            case STBI__F_paeth      : cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(0,prior[k],0)); break;
+            case STBI__F_avg_first  : cur[k] = raw[k]; break;
+            case STBI__F_paeth_first: cur[k] = raw[k]; break;
+         }
+      }
+
+      if (depth == 8) {
+         if (img_n != out_n)
+            cur[img_n] = 255; // first pixel
+         raw += img_n;
+         cur += out_n;
+         prior += out_n;
+      } else if (depth == 16) {
+         if (img_n != out_n) {
+            cur[filter_bytes]   = 255; // first pixel top byte
+            cur[filter_bytes+1] = 255; // first pixel bottom byte
+         }
+         raw += filter_bytes;
+         cur += output_bytes;
+         prior += output_bytes;
+      } else {
+         raw += 1;
+         cur += 1;
+         prior += 1;
+      }
+
+      // this is a little gross, so that we don't switch per-pixel or per-component
+      if (depth < 8 || img_n == out_n) {
+         int nk = (width - 1)*filter_bytes;
+         #define STBI__CASE(f) \
+             case f:     \
+                for (k=0; k < nk; ++k)
+         switch (filter) {
+            // "none" filter turns into a memcpy here; make that explicit.
+            case STBI__F_none:         memcpy(cur, raw, nk); break;
+            STBI__CASE(STBI__F_sub)          { cur[k] = STBI__BYTECAST(raw[k] + cur[k-filter_bytes]); } break;
+            STBI__CASE(STBI__F_up)           { cur[k] = STBI__BYTECAST(raw[k] + prior[k]); } break;
+            STBI__CASE(STBI__F_avg)          { cur[k] = STBI__BYTECAST(raw[k] + ((prior[k] + cur[k-filter_bytes])>>1)); } break;
+            STBI__CASE(STBI__F_paeth)        { cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k-filter_bytes],prior[k],prior[k-filter_bytes])); } break;
+            STBI__CASE(STBI__F_avg_first)    { cur[k] = STBI__BYTECAST(raw[k] + (cur[k-filter_bytes] >> 1)); } break;
+            STBI__CASE(STBI__F_paeth_first)  { cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k-filter_bytes],0,0)); } break;
+         }
+         #undef STBI__CASE
+         raw += nk;
+      } else {
+         STBI_ASSERT(img_n+1 == out_n);
+         #define STBI__CASE(f) \
+             case f:     \
+                for (i=x-1; i >= 1; --i, cur[filter_bytes]=255,raw+=filter_bytes,cur+=output_bytes,prior+=output_bytes) \
+                   for (k=0; k < filter_bytes; ++k)
+         switch (filter) {
+            STBI__CASE(STBI__F_none)         { cur[k] = raw[k]; } break;
+            STBI__CASE(STBI__F_sub)          { cur[k] = STBI__BYTECAST(raw[k] + cur[k- output_bytes]); } break;
+            STBI__CASE(STBI__F_up)           { cur[k] = STBI__BYTECAST(raw[k] + prior[k]); } break;
+            STBI__CASE(STBI__F_avg)          { cur[k] = STBI__BYTECAST(raw[k] + ((prior[k] + cur[k- output_bytes])>>1)); } break;
+            STBI__CASE(STBI__F_paeth)        { cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k- output_bytes],prior[k],prior[k- output_bytes])); } break;
+            STBI__CASE(STBI__F_avg_first)    { cur[k] = STBI__BYTECAST(raw[k] + (cur[k- output_bytes] >> 1)); } break;
+            STBI__CASE(STBI__F_paeth_first)  { cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k- output_bytes],0,0)); } break;
+         }
+         #undef STBI__CASE
+
+         // the loop above sets the high byte of the pixels' alpha, but for
+         // 16 bit png files we also need the low byte set. we'll do that here.
+         if (depth == 16) {
+            cur = a->out + stride*j; // start at the beginning of the row again
+            for (i=0; i < x; ++i,cur+=output_bytes) {
+               cur[filter_bytes+1] = 255;
+            }
+         }
+      }
+   }
+
+   // we make a separate pass to expand bits to pixels; for performance,
+   // this could run two scanlines behind the above code, so it won't
+   // intefere with filtering but will still be in the cache.
+   if (depth < 8) {
+      for (j=0; j < y; ++j) {
+         stbi_uc *cur = a->out + stride*j;
+         stbi_uc *in  = a->out + stride*j + x*out_n - img_width_bytes;
+         // unpack 1/2/4-bit into a 8-bit buffer. allows us to keep the common 8-bit path optimal at minimal cost for 1/2/4-bit
+         // png guarante byte alignment, if width is not multiple of 8/4/2 we'll decode dummy trailing data that will be skipped in the later loop
+         stbi_uc scale = (color == 0) ? stbi__depth_scale_table[depth] : 1; // scale grayscale values to 0..255 range
+
+         // note that the final byte might overshoot and write more data than desired.
+         // we can allocate enough data that this never writes out of memory, but it
+         // could also overwrite the next scanline. can it overwrite non-empty data
+         // on the next scanline? yes, consider 1-pixel-wide scanlines with 1-bit-per-pixel.
+         // so we need to explicitly clamp the final ones
+
+         if (depth == 4) {
+            for (k=x*img_n; k >= 2; k-=2, ++in) {
+               *cur++ = scale * ((*in >> 4)       );
+               *cur++ = scale * ((*in     ) & 0x0f);
+            }
+            if (k > 0) *cur++ = scale * ((*in >> 4)       );
+         } else if (depth == 2) {
+            for (k=x*img_n; k >= 4; k-=4, ++in) {
+               *cur++ = scale * ((*in >> 6)       );
+               *cur++ = scale * ((*in >> 4) & 0x03);
+               *cur++ = scale * ((*in >> 2) & 0x03);
+               *cur++ = scale * ((*in     ) & 0x03);
+            }
+            if (k > 0) *cur++ = scale * ((*in >> 6)       );
+            if (k > 1) *cur++ = scale * ((*in >> 4) & 0x03);
+            if (k > 2) *cur++ = scale * ((*in >> 2) & 0x03);
+         } else if (depth == 1) {
+            for (k=x*img_n; k >= 8; k-=8, ++in) {
+               *cur++ = scale * ((*in >> 7)       );
+               *cur++ = scale * ((*in >> 6) & 0x01);
+               *cur++ = scale * ((*in >> 5) & 0x01);
+               *cur++ = scale * ((*in >> 4) & 0x01);
+               *cur++ = scale * ((*in >> 3) & 0x01);
+               *cur++ = scale * ((*in >> 2) & 0x01);
+               *cur++ = scale * ((*in >> 1) & 0x01);
+               *cur++ = scale * ((*in     ) & 0x01);
+            }
+            if (k > 0) *cur++ = scale * ((*in >> 7)       );
+            if (k > 1) *cur++ = scale * ((*in >> 6) & 0x01);
+            if (k > 2) *cur++ = scale * ((*in >> 5) & 0x01);
+            if (k > 3) *cur++ = scale * ((*in >> 4) & 0x01);
+            if (k > 4) *cur++ = scale * ((*in >> 3) & 0x01);
+            if (k > 5) *cur++ = scale * ((*in >> 2) & 0x01);
+            if (k > 6) *cur++ = scale * ((*in >> 1) & 0x01);
+         }
+         if (img_n != out_n) {
+            int q;
+            // insert alpha = 255
+            cur = a->out + stride*j;
+            if (img_n == 1) {
+               for (q=x-1; q >= 0; --q) {
+                  cur[q*2+1] = 255;
+                  cur[q*2+0] = cur[q];
+               }
+            } else {
+               STBI_ASSERT(img_n == 3);
+               for (q=x-1; q >= 0; --q) {
+                  cur[q*4+3] = 255;
+                  cur[q*4+2] = cur[q*3+2];
+                  cur[q*4+1] = cur[q*3+1];
+                  cur[q*4+0] = cur[q*3+0];
+               }
+            }
+         }
+      }
+   } else if (depth == 16) {
+      // force the image data from big-endian to platform-native.
+      // this is done in a separate pass due to the decoding relying
+      // on the data being untouched, but could probably be done
+      // per-line during decode if care is taken.
+      stbi_uc *cur = a->out;
+      stbi__uint16 *cur16 = (stbi__uint16*)cur;
+
+      for(i=0; i < x*y*out_n; ++i,cur16++,cur+=2) {
+         *cur16 = (cur[0] << 8) | cur[1];
+      }
+   }
+
+   return 1;
+}
+
+static int stbi__create_png_image(stbi__png *a, stbi_uc *image_data, stbi__uint32 image_data_len, int out_n, int depth, int color, int interlaced)
+{
+   int bytes = (depth == 16 ? 2 : 1);
+   int out_bytes = out_n * bytes;
+   stbi_uc *final;
+   int p;
+   if (!interlaced)
+      return stbi__create_png_image_raw(a, image_data, image_data_len, out_n, a->s->img_x, a->s->img_y, depth, color);
+
+   // de-interlacing
+   final = (stbi_uc *) stbi__malloc_mad3(a->s->img_x, a->s->img_y, out_bytes, 0);
+   for (p=0; p < 7; ++p) {
+      int xorig[] = { 0,4,0,2,0,1,0 };
+      int yorig[] = { 0,0,4,0,2,0,1 };
+      int xspc[]  = { 8,8,4,4,2,2,1 };
+      int yspc[]  = { 8,8,8,4,4,2,2 };
+      int i,j,x,y;
+      // pass1_x[4] = 0, pass1_x[5] = 1, pass1_x[12] = 1
+      x = (a->s->img_x - xorig[p] + xspc[p]-1) / xspc[p];
+      y = (a->s->img_y - yorig[p] + yspc[p]-1) / yspc[p];
+      if (x && y) {
+         stbi__uint32 img_len = ((((a->s->img_n * x * depth) + 7) >> 3) + 1) * y;
+         if (!stbi__create_png_image_raw(a, image_data, image_data_len, out_n, x, y, depth, color)) {
+            STBI_FREE(final);
+            return 0;
+         }
+         for (j=0; j < y; ++j) {
+            for (i=0; i < x; ++i) {
+               int out_y = j*yspc[p]+yorig[p];
+               int out_x = i*xspc[p]+xorig[p];
+               memcpy(final + out_y*a->s->img_x*out_bytes + out_x*out_bytes,
+                      a->out + (j*x+i)*out_bytes, out_bytes);
+            }
+         }
+         STBI_FREE(a->out);
+         image_data += img_len;
+         image_data_len -= img_len;
+      }
+   }
+   a->out = final;
+
+   return 1;
+}
+
+static int stbi__compute_transparency(stbi__png *z, stbi_uc tc[3], int out_n)
+{
+   stbi__context *s = z->s;
+   stbi__uint32 i, pixel_count = s->img_x * s->img_y;
+   stbi_uc *p = z->out;
+
+   // compute color-based transparency, assuming we've
+   // already got 255 as the alpha value in the output
+   STBI_ASSERT(out_n == 2 || out_n == 4);
+
+   if (out_n == 2) {
+      for (i=0; i < pixel_count; ++i) {
+         p[1] = (p[0] == tc[0] ? 0 : 255);
+         p += 2;
+      }
+   } else {
+      for (i=0; i < pixel_count; ++i) {
+         if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2])
+            p[3] = 0;
+         p += 4;
+      }
+   }
+   return 1;
+}
+
+static int stbi__compute_transparency16(stbi__png *z, stbi__uint16 tc[3], int out_n)
+{
+   stbi__context *s = z->s;
+   stbi__uint32 i, pixel_count = s->img_x * s->img_y;
+   stbi__uint16 *p = (stbi__uint16*) z->out;
+
+   // compute color-based transparency, assuming we've
+   // already got 65535 as the alpha value in the output
+   STBI_ASSERT(out_n == 2 || out_n == 4);
+
+   if (out_n == 2) {
+      for (i = 0; i < pixel_count; ++i) {
+         p[1] = (p[0] == tc[0] ? 0 : 65535);
+         p += 2;
+      }
+   } else {
+      for (i = 0; i < pixel_count; ++i) {
+         if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2])
+            p[3] = 0;
+         p += 4;
+      }
+   }
+   return 1;
+}
+
+static int stbi__expand_png_palette(stbi__png *a, stbi_uc *palette, int len, int pal_img_n)
+{
+   stbi__uint32 i, pixel_count = a->s->img_x * a->s->img_y;
+   stbi_uc *p, *temp_out, *orig = a->out;
+
+   p = (stbi_uc *) stbi__malloc_mad2(pixel_count, pal_img_n, 0);
+   if (p == NULL) return stbi__err("outofmem", "Out of memory");
+
+   // between here and free(out) below, exitting would leak
+   temp_out = p;
+
+   if (pal_img_n == 3) {
+      for (i=0; i < pixel_count; ++i) {
+         int n = orig[i]*4;
+         p[0] = palette[n  ];
+         p[1] = palette[n+1];
+         p[2] = palette[n+2];
+         p += 3;
+      }
+   } else {
+      for (i=0; i < pixel_count; ++i) {
+         int n = orig[i]*4;
+         p[0] = palette[n  ];
+         p[1] = palette[n+1];
+         p[2] = palette[n+2];
+         p[3] = palette[n+3];
+         p += 4;
+      }
+   }
+   STBI_FREE(a->out);
+   a->out = temp_out;
+
+   STBI_NOTUSED(len);
+
+   return 1;
+}
+
+static int stbi__unpremultiply_on_load = 0;
+static int stbi__de_iphone_flag = 0;
+
+STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply)
+{
+   stbi__unpremultiply_on_load = flag_true_if_should_unpremultiply;
+}
+
+STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert)
+{
+   stbi__de_iphone_flag = flag_true_if_should_convert;
+}
+
+static void stbi__de_iphone(stbi__png *z)
+{
+   stbi__context *s = z->s;
+   stbi__uint32 i, pixel_count = s->img_x * s->img_y;
+   stbi_uc *p = z->out;
+
+   if (s->img_out_n == 3) {  // convert bgr to rgb
+      for (i=0; i < pixel_count; ++i) {
+         stbi_uc t = p[0];
+         p[0] = p[2];
+         p[2] = t;
+         p += 3;
+      }
+   } else {
+      STBI_ASSERT(s->img_out_n == 4);
+      if (stbi__unpremultiply_on_load) {
+         // convert bgr to rgb and unpremultiply
+         for (i=0; i < pixel_count; ++i) {
+            stbi_uc a = p[3];
+            stbi_uc t = p[0];
+            if (a) {
+               p[0] = p[2] * 255 / a;
+               p[1] = p[1] * 255 / a;
+               p[2] =  t   * 255 / a;
+            } else {
+               p[0] = p[2];
+               p[2] = t;
+            }
+            p += 4;
+         }
+      } else {
+         // convert bgr to rgb
+         for (i=0; i < pixel_count; ++i) {
+            stbi_uc t = p[0];
+            p[0] = p[2];
+            p[2] = t;
+            p += 4;
+         }
+      }
+   }
+}
+
+#define STBI__PNG_TYPE(a,b,c,d)  (((a) << 24) + ((b) << 16) + ((c) << 8) + (d))
+
+static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
+{
+   stbi_uc palette[1024], pal_img_n=0;
+   stbi_uc has_trans=0, tc[3];
+   stbi__uint16 tc16[3];
+   stbi__uint32 ioff=0, idata_limit=0, i, pal_len=0;
+   int first=1,k,interlace=0, color=0, is_iphone=0;
+   stbi__context *s = z->s;
+
+   z->expanded = NULL;
+   z->idata = NULL;
+   z->out = NULL;
+
+   if (!stbi__check_png_header(s)) return 0;
+
+   if (scan == STBI__SCAN_type) return 1;
+
+   for (;;) {
+      stbi__pngchunk c = stbi__get_chunk_header(s);
+      switch (c.type) {
+         case STBI__PNG_TYPE('C','g','B','I'):
+            is_iphone = 1;
+            stbi__skip(s, c.length);
+            break;
+         case STBI__PNG_TYPE('I','H','D','R'): {
+            int comp,filter;
+            if (!first) return stbi__err("multiple IHDR","Corrupt PNG");
+            first = 0;
+            if (c.length != 13) return stbi__err("bad IHDR len","Corrupt PNG");
+            s->img_x = stbi__get32be(s); if (s->img_x > (1 << 24)) return stbi__err("too large","Very large image (corrupt?)");
+            s->img_y = stbi__get32be(s); if (s->img_y > (1 << 24)) return stbi__err("too large","Very large image (corrupt?)");
+            z->depth = stbi__get8(s);  if (z->depth != 1 && z->depth != 2 && z->depth != 4 && z->depth != 8 && z->depth != 16)  return stbi__err("1/2/4/8/16-bit only","PNG not supported: 1/2/4/8/16-bit only");
+            color = stbi__get8(s);  if (color > 6)         return stbi__err("bad ctype","Corrupt PNG");
+			if (color == 3 && z->depth == 16)                  return stbi__err("bad ctype","Corrupt PNG");
+            if (color == 3) pal_img_n = 3; else if (color & 1) return stbi__err("bad ctype","Corrupt PNG");
+            comp  = stbi__get8(s);  if (comp) return stbi__err("bad comp method","Corrupt PNG");
+            filter= stbi__get8(s);  if (filter) return stbi__err("bad filter method","Corrupt PNG");
+            interlace = stbi__get8(s); if (interlace>1) return stbi__err("bad interlace method","Corrupt PNG");
+            if (!s->img_x || !s->img_y) return stbi__err("0-pixel image","Corrupt PNG");
+            if (!pal_img_n) {
+               s->img_n = (color & 2 ? 3 : 1) + (color & 4 ? 1 : 0);
+               if ((1 << 30) / s->img_x / s->img_n < s->img_y) return stbi__err("too large", "Image too large to decode");
+               if (scan == STBI__SCAN_header) return 1;
+            } else {
+               // if paletted, then pal_n is our final components, and
+               // img_n is # components to decompress/filter.
+               s->img_n = 1;
+               if ((1 << 30) / s->img_x / 4 < s->img_y) return stbi__err("too large","Corrupt PNG");
+               // if SCAN_header, have to scan to see if we have a tRNS
+            }
+            break;
+         }
+
+         case STBI__PNG_TYPE('P','L','T','E'):  {
+            if (first) return stbi__err("first not IHDR", "Corrupt PNG");
+            if (c.length > 256*3) return stbi__err("invalid PLTE","Corrupt PNG");
+            pal_len = c.length / 3;
+            if (pal_len * 3 != c.length) return stbi__err("invalid PLTE","Corrupt PNG");
+            for (i=0; i < pal_len; ++i) {
+               palette[i*4+0] = stbi__get8(s);
+               palette[i*4+1] = stbi__get8(s);
+               palette[i*4+2] = stbi__get8(s);
+               palette[i*4+3] = 255;
+            }
+            break;
+         }
+
+         case STBI__PNG_TYPE('t','R','N','S'): {
+            if (first) return stbi__err("first not IHDR", "Corrupt PNG");
+            if (z->idata) return stbi__err("tRNS after IDAT","Corrupt PNG");
+            if (pal_img_n) {
+               if (scan == STBI__SCAN_header) { s->img_n = 4; return 1; }
+               if (pal_len == 0) return stbi__err("tRNS before PLTE","Corrupt PNG");
+               if (c.length > pal_len) return stbi__err("bad tRNS len","Corrupt PNG");
+               pal_img_n = 4;
+               for (i=0; i < c.length; ++i)
+                  palette[i*4+3] = stbi__get8(s);
+            } else {
+               if (!(s->img_n & 1)) return stbi__err("tRNS with alpha","Corrupt PNG");
+               if (c.length != (stbi__uint32) s->img_n*2) return stbi__err("bad tRNS len","Corrupt PNG");
+               has_trans = 1;
+               if (z->depth == 16) {
+                  for (k = 0; k < s->img_n; ++k) tc16[k] = (stbi__uint16)stbi__get16be(s); // copy the values as-is
+               } else {
+                  for (k = 0; k < s->img_n; ++k) tc[k] = (stbi_uc)(stbi__get16be(s) & 255) * stbi__depth_scale_table[z->depth]; // non 8-bit images will be larger
+               }
+            }
+            break;
+         }
+
+         case STBI__PNG_TYPE('I','D','A','T'): {
+            if (first) return stbi__err("first not IHDR", "Corrupt PNG");
+            if (pal_img_n && !pal_len) return stbi__err("no PLTE","Corrupt PNG");
+            if (scan == STBI__SCAN_header) { s->img_n = pal_img_n; return 1; }
+            if ((int)(ioff + c.length) < (int)ioff) return 0;
+            if (ioff + c.length > idata_limit) {
+               stbi__uint32 idata_limit_old = idata_limit;
+               stbi_uc *p;
+               if (idata_limit == 0) idata_limit = c.length > 4096 ? c.length : 4096;
+               while (ioff + c.length > idata_limit)
+                  idata_limit *= 2;
+               STBI_NOTUSED(idata_limit_old);
+               p = (stbi_uc *) STBI_REALLOC_SIZED(z->idata, idata_limit_old, idata_limit); if (p == NULL) return stbi__err("outofmem", "Out of memory");
+               z->idata = p;
+            }
+            if (!stbi__getn(s, z->idata+ioff,c.length)) return stbi__err("outofdata","Corrupt PNG");
+            ioff += c.length;
+            break;
+         }
+
+         case STBI__PNG_TYPE('I','E','N','D'): {
+            stbi__uint32 raw_len, bpl;
+            if (first) return stbi__err("first not IHDR", "Corrupt PNG");
+            if (scan != STBI__SCAN_load) return 1;
+            if (z->idata == NULL) return stbi__err("no IDAT","Corrupt PNG");
+            // initial guess for decoded data size to avoid unnecessary reallocs
+            bpl = (s->img_x * z->depth + 7) / 8; // bytes per line, per component
+            raw_len = bpl * s->img_y * s->img_n /* pixels */ + s->img_y /* filter mode per row */;
+            z->expanded = (stbi_uc *) stbi_zlib_decode_malloc_guesssize_headerflag((char *) z->idata, ioff, raw_len, (int *) &raw_len, !is_iphone);
+            if (z->expanded == NULL) return 0; // zlib should set error
+            STBI_FREE(z->idata); z->idata = NULL;
+            if ((req_comp == s->img_n+1 && req_comp != 3 && !pal_img_n) || has_trans)
+               s->img_out_n = s->img_n+1;
+            else
+               s->img_out_n = s->img_n;
+            if (!stbi__create_png_image(z, z->expanded, raw_len, s->img_out_n, z->depth, color, interlace)) return 0;
+            if (has_trans) {
+               if (z->depth == 16) {
+                  if (!stbi__compute_transparency16(z, tc16, s->img_out_n)) return 0;
+               } else {
+                  if (!stbi__compute_transparency(z, tc, s->img_out_n)) return 0;
+               }
+            }
+            if (is_iphone && stbi__de_iphone_flag && s->img_out_n > 2)
+               stbi__de_iphone(z);
+            if (pal_img_n) {
+               // pal_img_n == 3 or 4
+               s->img_n = pal_img_n; // record the actual colors we had
+               s->img_out_n = pal_img_n;
+               if (req_comp >= 3) s->img_out_n = req_comp;
+               if (!stbi__expand_png_palette(z, palette, pal_len, s->img_out_n))
+                  return 0;
+            }
+            STBI_FREE(z->expanded); z->expanded = NULL;
+            return 1;
+         }
+
+         default:
+            // if critical, fail
+            if (first) return stbi__err("first not IHDR", "Corrupt PNG");
+            if ((c.type & (1 << 29)) == 0) {
+               #ifndef STBI_NO_FAILURE_STRINGS
+               // not threadsafe
+               static char invalid_chunk[] = "XXXX PNG chunk not known";
+               invalid_chunk[0] = STBI__BYTECAST(c.type >> 24);
+               invalid_chunk[1] = STBI__BYTECAST(c.type >> 16);
+               invalid_chunk[2] = STBI__BYTECAST(c.type >>  8);
+               invalid_chunk[3] = STBI__BYTECAST(c.type >>  0);
+               #endif
+               return stbi__err(invalid_chunk, "PNG not supported: unknown PNG chunk type");
+            }
+            stbi__skip(s, c.length);
+            break;
+      }
+      // end of PNG chunk, read and skip CRC
+      stbi__get32be(s);
+   }
+}
+
+static void *stbi__do_png(stbi__png *p, int *x, int *y, int *n, int req_comp, stbi__result_info *ri)
+{
+   void *result=NULL;
+   if (req_comp < 0 || req_comp > 4) return stbi__errpuc("bad req_comp", "Internal error");
+   if (stbi__parse_png_file(p, STBI__SCAN_load, req_comp)) {
+      ri->bits_per_channel = p->depth;
+      result = p->out;
+      p->out = NULL;
+      if (req_comp && req_comp != p->s->img_out_n) {
+         if (ri->bits_per_channel == 8)
+            result = stbi__convert_format((unsigned char *) result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y);
+         else
+            result = stbi__convert_format16((stbi__uint16 *) result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y);
+         p->s->img_out_n = req_comp;
+         if (result == NULL) return result;
+      }
+      *x = p->s->img_x;
+      *y = p->s->img_y;
+      if (n) *n = p->s->img_n;
+   }
+   STBI_FREE(p->out);      p->out      = NULL;
+   STBI_FREE(p->expanded); p->expanded = NULL;
+   STBI_FREE(p->idata);    p->idata    = NULL;
+
+   return result;
+}
+
+static void *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
+{
+   stbi__png p;
+   p.s = s;
+   return stbi__do_png(&p, x,y,comp,req_comp, ri);
+}
+
+static int stbi__png_test(stbi__context *s)
+{
+   int r;
+   r = stbi__check_png_header(s);
+   stbi__rewind(s);
+   return r;
+}
+
+static int stbi__png_info_raw(stbi__png *p, int *x, int *y, int *comp)
+{
+   if (!stbi__parse_png_file(p, STBI__SCAN_header, 0)) {
+      stbi__rewind( p->s );
+      return 0;
+   }
+   if (x) *x = p->s->img_x;
+   if (y) *y = p->s->img_y;
+   if (comp) *comp = p->s->img_n;
+   return 1;
+}
+
+static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp)
+{
+   stbi__png p;
+   p.s = s;
+   return stbi__png_info_raw(&p, x, y, comp);
+}
+#endif
+
+// Microsoft/Windows BMP image
+
+#ifndef STBI_NO_BMP
+static int stbi__bmp_test_raw(stbi__context *s)
+{
+   int r;
+   int sz;
+   if (stbi__get8(s) != 'B') return 0;
+   if (stbi__get8(s) != 'M') return 0;
+   stbi__get32le(s); // discard filesize
+   stbi__get16le(s); // discard reserved
+   stbi__get16le(s); // discard reserved
+   stbi__get32le(s); // discard data offset
+   sz = stbi__get32le(s);
+   r = (sz == 12 || sz == 40 || sz == 56 || sz == 108 || sz == 124);
+   return r;
+}
+
+static int stbi__bmp_test(stbi__context *s)
+{
+   int r = stbi__bmp_test_raw(s);
+   stbi__rewind(s);
+   return r;
+}
+
+
+// returns 0..31 for the highest set bit
+static int stbi__high_bit(unsigned int z)
+{
+   int n=0;
+   if (z == 0) return -1;
+   if (z >= 0x10000) n += 16, z >>= 16;
+   if (z >= 0x00100) n +=  8, z >>=  8;
+   if (z >= 0x00010) n +=  4, z >>=  4;
+   if (z >= 0x00004) n +=  2, z >>=  2;
+   if (z >= 0x00002) n +=  1, z >>=  1;
+   return n;
+}
+
+static int stbi__bitcount(unsigned int a)
+{
+   a = (a & 0x55555555) + ((a >>  1) & 0x55555555); // max 2
+   a = (a & 0x33333333) + ((a >>  2) & 0x33333333); // max 4
+   a = (a + (a >> 4)) & 0x0f0f0f0f; // max 8 per 4, now 8 bits
+   a = (a + (a >> 8)); // max 16 per 8 bits
+   a = (a + (a >> 16)); // max 32 per 8 bits
+   return a & 0xff;
+}
+
+static int stbi__shiftsigned(int v, int shift, int bits)
+{
+   int result;
+   int z=0;
+
+   if (shift < 0) v <<= -shift;
+   else v >>= shift;
+   result = v;
+
+   z = bits;
+   while (z < 8) {
+      result += v >> z;
+      z += bits;
+   }
+   return result;
+}
+
+typedef struct
+{
+   int bpp, offset, hsz;
+   unsigned int mr,mg,mb,ma, all_a;
+} stbi__bmp_data;
+
+static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info)
+{
+   int hsz;
+   if (stbi__get8(s) != 'B' || stbi__get8(s) != 'M') return stbi__errpuc("not BMP", "Corrupt BMP");
+   stbi__get32le(s); // discard filesize
+   stbi__get16le(s); // discard reserved
+   stbi__get16le(s); // discard reserved
+   info->offset = stbi__get32le(s);
+   info->hsz = hsz = stbi__get32le(s);
+   info->mr = info->mg = info->mb = info->ma = 0;
+
+   if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) return stbi__errpuc("unknown BMP", "BMP type not supported: unknown");
+   if (hsz == 12) {
+      s->img_x = stbi__get16le(s);
+      s->img_y = stbi__get16le(s);
+   } else {
+      s->img_x = stbi__get32le(s);
+      s->img_y = stbi__get32le(s);
+   }
+   if (stbi__get16le(s) != 1) return stbi__errpuc("bad BMP", "bad BMP");
+   info->bpp = stbi__get16le(s);
+   if (info->bpp == 1) return stbi__errpuc("monochrome", "BMP type not supported: 1-bit");
+   if (hsz != 12) {
+      int compress = stbi__get32le(s);
+      if (compress == 1 || compress == 2) return stbi__errpuc("BMP RLE", "BMP type not supported: RLE");
+      stbi__get32le(s); // discard sizeof
+      stbi__get32le(s); // discard hres
+      stbi__get32le(s); // discard vres
+      stbi__get32le(s); // discard colorsused
+      stbi__get32le(s); // discard max important
+      if (hsz == 40 || hsz == 56) {
+         if (hsz == 56) {
+            stbi__get32le(s);
+            stbi__get32le(s);
+            stbi__get32le(s);
+            stbi__get32le(s);
+         }
+         if (info->bpp == 16 || info->bpp == 32) {
+            if (compress == 0) {
+               if (info->bpp == 32) {
+                  info->mr = 0xffu << 16;
+                  info->mg = 0xffu <<  8;
+                  info->mb = 0xffu <<  0;
+                  info->ma = 0xffu << 24;
+                  info->all_a = 0; // if all_a is 0 at end, then we loaded alpha channel but it was all 0
+               } else {
+                  info->mr = 31u << 10;
+                  info->mg = 31u <<  5;
+                  info->mb = 31u <<  0;
+               }
+            } else if (compress == 3) {
+               info->mr = stbi__get32le(s);
+               info->mg = stbi__get32le(s);
+               info->mb = stbi__get32le(s);
+               // not documented, but generated by photoshop and handled by mspaint
+               if (info->mr == info->mg && info->mg == info->mb) {
+                  // ?!?!?
+                  return stbi__errpuc("bad BMP", "bad BMP");
+               }
+            } else
+               return stbi__errpuc("bad BMP", "bad BMP");
+         }
+      } else {
+         int i;
+         if (hsz != 108 && hsz != 124)
+            return stbi__errpuc("bad BMP", "bad BMP");
+         info->mr = stbi__get32le(s);
+         info->mg = stbi__get32le(s);
+         info->mb = stbi__get32le(s);
+         info->ma = stbi__get32le(s);
+         stbi__get32le(s); // discard color space
+         for (i=0; i < 12; ++i)
+            stbi__get32le(s); // discard color space parameters
+         if (hsz == 124) {
+            stbi__get32le(s); // discard rendering intent
+            stbi__get32le(s); // discard offset of profile data
+            stbi__get32le(s); // discard size of profile data
+            stbi__get32le(s); // discard reserved
+         }
+      }
+   }
+   return (void *) 1;
+}
+
+
+static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
+{
+   stbi_uc *out;
+   unsigned int mr=0,mg=0,mb=0,ma=0, all_a;
+   stbi_uc pal[256][4];
+   int psize=0,i,j,width;
+   int flip_vertically, pad, target;
+   stbi__bmp_data info;
+   STBI_NOTUSED(ri);
+
+   info.all_a = 255;
+   if (stbi__bmp_parse_header(s, &info) == NULL)
+      return NULL; // error code already set
+
+   flip_vertically = ((int) s->img_y) > 0;
+   s->img_y = abs((int) s->img_y);
+
+   mr = info.mr;
+   mg = info.mg;
+   mb = info.mb;
+   ma = info.ma;
+   all_a = info.all_a;
+
+   if (info.hsz == 12) {
+      if (info.bpp < 24)
+         psize = (info.offset - 14 - 24) / 3;
+   } else {
+      if (info.bpp < 16)
+         psize = (info.offset - 14 - info.hsz) >> 2;
+   }
+
+   s->img_n = ma ? 4 : 3;
+   if (req_comp && req_comp >= 3) // we can directly decode 3 or 4
+      target = req_comp;
+   else
+      target = s->img_n; // if they want monochrome, we'll post-convert
+
+   // sanity-check size
+   if (!stbi__mad3sizes_valid(target, s->img_x, s->img_y, 0))
+      return stbi__errpuc("too large", "Corrupt BMP");
+
+   out = (stbi_uc *) stbi__malloc_mad3(target, s->img_x, s->img_y, 0);
+   if (!out) return stbi__errpuc("outofmem", "Out of memory");
+   if (info.bpp < 16) {
+      int z=0;
+      if (psize == 0 || psize > 256) { STBI_FREE(out); return stbi__errpuc("invalid", "Corrupt BMP"); }
+      for (i=0; i < psize; ++i) {
+         pal[i][2] = stbi__get8(s);
+         pal[i][1] = stbi__get8(s);
+         pal[i][0] = stbi__get8(s);
+         if (info.hsz != 12) stbi__get8(s);
+         pal[i][3] = 255;
+      }
+      stbi__skip(s, info.offset - 14 - info.hsz - psize * (info.hsz == 12 ? 3 : 4));
+      if (info.bpp == 4) width = (s->img_x + 1) >> 1;
+      else if (info.bpp == 8) width = s->img_x;
+      else { STBI_FREE(out); return stbi__errpuc("bad bpp", "Corrupt BMP"); }
+      pad = (-width)&3;
+      for (j=0; j < (int) s->img_y; ++j) {
+         for (i=0; i < (int) s->img_x; i += 2) {
+            int v=stbi__get8(s),v2=0;
+            if (info.bpp == 4) {
+               v2 = v & 15;
+               v >>= 4;
+            }
+            out[z++] = pal[v][0];
+            out[z++] = pal[v][1];
+            out[z++] = pal[v][2];
+            if (target == 4) out[z++] = 255;
+            if (i+1 == (int) s->img_x) break;
+            v = (info.bpp == 8) ? stbi__get8(s) : v2;
+            out[z++] = pal[v][0];
+            out[z++] = pal[v][1];
+            out[z++] = pal[v][2];
+            if (target == 4) out[z++] = 255;
+         }
+         stbi__skip(s, pad);
+      }
+   } else {
+      int rshift=0,gshift=0,bshift=0,ashift=0,rcount=0,gcount=0,bcount=0,acount=0;
+      int z = 0;
+      int easy=0;
+      stbi__skip(s, info.offset - 14 - info.hsz);
+      if (info.bpp == 24) width = 3 * s->img_x;
+      else if (info.bpp == 16) width = 2*s->img_x;
+      else /* bpp = 32 and pad = 0 */ width=0;
+      pad = (-width) & 3;
+      if (info.bpp == 24) {
+         easy = 1;
+      } else if (info.bpp == 32) {
+         if (mb == 0xff && mg == 0xff00 && mr == 0x00ff0000 && ma == 0xff000000)
+            easy = 2;
+      }
+      if (!easy) {
+         if (!mr || !mg || !mb) { STBI_FREE(out); return stbi__errpuc("bad masks", "Corrupt BMP"); }
+         // right shift amt to put high bit in position #7
+         rshift = stbi__high_bit(mr)-7; rcount = stbi__bitcount(mr);
+         gshift = stbi__high_bit(mg)-7; gcount = stbi__bitcount(mg);
+         bshift = stbi__high_bit(mb)-7; bcount = stbi__bitcount(mb);
+         ashift = stbi__high_bit(ma)-7; acount = stbi__bitcount(ma);
+      }
+      for (j=0; j < (int) s->img_y; ++j) {
+         if (easy) {
+            for (i=0; i < (int) s->img_x; ++i) {
+               unsigned char a;
+               out[z+2] = stbi__get8(s);
+               out[z+1] = stbi__get8(s);
+               out[z+0] = stbi__get8(s);
+               z += 3;
+               a = (easy == 2 ? stbi__get8(s) : 255);
+               all_a |= a;
+               if (target == 4) out[z++] = a;
+            }
+         } else {
+            int bpp = info.bpp;
+            for (i=0; i < (int) s->img_x; ++i) {
+               stbi__uint32 v = (bpp == 16 ? (stbi__uint32) stbi__get16le(s) : stbi__get32le(s));
+               int a;
+               out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mr, rshift, rcount));
+               out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mg, gshift, gcount));
+               out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mb, bshift, bcount));
+               a = (ma ? stbi__shiftsigned(v & ma, ashift, acount) : 255);
+               all_a |= a;
+               if (target == 4) out[z++] = STBI__BYTECAST(a);
+            }
+         }
+         stbi__skip(s, pad);
+      }
+   }
+
+   // if alpha channel is all 0s, replace with all 255s
+   if (target == 4 && all_a == 0)
+      for (i=4*s->img_x*s->img_y-1; i >= 0; i -= 4)
+         out[i] = 255;
+
+   if (flip_vertically) {
+      stbi_uc t;
+      for (j=0; j < (int) s->img_y>>1; ++j) {
+         stbi_uc *p1 = out +      j     *s->img_x*target;
+         stbi_uc *p2 = out + (s->img_y-1-j)*s->img_x*target;
+         for (i=0; i < (int) s->img_x*target; ++i) {
+            t = p1[i], p1[i] = p2[i], p2[i] = t;
+         }
+      }
+   }
+
+   if (req_comp && req_comp != target) {
+      out = stbi__convert_format(out, target, req_comp, s->img_x, s->img_y);
+      if (out == NULL) return out; // stbi__convert_format frees input on failure
+   }
+
+   *x = s->img_x;
+   *y = s->img_y;
+   if (comp) *comp = s->img_n;
+   return out;
+}
+#endif
+
+// Targa Truevision - TGA
+// by Jonathan Dummer
+#ifndef STBI_NO_TGA
+// returns STBI_rgb or whatever, 0 on error
+static int stbi__tga_get_comp(int bits_per_pixel, int is_grey, int* is_rgb16)
+{
+   // only RGB or RGBA (incl. 16bit) or grey allowed
+   if(is_rgb16) *is_rgb16 = 0;
+   switch(bits_per_pixel) {
+      case 8:  return STBI_grey;
+      case 16: if(is_grey) return STBI_grey_alpha;
+            // else: fall-through
+      case 15: if(is_rgb16) *is_rgb16 = 1;
+            return STBI_rgb;
+      case 24: // fall-through
+      case 32: return bits_per_pixel/8;
+      default: return 0;
+   }
+}
+
+static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp)
+{
+    int tga_w, tga_h, tga_comp, tga_image_type, tga_bits_per_pixel, tga_colormap_bpp;
+    int sz, tga_colormap_type;
+    stbi__get8(s);                   // discard Offset
+    tga_colormap_type = stbi__get8(s); // colormap type
+    if( tga_colormap_type > 1 ) {
+        stbi__rewind(s);
+        return 0;      // only RGB or indexed allowed
+    }
+    tga_image_type = stbi__get8(s); // image type
+    if ( tga_colormap_type == 1 ) { // colormapped (paletted) image
+        if (tga_image_type != 1 && tga_image_type != 9) {
+            stbi__rewind(s);
+            return 0;
+        }
+        stbi__skip(s,4);       // skip index of first colormap entry and number of entries
+        sz = stbi__get8(s);    //   check bits per palette color entry
+        if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) {
+            stbi__rewind(s);
+            return 0;
+        }
+        stbi__skip(s,4);       // skip image x and y origin
+        tga_colormap_bpp = sz;
+    } else { // "normal" image w/o colormap - only RGB or grey allowed, +/- RLE
+        if ( (tga_image_type != 2) && (tga_image_type != 3) && (tga_image_type != 10) && (tga_image_type != 11) ) {
+            stbi__rewind(s);
+            return 0; // only RGB or grey allowed, +/- RLE
+        }
+        stbi__skip(s,9); // skip colormap specification and image x/y origin
+        tga_colormap_bpp = 0;
+    }
+    tga_w = stbi__get16le(s);
+    if( tga_w < 1 ) {
+        stbi__rewind(s);
+        return 0;   // test width
+    }
+    tga_h = stbi__get16le(s);
+    if( tga_h < 1 ) {
+        stbi__rewind(s);
+        return 0;   // test height
+    }
+    tga_bits_per_pixel = stbi__get8(s); // bits per pixel
+    stbi__get8(s); // ignore alpha bits
+    if (tga_colormap_bpp != 0) {
+        if((tga_bits_per_pixel != 8) && (tga_bits_per_pixel != 16)) {
+            // when using a colormap, tga_bits_per_pixel is the size of the indexes
+            // I don't think anything but 8 or 16bit indexes makes sense
+            stbi__rewind(s);
+            return 0;
+        }
+        tga_comp = stbi__tga_get_comp(tga_colormap_bpp, 0, NULL);
+    } else {
+        tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3) || (tga_image_type == 11), NULL);
+    }
+    if(!tga_comp) {
+      stbi__rewind(s);
+      return 0;
+    }
+    if (x) *x = tga_w;
+    if (y) *y = tga_h;
+    if (comp) *comp = tga_comp;
+    return 1;                   // seems to have passed everything
+}
+
+static int stbi__tga_test(stbi__context *s)
+{
+   int res = 0;
+   int sz, tga_color_type;
+   stbi__get8(s);      //   discard Offset
+   tga_color_type = stbi__get8(s);   //   color type
+   if ( tga_color_type > 1 ) goto errorEnd;   //   only RGB or indexed allowed
+   sz = stbi__get8(s);   //   image type
+   if ( tga_color_type == 1 ) { // colormapped (paletted) image
+      if (sz != 1 && sz != 9) goto errorEnd; // colortype 1 demands image type 1 or 9
+      stbi__skip(s,4);       // skip index of first colormap entry and number of entries
+      sz = stbi__get8(s);    //   check bits per palette color entry
+      if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) goto errorEnd;
+      stbi__skip(s,4);       // skip image x and y origin
+   } else { // "normal" image w/o colormap
+      if ( (sz != 2) && (sz != 3) && (sz != 10) && (sz != 11) ) goto errorEnd; // only RGB or grey allowed, +/- RLE
+      stbi__skip(s,9); // skip colormap specification and image x/y origin
+   }
+   if ( stbi__get16le(s) < 1 ) goto errorEnd;      //   test width
+   if ( stbi__get16le(s) < 1 ) goto errorEnd;      //   test height
+   sz = stbi__get8(s);   //   bits per pixel
+   if ( (tga_color_type == 1) && (sz != 8) && (sz != 16) ) goto errorEnd; // for colormapped images, bpp is size of an index
+   if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) goto errorEnd;
+
+   res = 1; // if we got this far, everything's good and we can return 1 instead of 0
+
+errorEnd:
+   stbi__rewind(s);
+   return res;
+}
+
+// read 16bit value and convert to 24bit RGB
+static void stbi__tga_read_rgb16(stbi__context *s, stbi_uc* out)
+{
+   stbi__uint16 px = (stbi__uint16)stbi__get16le(s);
+   stbi__uint16 fiveBitMask = 31;
+   // we have 3 channels with 5bits each
+   int r = (px >> 10) & fiveBitMask;
+   int g = (px >> 5) & fiveBitMask;
+   int b = px & fiveBitMask;
+   // Note that this saves the data in RGB(A) order, so it doesn't need to be swapped later
+   out[0] = (stbi_uc)((r * 255)/31);
+   out[1] = (stbi_uc)((g * 255)/31);
+   out[2] = (stbi_uc)((b * 255)/31);
+
+   // some people claim that the most significant bit might be used for alpha
+   // (possibly if an alpha-bit is set in the "image descriptor byte")
+   // but that only made 16bit test images completely translucent..
+   // so let's treat all 15 and 16bit TGAs as RGB with no alpha.
+}
+
+static void *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
+{
+   //   read in the TGA header stuff
+   int tga_offset = stbi__get8(s);
+   int tga_indexed = stbi__get8(s);
+   int tga_image_type = stbi__get8(s);
+   int tga_is_RLE = 0;
+   int tga_palette_start = stbi__get16le(s);
+   int tga_palette_len = stbi__get16le(s);
+   int tga_palette_bits = stbi__get8(s);
+   int tga_x_origin = stbi__get16le(s);
+   int tga_y_origin = stbi__get16le(s);
+   int tga_width = stbi__get16le(s);
+   int tga_height = stbi__get16le(s);
+   int tga_bits_per_pixel = stbi__get8(s);
+   int tga_comp, tga_rgb16=0;
+   int tga_inverted = stbi__get8(s);
+   // int tga_alpha_bits = tga_inverted & 15; // the 4 lowest bits - unused (useless?)
+   //   image data
+   unsigned char *tga_data;
+   unsigned char *tga_palette = NULL;
+   int i, j;
+   unsigned char raw_data[4];
+   int RLE_count = 0;
+   int RLE_repeating = 0;
+   int read_next_pixel = 1;
+   STBI_NOTUSED(ri);
+
+   //   do a tiny bit of precessing
+   if ( tga_image_type >= 8 )
+   {
+      tga_image_type -= 8;
+      tga_is_RLE = 1;
+   }
+   tga_inverted = 1 - ((tga_inverted >> 5) & 1);
+
+   //   If I'm paletted, then I'll use the number of bits from the palette
+   if ( tga_indexed ) tga_comp = stbi__tga_get_comp(tga_palette_bits, 0, &tga_rgb16);
+   else tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3), &tga_rgb16);
+
+   if(!tga_comp) // shouldn't really happen, stbi__tga_test() should have ensured basic consistency
+      return stbi__errpuc("bad format", "Can't find out TGA pixelformat");
+
+   //   tga info
+   *x = tga_width;
+   *y = tga_height;
+   if (comp) *comp = tga_comp;
+
+   if (!stbi__mad3sizes_valid(tga_width, tga_height, tga_comp, 0))
+      return stbi__errpuc("too large", "Corrupt TGA");
+
+   tga_data = (unsigned char*)stbi__malloc_mad3(tga_width, tga_height, tga_comp, 0);
+   if (!tga_data) return stbi__errpuc("outofmem", "Out of memory");
+
+   // skip to the data's starting position (offset usually = 0)
+   stbi__skip(s, tga_offset );
+
+   if ( !tga_indexed && !tga_is_RLE && !tga_rgb16 ) {
+      for (i=0; i < tga_height; ++i) {
+         int row = tga_inverted ? tga_height -i - 1 : i;
+         stbi_uc *tga_row = tga_data + row*tga_width*tga_comp;
+         stbi__getn(s, tga_row, tga_width * tga_comp);
+      }
+   } else  {
+      //   do I need to load a palette?
+      if ( tga_indexed)
+      {
+         //   any data to skip? (offset usually = 0)
+         stbi__skip(s, tga_palette_start );
+         //   load the palette
+         tga_palette = (unsigned char*)stbi__malloc_mad2(tga_palette_len, tga_comp, 0);
+         if (!tga_palette) {
+            STBI_FREE(tga_data);
+            return stbi__errpuc("outofmem", "Out of memory");
+         }
+         if (tga_rgb16) {
+            stbi_uc *pal_entry = tga_palette;
+            STBI_ASSERT(tga_comp == STBI_rgb);
+            for (i=0; i < tga_palette_len; ++i) {
+               stbi__tga_read_rgb16(s, pal_entry);
+               pal_entry += tga_comp;
+            }
+         } else if (!stbi__getn(s, tga_palette, tga_palette_len * tga_comp)) {
+               STBI_FREE(tga_data);
+               STBI_FREE(tga_palette);
+               return stbi__errpuc("bad palette", "Corrupt TGA");
+         }
+      }
+      //   load the data
+      for (i=0; i < tga_width * tga_height; ++i)
+      {
+         //   if I'm in RLE mode, do I need to get a RLE stbi__pngchunk?
+         if ( tga_is_RLE )
+         {
+            if ( RLE_count == 0 )
+            {
+               //   yep, get the next byte as a RLE command
+               int RLE_cmd = stbi__get8(s);
+               RLE_count = 1 + (RLE_cmd & 127);
+               RLE_repeating = RLE_cmd >> 7;
+               read_next_pixel = 1;
+            } else if ( !RLE_repeating )
+            {
+               read_next_pixel = 1;
+            }
+         } else
+         {
+            read_next_pixel = 1;
+         }
+         //   OK, if I need to read a pixel, do it now
+         if ( read_next_pixel )
+         {
+            //   load however much data we did have
+            if ( tga_indexed )
+            {
+               // read in index, then perform the lookup
+               int pal_idx = (tga_bits_per_pixel == 8) ? stbi__get8(s) : stbi__get16le(s);
+               if ( pal_idx >= tga_palette_len ) {
+                  // invalid index
+                  pal_idx = 0;
+               }
+               pal_idx *= tga_comp;
+               for (j = 0; j < tga_comp; ++j) {
+                  raw_data[j] = tga_palette[pal_idx+j];
+               }
+            } else if(tga_rgb16) {
+               STBI_ASSERT(tga_comp == STBI_rgb);
+               stbi__tga_read_rgb16(s, raw_data);
+            } else {
+               //   read in the data raw
+               for (j = 0; j < tga_comp; ++j) {
+                  raw_data[j] = stbi__get8(s);
+               }
+            }
+            //   clear the reading flag for the next pixel
+            read_next_pixel = 0;
+         } // end of reading a pixel
+
+         // copy data
+         for (j = 0; j < tga_comp; ++j)
+           tga_data[i*tga_comp+j] = raw_data[j];
+
+         //   in case we're in RLE mode, keep counting down
+         --RLE_count;
+      }
+      //   do I need to invert the image?
+      if ( tga_inverted )
+      {
+         for (j = 0; j*2 < tga_height; ++j)
+         {
+            int index1 = j * tga_width * tga_comp;
+            int index2 = (tga_height - 1 - j) * tga_width * tga_comp;
+            for (i = tga_width * tga_comp; i > 0; --i)
+            {
+               unsigned char temp = tga_data[index1];
+               tga_data[index1] = tga_data[index2];
+               tga_data[index2] = temp;
+               ++index1;
+               ++index2;
+            }
+         }
+      }
+      //   clear my palette, if I had one
+      if ( tga_palette != NULL )
+      {
+         STBI_FREE( tga_palette );
+      }
+   }
+
+   // swap RGB - if the source data was RGB16, it already is in the right order
+   if (tga_comp >= 3 && !tga_rgb16)
+   {
+      unsigned char* tga_pixel = tga_data;
+      for (i=0; i < tga_width * tga_height; ++i)
+      {
+         unsigned char temp = tga_pixel[0];
+         tga_pixel[0] = tga_pixel[2];
+         tga_pixel[2] = temp;
+         tga_pixel += tga_comp;
+      }
+   }
+
+   // convert to target component count
+   if (req_comp && req_comp != tga_comp)
+      tga_data = stbi__convert_format(tga_data, tga_comp, req_comp, tga_width, tga_height);
+
+   //   the things I do to get rid of an error message, and yet keep
+   //   Microsoft's C compilers happy... [8^(
+   tga_palette_start = tga_palette_len = tga_palette_bits =
+         tga_x_origin = tga_y_origin = 0;
+   //   OK, done
+   return tga_data;
+}
+#endif
+
+// *************************************************************************************************
+// Photoshop PSD loader -- PD by Thatcher Ulrich, integration by Nicolas Schulz, tweaked by STB
+
+#ifndef STBI_NO_PSD
+static int stbi__psd_test(stbi__context *s)
+{
+   int r = (stbi__get32be(s) == 0x38425053);
+   stbi__rewind(s);
+   return r;
+}
+
+static int stbi__psd_decode_rle(stbi__context *s, stbi_uc *p, int pixelCount)
+{
+   int count, nleft, len;
+
+   count = 0;
+   while ((nleft = pixelCount - count) > 0) {
+      len = stbi__get8(s);
+      if (len == 128) {
+         // No-op.
+      } else if (len < 128) {
+         // Copy next len+1 bytes literally.
+         len++;
+         if (len > nleft) return 0; // corrupt data
+         count += len;
+         while (len) {
+            *p = stbi__get8(s);
+            p += 4;
+            len--;
+         }
+      } else if (len > 128) {
+         stbi_uc   val;
+         // Next -len+1 bytes in the dest are replicated from next source byte.
+         // (Interpret len as a negative 8-bit int.)
+         len = 257 - len;
+         if (len > nleft) return 0; // corrupt data
+         val = stbi__get8(s);
+         count += len;
+         while (len) {
+            *p = val;
+            p += 4;
+            len--;
+         }
+      }
+   }
+
+   return 1;
+}
+
+static void *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc)
+{
+   int pixelCount;
+   int channelCount, compression;
+   int channel, i;
+   int bitdepth;
+   int w,h;
+   stbi_uc *out;
+   STBI_NOTUSED(ri);
+
+   // Check identifier
+   if (stbi__get32be(s) != 0x38425053)   // "8BPS"
+      return stbi__errpuc("not PSD", "Corrupt PSD image");
+
+   // Check file type version.
+   if (stbi__get16be(s) != 1)
+      return stbi__errpuc("wrong version", "Unsupported version of PSD image");
+
+   // Skip 6 reserved bytes.
+   stbi__skip(s, 6 );
+
+   // Read the number of channels (R, G, B, A, etc).
+   channelCount = stbi__get16be(s);
+   if (channelCount < 0 || channelCount > 16)
+      return stbi__errpuc("wrong channel count", "Unsupported number of channels in PSD image");
+
+   // Read the rows and columns of the image.
+   h = stbi__get32be(s);
+   w = stbi__get32be(s);
+
+   // Make sure the depth is 8 bits.
+   bitdepth = stbi__get16be(s);
+   if (bitdepth != 8 && bitdepth != 16)
+      return stbi__errpuc("unsupported bit depth", "PSD bit depth is not 8 or 16 bit");
+
+   // Make sure the color mode is RGB.
+   // Valid options are:
+   //   0: Bitmap
+   //   1: Grayscale
+   //   2: Indexed color
+   //   3: RGB color
+   //   4: CMYK color
+   //   7: Multichannel
+   //   8: Duotone
+   //   9: Lab color
+   if (stbi__get16be(s) != 3)
+      return stbi__errpuc("wrong color format", "PSD is not in RGB color format");
+
+   // Skip the Mode Data.  (It's the palette for indexed color; other info for other modes.)
+   stbi__skip(s,stbi__get32be(s) );
+
+   // Skip the image resources.  (resolution, pen tool paths, etc)
+   stbi__skip(s, stbi__get32be(s) );
+
+   // Skip the reserved data.
+   stbi__skip(s, stbi__get32be(s) );
+
+   // Find out if the data is compressed.
+   // Known values:
+   //   0: no compression
+   //   1: RLE compressed
+   compression = stbi__get16be(s);
+   if (compression > 1)
+      return stbi__errpuc("bad compression", "PSD has an unknown compression format");
+
+   // Check size
+   if (!stbi__mad3sizes_valid(4, w, h, 0))
+      return stbi__errpuc("too large", "Corrupt PSD");
+
+   // Create the destination image.
+
+   if (!compression && bitdepth == 16 && bpc == 16) {
+      out = (stbi_uc *) stbi__malloc_mad3(8, w, h, 0);
+      ri->bits_per_channel = 16;
+   } else
+      out = (stbi_uc *) stbi__malloc(4 * w*h);
+
+   if (!out) return stbi__errpuc("outofmem", "Out of memory");
+   pixelCount = w*h;
+
+   // Initialize the data to zero.
+   //memset( out, 0, pixelCount * 4 );
+
+   // Finally, the image data.
+   if (compression) {
+      // RLE as used by .PSD and .TIFF
+      // Loop until you get the number of unpacked bytes you are expecting:
+      //     Read the next source byte into n.
+      //     If n is between 0 and 127 inclusive, copy the next n+1 bytes literally.
+      //     Else if n is between -127 and -1 inclusive, copy the next byte -n+1 times.
+      //     Else if n is 128, noop.
+      // Endloop
+
+      // The RLE-compressed data is preceeded by a 2-byte data count for each row in the data,
+      // which we're going to just skip.
+      stbi__skip(s, h * channelCount * 2 );
+
+      // Read the RLE data by channel.
+      for (channel = 0; channel < 4; channel++) {
+         stbi_uc *p;
+
+         p = out+channel;
+         if (channel >= channelCount) {
+            // Fill this channel with default data.
+            for (i = 0; i < pixelCount; i++, p += 4)
+               *p = (channel == 3 ? 255 : 0);
+         } else {
+            // Read the RLE data.
+            if (!stbi__psd_decode_rle(s, p, pixelCount)) {
+               STBI_FREE(out);
+               return stbi__errpuc("corrupt", "bad RLE data");
+            }
+         }
+      }
+
+   } else {
+      // We're at the raw image data.  It's each channel in order (Red, Green, Blue, Alpha, ...)
+      // where each channel consists of an 8-bit (or 16-bit) value for each pixel in the image.
+
+      // Read the data by channel.
+      for (channel = 0; channel < 4; channel++) {
+         if (channel >= channelCount) {
+            // Fill this channel with default data.
+            if (bitdepth == 16 && bpc == 16) {
+               stbi__uint16 *q = ((stbi__uint16 *) out) + channel;
+               stbi_uc val = channel == 3 ? 65535 : 0;
+               for (i = 0; i < pixelCount; i++, q += 4)
+                  *q = val;
+            } else {
+               stbi_uc *p = out+channel;
+               stbi_uc val = channel == 3 ? 255 : 0;
+               for (i = 0; i < pixelCount; i++, p += 4)
+                  *p = val;
+            }
+         } else {
+            if (ri->bits_per_channel == 16) {    // output bpc
+               stbi__uint16 *q = ((stbi__uint16 *) out) + channel;
+               for (i = 0; i < pixelCount; i++, q += 4)
+                  *q = (stbi__uint16) stbi__get16be(s);
+            } else {
+               stbi_uc *p = out+channel;
+               if (bitdepth == 16) {  // input bpc
+                  for (i = 0; i < pixelCount; i++, p += 4)
+                     *p = (stbi_uc) (stbi__get16be(s) >> 8);
+               } else {
+                  for (i = 0; i < pixelCount; i++, p += 4)
+                     *p = stbi__get8(s);
+               }
+            }
+         }
+      }
+   }
+
+   // remove weird white matte from PSD
+   if (channelCount >= 4) {
+      if (ri->bits_per_channel == 16) {
+         for (i=0; i < w*h; ++i) {
+            stbi__uint16 *pixel = (stbi__uint16 *) out + 4*i;
+            if (pixel[3] != 0 && pixel[3] != 65535) {
+               float a = pixel[3] / 65535.0f;
+               float ra = 1.0f / a;
+               float inv_a = 65535.0f * (1 - ra);
+               pixel[0] = (stbi__uint16) (pixel[0]*ra + inv_a);
+               pixel[1] = (stbi__uint16) (pixel[1]*ra + inv_a);
+               pixel[2] = (stbi__uint16) (pixel[2]*ra + inv_a);
+            }
+         }
+      } else {
+         for (i=0; i < w*h; ++i) {
+            unsigned char *pixel = out + 4*i;
+            if (pixel[3] != 0 && pixel[3] != 255) {
+               float a = pixel[3] / 255.0f;
+               float ra = 1.0f / a;
+               float inv_a = 255.0f * (1 - ra);
+               pixel[0] = (unsigned char) (pixel[0]*ra + inv_a);
+               pixel[1] = (unsigned char) (pixel[1]*ra + inv_a);
+               pixel[2] = (unsigned char) (pixel[2]*ra + inv_a);
+            }
+         }
+      }
+   }
+
+   // convert to desired output format
+   if (req_comp && req_comp != 4) {
+      if (ri->bits_per_channel == 16)
+         out = (stbi_uc *) stbi__convert_format16((stbi__uint16 *) out, 4, req_comp, w, h);
+      else
+         out = stbi__convert_format(out, 4, req_comp, w, h);
+      if (out == NULL) return out; // stbi__convert_format frees input on failure
+   }
+
+   if (comp) *comp = 4;
+   *y = h;
+   *x = w;
+
+   return out;
+}
+#endif
+
+// *************************************************************************************************
+// Softimage PIC loader
+// by Tom Seddon
+//
+// See http://softimage.wiki.softimage.com/index.php/INFO:_PIC_file_format
+// See http://ozviz.wasp.uwa.edu.au/~pbourke/dataformats/softimagepic/
+
+#ifndef STBI_NO_PIC
+static int stbi__pic_is4(stbi__context *s,const char *str)
+{
+   int i;
+   for (i=0; i<4; ++i)
+      if (stbi__get8(s) != (stbi_uc)str[i])
+         return 0;
+
+   return 1;
+}
+
+static int stbi__pic_test_core(stbi__context *s)
+{
+   int i;
+
+   if (!stbi__pic_is4(s,"\x53\x80\xF6\x34"))
+      return 0;
+
+   for(i=0;i<84;++i)
+      stbi__get8(s);
+
+   if (!stbi__pic_is4(s,"PICT"))
+      return 0;
+
+   return 1;
+}
+
+typedef struct
+{
+   stbi_uc size,type,channel;
+} stbi__pic_packet;
+
+static stbi_uc *stbi__readval(stbi__context *s, int channel, stbi_uc *dest)
+{
+   int mask=0x80, i;
+
+   for (i=0; i<4; ++i, mask>>=1) {
+      if (channel & mask) {
+         if (stbi__at_eof(s)) return stbi__errpuc("bad file","PIC file too short");
+         dest[i]=stbi__get8(s);
+      }
+   }
+
+   return dest;
+}
+
+static void stbi__copyval(int channel,stbi_uc *dest,const stbi_uc *src)
+{
+   int mask=0x80,i;
+
+   for (i=0;i<4; ++i, mask>>=1)
+      if (channel&mask)
+         dest[i]=src[i];
+}
+
+static stbi_uc *stbi__pic_load_core(stbi__context *s,int width,int height,int *comp, stbi_uc *result)
+{
+   int act_comp=0,num_packets=0,y,chained;
+   stbi__pic_packet packets[10];
+
+   // this will (should...) cater for even some bizarre stuff like having data
+    // for the same channel in multiple packets.
+   do {
+      stbi__pic_packet *packet;
+
+      if (num_packets==sizeof(packets)/sizeof(packets[0]))
+         return stbi__errpuc("bad format","too many packets");
+
+      packet = &packets[num_packets++];
+
+      chained = stbi__get8(s);
+      packet->size    = stbi__get8(s);
+      packet->type    = stbi__get8(s);
+      packet->channel = stbi__get8(s);
+
+      act_comp |= packet->channel;
+
+      if (stbi__at_eof(s))          return stbi__errpuc("bad file","file too short (reading packets)");
+      if (packet->size != 8)  return stbi__errpuc("bad format","packet isn't 8bpp");
+   } while (chained);
+
+   *comp = (act_comp & 0x10 ? 4 : 3); // has alpha channel?
+
+   for(y=0; y<height; ++y) {
+      int packet_idx;
+
+      for(packet_idx=0; packet_idx < num_packets; ++packet_idx) {
+         stbi__pic_packet *packet = &packets[packet_idx];
+         stbi_uc *dest = result+y*width*4;
+
+         switch (packet->type) {
+            default:
+               return stbi__errpuc("bad format","packet has bad compression type");
+
+            case 0: {//uncompressed
+               int x;
+
+               for(x=0;x<width;++x, dest+=4)
+                  if (!stbi__readval(s,packet->channel,dest))
+                     return 0;
+               break;
+            }
+
+            case 1://Pure RLE
+               {
+                  int left=width, i;
+
+                  while (left>0) {
+                     stbi_uc count,value[4];
+
+                     count=stbi__get8(s);
+                     if (stbi__at_eof(s))   return stbi__errpuc("bad file","file too short (pure read count)");
+
+                     if (count > left)
+                        count = (stbi_uc) left;
+
+                     if (!stbi__readval(s,packet->channel,value))  return 0;
+
+                     for(i=0; i<count; ++i,dest+=4)
+                        stbi__copyval(packet->channel,dest,value);
+                     left -= count;
+                  }
+               }
+               break;
+
+            case 2: {//Mixed RLE
+               int left=width;
+               while (left>0) {
+                  int count = stbi__get8(s), i;
+                  if (stbi__at_eof(s))  return stbi__errpuc("bad file","file too short (mixed read count)");
+
+                  if (count >= 128) { // Repeated
+                     stbi_uc value[4];
+
+                     if (count==128)
+                        count = stbi__get16be(s);
+                     else
+                        count -= 127;
+                     if (count > left)
+                        return stbi__errpuc("bad file","scanline overrun");
+
+                     if (!stbi__readval(s,packet->channel,value))
+                        return 0;
+
+                     for(i=0;i<count;++i, dest += 4)
+                        stbi__copyval(packet->channel,dest,value);
+                  } else { // Raw
+                     ++count;
+                     if (count>left) return stbi__errpuc("bad file","scanline overrun");
+
+                     for(i=0;i<count;++i, dest+=4)
+                        if (!stbi__readval(s,packet->channel,dest))
+                           return 0;
+                  }
+                  left-=count;
+               }
+               break;
+            }
+         }
+      }
+   }
+
+   return result;
+}
+
+static void *stbi__pic_load(stbi__context *s,int *px,int *py,int *comp,int req_comp, stbi__result_info *ri)
+{
+   stbi_uc *result;
+   int i, x,y;
+   STBI_NOTUSED(ri);
+
+   for (i=0; i<92; ++i)
+      stbi__get8(s);
+
+   x = stbi__get16be(s);
+   y = stbi__get16be(s);
+   if (stbi__at_eof(s))  return stbi__errpuc("bad file","file too short (pic header)");
+   if (!stbi__mad3sizes_valid(x, y, 4, 0)) return stbi__errpuc("too large", "PIC image too large to decode");
+
+   stbi__get32be(s); //skip `ratio'
+   stbi__get16be(s); //skip `fields'
+   stbi__get16be(s); //skip `pad'
+
+   // intermediate buffer is RGBA
+   result = (stbi_uc *) stbi__malloc_mad3(x, y, 4, 0);
+   memset(result, 0xff, x*y*4);
+
+   if (!stbi__pic_load_core(s,x,y,comp, result)) {
+      STBI_FREE(result);
+      result=0;
+   }
+   *px = x;
+   *py = y;
+   if (req_comp == 0) req_comp = *comp;
+   result=stbi__convert_format(result,4,req_comp,x,y);
+
+   return result;
+}
+
+static int stbi__pic_test(stbi__context *s)
+{
+   int r = stbi__pic_test_core(s);
+   stbi__rewind(s);
+   return r;
+}
+#endif
+
+// *************************************************************************************************
+// GIF loader -- public domain by Jean-Marc Lienher -- simplified/shrunk by stb
+
+#ifndef STBI_NO_GIF
+typedef struct
+{
+   stbi__int16 prefix;
+   stbi_uc first;
+   stbi_uc suffix;
+} stbi__gif_lzw;
+
+typedef struct
+{
+   int w,h;
+   stbi_uc *out, *old_out;             // output buffer (always 4 components)
+   int flags, bgindex, ratio, transparent, eflags, delay;
+   stbi_uc  pal[256][4];
+   stbi_uc lpal[256][4];
+   stbi__gif_lzw codes[4096];
+   stbi_uc *color_table;
+   int parse, step;
+   int lflags;
+   int start_x, start_y;
+   int max_x, max_y;
+   int cur_x, cur_y;
+   int line_size;
+} stbi__gif;
+
+static int stbi__gif_test_raw(stbi__context *s)
+{
+   int sz;
+   if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') return 0;
+   sz = stbi__get8(s);
+   if (sz != '9' && sz != '7') return 0;
+   if (stbi__get8(s) != 'a') return 0;
+   return 1;
+}
+
+static int stbi__gif_test(stbi__context *s)
+{
+   int r = stbi__gif_test_raw(s);
+   stbi__rewind(s);
+   return r;
+}
+
+static void stbi__gif_parse_colortable(stbi__context *s, stbi_uc pal[256][4], int num_entries, int transp)
+{
+   int i;
+   for (i=0; i < num_entries; ++i) {
+      pal[i][2] = stbi__get8(s);
+      pal[i][1] = stbi__get8(s);
+      pal[i][0] = stbi__get8(s);
+      pal[i][3] = transp == i ? 0 : 255;
+   }
+}
+
+static int stbi__gif_header(stbi__context *s, stbi__gif *g, int *comp, int is_info)
+{
+   stbi_uc version;
+   if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8')
+      return stbi__err("not GIF", "Corrupt GIF");
+
+   version = stbi__get8(s);
+   if (version != '7' && version != '9')    return stbi__err("not GIF", "Corrupt GIF");
+   if (stbi__get8(s) != 'a')                return stbi__err("not GIF", "Corrupt GIF");
+
+   stbi__g_failure_reason = "";
+   g->w = stbi__get16le(s);
+   g->h = stbi__get16le(s);
+   g->flags = stbi__get8(s);
+   g->bgindex = stbi__get8(s);
+   g->ratio = stbi__get8(s);
+   g->transparent = -1;
+
+   if (comp != 0) *comp = 4;  // can't actually tell whether it's 3 or 4 until we parse the comments
+
+   if (is_info) return 1;
+
+   if (g->flags & 0x80)
+      stbi__gif_parse_colortable(s,g->pal, 2 << (g->flags & 7), -1);
+
+   return 1;
+}
+
+static int stbi__gif_info_raw(stbi__context *s, int *x, int *y, int *comp)
+{
+   stbi__gif* g = (stbi__gif*) stbi__malloc(sizeof(stbi__gif));
+   if (!stbi__gif_header(s, g, comp, 1)) {
+      STBI_FREE(g);
+      stbi__rewind( s );
+      return 0;
+   }
+   if (x) *x = g->w;
+   if (y) *y = g->h;
+   STBI_FREE(g);
+   return 1;
+}
+
+static void stbi__out_gif_code(stbi__gif *g, stbi__uint16 code)
+{
+   stbi_uc *p, *c;
+
+   // recurse to decode the prefixes, since the linked-list is backwards,
+   // and working backwards through an interleaved image would be nasty
+   if (g->codes[code].prefix >= 0)
+      stbi__out_gif_code(g, g->codes[code].prefix);
+
+   if (g->cur_y >= g->max_y) return;
+
+   p = &g->out[g->cur_x + g->cur_y];
+   c = &g->color_table[g->codes[code].suffix * 4];
+
+   if (c[3] >= 128) {
+      p[0] = c[2];
+      p[1] = c[1];
+      p[2] = c[0];
+      p[3] = c[3];
+   }
+   g->cur_x += 4;
+
+   if (g->cur_x >= g->max_x) {
+      g->cur_x = g->start_x;
+      g->cur_y += g->step;
+
+      while (g->cur_y >= g->max_y && g->parse > 0) {
+         g->step = (1 << g->parse) * g->line_size;
+         g->cur_y = g->start_y + (g->step >> 1);
+         --g->parse;
+      }
+   }
+}
+
+static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g)
+{
+   stbi_uc lzw_cs;
+   stbi__int32 len, init_code;
+   stbi__uint32 first;
+   stbi__int32 codesize, codemask, avail, oldcode, bits, valid_bits, clear;
+   stbi__gif_lzw *p;
+
+   lzw_cs = stbi__get8(s);
+   if (lzw_cs > 12) return NULL;
+   clear = 1 << lzw_cs;
+   first = 1;
+   codesize = lzw_cs + 1;
+   codemask = (1 << codesize) - 1;
+   bits = 0;
+   valid_bits = 0;
+   for (init_code = 0; init_code < clear; init_code++) {
+      g->codes[init_code].prefix = -1;
+      g->codes[init_code].first = (stbi_uc) init_code;
+      g->codes[init_code].suffix = (stbi_uc) init_code;
+   }
+
+   // support no starting clear code
+   avail = clear+2;
+   oldcode = -1;
+
+   len = 0;
+   for(;;) {
+      if (valid_bits < codesize) {
+         if (len == 0) {
+            len = stbi__get8(s); // start new block
+            if (len == 0)
+               return g->out;
+         }
+         --len;
+         bits |= (stbi__int32) stbi__get8(s) << valid_bits;
+         valid_bits += 8;
+      } else {
+         stbi__int32 code = bits & codemask;
+         bits >>= codesize;
+         valid_bits -= codesize;
+         // @OPTIMIZE: is there some way we can accelerate the non-clear path?
+         if (code == clear) {  // clear code
+            codesize = lzw_cs + 1;
+            codemask = (1 << codesize) - 1;
+            avail = clear + 2;
+            oldcode = -1;
+            first = 0;
+         } else if (code == clear + 1) { // end of stream code
+            stbi__skip(s, len);
+            while ((len = stbi__get8(s)) > 0)
+               stbi__skip(s,len);
+            return g->out;
+         } else if (code <= avail) {
+            if (first) return stbi__errpuc("no clear code", "Corrupt GIF");
+
+            if (oldcode >= 0) {
+               p = &g->codes[avail++];
+               if (avail > 4096)        return stbi__errpuc("too many codes", "Corrupt GIF");
+               p->prefix = (stbi__int16) oldcode;
+               p->first = g->codes[oldcode].first;
+               p->suffix = (code == avail) ? p->first : g->codes[code].first;
+            } else if (code == avail)
+               return stbi__errpuc("illegal code in raster", "Corrupt GIF");
+
+            stbi__out_gif_code(g, (stbi__uint16) code);
+
+            if ((avail & codemask) == 0 && avail <= 0x0FFF) {
+               codesize++;
+               codemask = (1 << codesize) - 1;
+            }
+
+            oldcode = code;
+         } else {
+            return stbi__errpuc("illegal code in raster", "Corrupt GIF");
+         }
+      }
+   }
+}
+
+static void stbi__fill_gif_background(stbi__gif *g, int x0, int y0, int x1, int y1)
+{
+   int x, y;
+   stbi_uc *c = g->pal[g->bgindex];
+   for (y = y0; y < y1; y += 4 * g->w) {
+      for (x = x0; x < x1; x += 4) {
+         stbi_uc *p  = &g->out[y + x];
+         p[0] = c[2];
+         p[1] = c[1];
+         p[2] = c[0];
+         p[3] = 0;
+      }
+   }
+}
+
+// this function is designed to support animated gifs, although stb_image doesn't support it
+static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, int req_comp)
+{
+   int i;
+   stbi_uc *prev_out = 0;
+
+   if (g->out == 0 && !stbi__gif_header(s, g, comp,0))
+      return 0; // stbi__g_failure_reason set by stbi__gif_header
+
+   if (!stbi__mad3sizes_valid(g->w, g->h, 4, 0))
+      return stbi__errpuc("too large", "GIF too large");
+
+   prev_out = g->out;
+   g->out = (stbi_uc *) stbi__malloc_mad3(4, g->w, g->h, 0);
+   if (g->out == 0) return stbi__errpuc("outofmem", "Out of memory");
+
+   switch ((g->eflags & 0x1C) >> 2) {
+      case 0: // unspecified (also always used on 1st frame)
+         stbi__fill_gif_background(g, 0, 0, 4 * g->w, 4 * g->w * g->h);
+         break;
+      case 1: // do not dispose
+         if (prev_out) memcpy(g->out, prev_out, 4 * g->w * g->h);
+         g->old_out = prev_out;
+         break;
+      case 2: // dispose to background
+         if (prev_out) memcpy(g->out, prev_out, 4 * g->w * g->h);
+         stbi__fill_gif_background(g, g->start_x, g->start_y, g->max_x, g->max_y);
+         break;
+      case 3: // dispose to previous
+         if (g->old_out) {
+            for (i = g->start_y; i < g->max_y; i += 4 * g->w)
+               memcpy(&g->out[i + g->start_x], &g->old_out[i + g->start_x], g->max_x - g->start_x);
+         }
+         break;
+   }
+
+   for (;;) {
+      switch (stbi__get8(s)) {
+         case 0x2C: /* Image Descriptor */
+         {
+            int prev_trans = -1;
+            stbi__int32 x, y, w, h;
+            stbi_uc *o;
+
+            x = stbi__get16le(s);
+            y = stbi__get16le(s);
+            w = stbi__get16le(s);
+            h = stbi__get16le(s);
+            if (((x + w) > (g->w)) || ((y + h) > (g->h)))
+               return stbi__errpuc("bad Image Descriptor", "Corrupt GIF");
+
+            g->line_size = g->w * 4;
+            g->start_x = x * 4;
+            g->start_y = y * g->line_size;
+            g->max_x   = g->start_x + w * 4;
+            g->max_y   = g->start_y + h * g->line_size;
+            g->cur_x   = g->start_x;
+            g->cur_y   = g->start_y;
+
+            g->lflags = stbi__get8(s);
+
+            if (g->lflags & 0x40) {
+               g->step = 8 * g->line_size; // first interlaced spacing
+               g->parse = 3;
+            } else {
+               g->step = g->line_size;
+               g->parse = 0;
+            }
+
+            if (g->lflags & 0x80) {
+               stbi__gif_parse_colortable(s,g->lpal, 2 << (g->lflags & 7), g->eflags & 0x01 ? g->transparent : -1);
+               g->color_table = (stbi_uc *) g->lpal;
+            } else if (g->flags & 0x80) {
+               if (g->transparent >= 0 && (g->eflags & 0x01)) {
+                  prev_trans = g->pal[g->transparent][3];
+                  g->pal[g->transparent][3] = 0;
+               }
+               g->color_table = (stbi_uc *) g->pal;
+            } else
+               return stbi__errpuc("missing color table", "Corrupt GIF");
+
+            o = stbi__process_gif_raster(s, g);
+            if (o == NULL) return NULL;
+
+            if (prev_trans != -1)
+               g->pal[g->transparent][3] = (stbi_uc) prev_trans;
+
+            return o;
+         }
+
+         case 0x21: // Comment Extension.
+         {
+            int len;
+            if (stbi__get8(s) == 0xF9) { // Graphic Control Extension.
+               len = stbi__get8(s);
+               if (len == 4) {
+                  g->eflags = stbi__get8(s);
+                  g->delay = stbi__get16le(s);
+                  g->transparent = stbi__get8(s);
+               } else {
+                  stbi__skip(s, len);
+                  break;
+               }
+            }
+            while ((len = stbi__get8(s)) != 0)
+               stbi__skip(s, len);
+            break;
+         }
+
+         case 0x3B: // gif stream termination code
+            return (stbi_uc *) s; // using '1' causes warning on some compilers
+
+         default:
+            return stbi__errpuc("unknown code", "Corrupt GIF");
+      }
+   }
+
+   STBI_NOTUSED(req_comp);
+}
+
+static void *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
+{
+   stbi_uc *u = 0;
+   stbi__gif* g = (stbi__gif*) stbi__malloc(sizeof(stbi__gif));
+   memset(g, 0, sizeof(*g));
+   STBI_NOTUSED(ri);
+
+   u = stbi__gif_load_next(s, g, comp, req_comp);
+   if (u == (stbi_uc *) s) u = 0;  // end of animated gif marker
+   if (u) {
+      *x = g->w;
+      *y = g->h;
+      if (req_comp && req_comp != 4)
+         u = stbi__convert_format(u, 4, req_comp, g->w, g->h);
+   }
+   else if (g->out)
+      STBI_FREE(g->out);
+   STBI_FREE(g);
+   return u;
+}
+
+static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp)
+{
+   return stbi__gif_info_raw(s,x,y,comp);
+}
+#endif
+
+// *************************************************************************************************
+// Radiance RGBE HDR loader
+// originally by Nicolas Schulz
+#ifndef STBI_NO_HDR
+static int stbi__hdr_test_core(stbi__context *s, const char *signature)
+{
+   int i;
+   for (i=0; signature[i]; ++i)
+      if (stbi__get8(s) != signature[i])
+          return 0;
+   stbi__rewind(s);
+   return 1;
+}
+
+static int stbi__hdr_test(stbi__context* s)
+{
+   int r = stbi__hdr_test_core(s, "#?RADIANCE\n");
+   stbi__rewind(s);
+   if(!r) {
+       r = stbi__hdr_test_core(s, "#?RGBE\n");
+       stbi__rewind(s);
+   }
+   return r;
+}
+
+#define STBI__HDR_BUFLEN  1024
+static char *stbi__hdr_gettoken(stbi__context *z, char *buffer)
+{
+   int len=0;
+   char c = '\0';
+
+   c = (char) stbi__get8(z);
+
+   while (!stbi__at_eof(z) && c != '\n') {
+      buffer[len++] = c;
+      if (len == STBI__HDR_BUFLEN-1) {
+         // flush to end of line
+         while (!stbi__at_eof(z) && stbi__get8(z) != '\n')
+            ;
+         break;
+      }
+      c = (char) stbi__get8(z);
+   }
+
+   buffer[len] = 0;
+   return buffer;
+}
+
+static void stbi__hdr_convert(float *output, stbi_uc *input, int req_comp)
+{
+   if ( input[3] != 0 ) {
+      float f1;
+      // Exponent
+      f1 = (float) ldexp(1.0f, input[3] - (int)(128 + 8));
+      if (req_comp <= 2)
+         output[0] = (input[0] + input[1] + input[2]) * f1 / 3;
+      else {
+         output[0] = input[0] * f1;
+         output[1] = input[1] * f1;
+         output[2] = input[2] * f1;
+      }
+      if (req_comp == 2) output[1] = 1;
+      if (req_comp == 4) output[3] = 1;
+   } else {
+      switch (req_comp) {
+         case 4: output[3] = 1; /* fallthrough */
+         case 3: output[0] = output[1] = output[2] = 0;
+                 break;
+         case 2: output[1] = 1; /* fallthrough */
+         case 1: output[0] = 0;
+                 break;
+      }
+   }
+}
+
+static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
+{
+   char buffer[STBI__HDR_BUFLEN];
+   char *token;
+   int valid = 0;
+   int width, height;
+   stbi_uc *scanline;
+   float *hdr_data;
+   int len;
+   unsigned char count, value;
+   int i, j, k, c1,c2, z;
+   const char *headerToken;
+   STBI_NOTUSED(ri);
+
+   // Check identifier
+   headerToken = stbi__hdr_gettoken(s,buffer);
+   if (strcmp(headerToken, "#?RADIANCE") != 0 && strcmp(headerToken, "#?RGBE") != 0)
+      return stbi__errpf("not HDR", "Corrupt HDR image");
+
+   // Parse header
+   for(;;) {
+      token = stbi__hdr_gettoken(s,buffer);
+      if (token[0] == 0) break;
+      if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) valid = 1;
+   }
+
+   if (!valid)    return stbi__errpf("unsupported format", "Unsupported HDR format");
+
+   // Parse width and height
+   // can't use sscanf() if we're not using stdio!
+   token = stbi__hdr_gettoken(s,buffer);
+   if (strncmp(token, "-Y ", 3))  return stbi__errpf("unsupported data layout", "Unsupported HDR format");
+   token += 3;
+   height = (int) strtol(token, &token, 10);
+   while (*token == ' ') ++token;
+   if (strncmp(token, "+X ", 3))  return stbi__errpf("unsupported data layout", "Unsupported HDR format");
+   token += 3;
+   width = (int) strtol(token, NULL, 10);
+
+   *x = width;
+   *y = height;
+
+   if (comp) *comp = 3;
+   if (req_comp == 0) req_comp = 3;
+
+   if (!stbi__mad4sizes_valid(width, height, req_comp, sizeof(float), 0))
+      return stbi__errpf("too large", "HDR image is too large");
+
+   // Read data
+   hdr_data = (float *) stbi__malloc_mad4(width, height, req_comp, sizeof(float), 0);
+   if (!hdr_data)
+      return stbi__errpf("outofmem", "Out of memory");
+
+   // Load image data
+   // image data is stored as some number of sca
+   if ( width < 8 || width >= 32768) {
+      // Read flat data
+      for (j=0; j < height; ++j) {
+         for (i=0; i < width; ++i) {
+            stbi_uc rgbe[4];
+           main_decode_loop:
+            stbi__getn(s, rgbe, 4);
+            stbi__hdr_convert(hdr_data + j * width * req_comp + i * req_comp, rgbe, req_comp);
+         }
+      }
+   } else {
+      // Read RLE-encoded data
+      scanline = NULL;
+
+      for (j = 0; j < height; ++j) {
+         c1 = stbi__get8(s);
+         c2 = stbi__get8(s);
+         len = stbi__get8(s);
+         if (c1 != 2 || c2 != 2 || (len & 0x80)) {
+            // not run-length encoded, so we have to actually use THIS data as a decoded
+            // pixel (note this can't be a valid pixel--one of RGB must be >= 128)
+            stbi_uc rgbe[4];
+            rgbe[0] = (stbi_uc) c1;
+            rgbe[1] = (stbi_uc) c2;
+            rgbe[2] = (stbi_uc) len;
+            rgbe[3] = (stbi_uc) stbi__get8(s);
+            stbi__hdr_convert(hdr_data, rgbe, req_comp);
+            i = 1;
+            j = 0;
+            STBI_FREE(scanline);
+            goto main_decode_loop; // yes, this makes no sense
+         }
+         len <<= 8;
+         len |= stbi__get8(s);
+         if (len != width) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("invalid decoded scanline length", "corrupt HDR"); }
+         if (scanline == NULL) {
+            scanline = (stbi_uc *) stbi__malloc_mad2(width, 4, 0);
+            if (!scanline) {
+               STBI_FREE(hdr_data);
+               return stbi__errpf("outofmem", "Out of memory");
+            }
+         }
+
+         for (k = 0; k < 4; ++k) {
+            int nleft;
+            i = 0;
+            while ((nleft = width - i) > 0) {
+               count = stbi__get8(s);
+               if (count > 128) {
+                  // Run
+                  value = stbi__get8(s);
+                  count -= 128;
+                  if (count > nleft) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("corrupt", "bad RLE data in HDR"); }
+                  for (z = 0; z < count; ++z)
+                     scanline[i++ * 4 + k] = value;
+               } else {
+                  // Dump
+                  if (count > nleft) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("corrupt", "bad RLE data in HDR"); }
+                  for (z = 0; z < count; ++z)
+                     scanline[i++ * 4 + k] = stbi__get8(s);
+               }
+            }
+         }
+         for (i=0; i < width; ++i)
+            stbi__hdr_convert(hdr_data+(j*width + i)*req_comp, scanline + i*4, req_comp);
+      }
+      if (scanline)
+         STBI_FREE(scanline);
+   }
+
+   return hdr_data;
+}
+
+static int stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp)
+{
+   char buffer[STBI__HDR_BUFLEN];
+   char *token;
+   int valid = 0;
+
+   if (stbi__hdr_test(s) == 0) {
+       stbi__rewind( s );
+       return 0;
+   }
+
+   for(;;) {
+      token = stbi__hdr_gettoken(s,buffer);
+      if (token[0] == 0) break;
+      if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) valid = 1;
+   }
+
+   if (!valid) {
+       stbi__rewind( s );
+       return 0;
+   }
+   token = stbi__hdr_gettoken(s,buffer);
+   if (strncmp(token, "-Y ", 3)) {
+       stbi__rewind( s );
+       return 0;
+   }
+   token += 3;
+   *y = (int) strtol(token, &token, 10);
+   while (*token == ' ') ++token;
+   if (strncmp(token, "+X ", 3)) {
+       stbi__rewind( s );
+       return 0;
+   }
+   token += 3;
+   *x = (int) strtol(token, NULL, 10);
+   *comp = 3;
+   return 1;
+}
+#endif // STBI_NO_HDR
+
+#ifndef STBI_NO_BMP
+static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp)
+{
+   void *p;
+   stbi__bmp_data info;
+
+   info.all_a = 255;
+   p = stbi__bmp_parse_header(s, &info);
+   stbi__rewind( s );
+   if (p == NULL)
+      return 0;
+   *x = s->img_x;
+   *y = s->img_y;
+   *comp = info.ma ? 4 : 3;
+   return 1;
+}
+#endif
+
+#ifndef STBI_NO_PSD
+static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp)
+{
+   int channelCount;
+   if (stbi__get32be(s) != 0x38425053) {
+       stbi__rewind( s );
+       return 0;
+   }
+   if (stbi__get16be(s) != 1) {
+       stbi__rewind( s );
+       return 0;
+   }
+   stbi__skip(s, 6);
+   channelCount = stbi__get16be(s);
+   if (channelCount < 0 || channelCount > 16) {
+       stbi__rewind( s );
+       return 0;
+   }
+   *y = stbi__get32be(s);
+   *x = stbi__get32be(s);
+   if (stbi__get16be(s) != 8) {
+       stbi__rewind( s );
+       return 0;
+   }
+   if (stbi__get16be(s) != 3) {
+       stbi__rewind( s );
+       return 0;
+   }
+   *comp = 4;
+   return 1;
+}
+#endif
+
+#ifndef STBI_NO_PIC
+static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp)
+{
+   int act_comp=0,num_packets=0,chained;
+   stbi__pic_packet packets[10];
+
+   if (!stbi__pic_is4(s,"\x53\x80\xF6\x34")) {
+      stbi__rewind(s);
+      return 0;
+   }
+
+   stbi__skip(s, 88);
+
+   *x = stbi__get16be(s);
+   *y = stbi__get16be(s);
+   if (stbi__at_eof(s)) {
+      stbi__rewind( s);
+      return 0;
+   }
+   if ( (*x) != 0 && (1 << 28) / (*x) < (*y)) {
+      stbi__rewind( s );
+      return 0;
+   }
+
+   stbi__skip(s, 8);
+
+   do {
+      stbi__pic_packet *packet;
+
+      if (num_packets==sizeof(packets)/sizeof(packets[0]))
+         return 0;
+
+      packet = &packets[num_packets++];
+      chained = stbi__get8(s);
+      packet->size    = stbi__get8(s);
+      packet->type    = stbi__get8(s);
+      packet->channel = stbi__get8(s);
+      act_comp |= packet->channel;
+
+      if (stbi__at_eof(s)) {
+          stbi__rewind( s );
+          return 0;
+      }
+      if (packet->size != 8) {
+          stbi__rewind( s );
+          return 0;
+      }
+   } while (chained);
+
+   *comp = (act_comp & 0x10 ? 4 : 3);
+
+   return 1;
+}
+#endif
+
+// *************************************************************************************************
+// Portable Gray Map and Portable Pixel Map loader
+// by Ken Miller
+//
+// PGM: http://netpbm.sourceforge.net/doc/pgm.html
+// PPM: http://netpbm.sourceforge.net/doc/ppm.html
+//
+// Known limitations:
+//    Does not support comments in the header section
+//    Does not support ASCII image data (formats P2 and P3)
+//    Does not support 16-bit-per-channel
+
+#ifndef STBI_NO_PNM
+
+static int      stbi__pnm_test(stbi__context *s)
+{
+   char p, t;
+   p = (char) stbi__get8(s);
+   t = (char) stbi__get8(s);
+   if (p != 'P' || (t != '5' && t != '6')) {
+       stbi__rewind( s );
+       return 0;
+   }
+   return 1;
+}
+
+static void *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
+{
+   stbi_uc *out;
+   STBI_NOTUSED(ri);
+
+   if (!stbi__pnm_info(s, (int *)&s->img_x, (int *)&s->img_y, (int *)&s->img_n))
+      return 0;
+
+   *x = s->img_x;
+   *y = s->img_y;
+   *comp = s->img_n;
+
+   if (!stbi__mad3sizes_valid(s->img_n, s->img_x, s->img_y, 0))
+      return stbi__errpuc("too large", "PNM too large");
+
+   out = (stbi_uc *) stbi__malloc_mad3(s->img_n, s->img_x, s->img_y, 0);
+   if (!out) return stbi__errpuc("outofmem", "Out of memory");
+   stbi__getn(s, out, s->img_n * s->img_x * s->img_y);
+
+   if (req_comp && req_comp != s->img_n) {
+      out = stbi__convert_format(out, s->img_n, req_comp, s->img_x, s->img_y);
+      if (out == NULL) return out; // stbi__convert_format frees input on failure
+   }
+   return out;
+}
+
+static int      stbi__pnm_isspace(char c)
+{
+   return c == ' ' || c == '\t' || c == '\n' || c == '\v' || c == '\f' || c == '\r';
+}
+
+static void     stbi__pnm_skip_whitespace(stbi__context *s, char *c)
+{
+   for (;;) {
+      while (!stbi__at_eof(s) && stbi__pnm_isspace(*c))
+         *c = (char) stbi__get8(s);
+
+      if (stbi__at_eof(s) || *c != '#')
+         break;
+
+      while (!stbi__at_eof(s) && *c != '\n' && *c != '\r' )
+         *c = (char) stbi__get8(s);
+   }
+}
+
+static int      stbi__pnm_isdigit(char c)
+{
+   return c >= '0' && c <= '9';
+}
+
+static int      stbi__pnm_getinteger(stbi__context *s, char *c)
+{
+   int value = 0;
+
+   while (!stbi__at_eof(s) && stbi__pnm_isdigit(*c)) {
+      value = value*10 + (*c - '0');
+      *c = (char) stbi__get8(s);
+   }
+
+   return value;
+}
+
+static int      stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp)
+{
+   int maxv;
+   char c, p, t;
+
+   stbi__rewind( s );
+
+   // Get identifier
+   p = (char) stbi__get8(s);
+   t = (char) stbi__get8(s);
+   if (p != 'P' || (t != '5' && t != '6')) {
+       stbi__rewind( s );
+       return 0;
+   }
+
+   *comp = (t == '6') ? 3 : 1;  // '5' is 1-component .pgm; '6' is 3-component .ppm
+
+   c = (char) stbi__get8(s);
+   stbi__pnm_skip_whitespace(s, &c);
+
+   *x = stbi__pnm_getinteger(s, &c); // read width
+   stbi__pnm_skip_whitespace(s, &c);
+
+   *y = stbi__pnm_getinteger(s, &c); // read height
+   stbi__pnm_skip_whitespace(s, &c);
+
+   maxv = stbi__pnm_getinteger(s, &c);  // read max value
+
+   if (maxv > 255)
+      return stbi__err("max value > 255", "PPM image not 8-bit");
+   else
+      return 1;
+}
+#endif
+
+static int stbi__info_main(stbi__context *s, int *x, int *y, int *comp)
+{
+   #ifndef STBI_NO_JPEG
+   if (stbi__jpeg_info(s, x, y, comp)) return 1;
+   #endif
+
+   #ifndef STBI_NO_PNG
+   if (stbi__png_info(s, x, y, comp))  return 1;
+   #endif
+
+   #ifndef STBI_NO_GIF
+   if (stbi__gif_info(s, x, y, comp))  return 1;
+   #endif
+
+   #ifndef STBI_NO_BMP
+   if (stbi__bmp_info(s, x, y, comp))  return 1;
+   #endif
+
+   #ifndef STBI_NO_PSD
+   if (stbi__psd_info(s, x, y, comp))  return 1;
+   #endif
+
+   #ifndef STBI_NO_PIC
+   if (stbi__pic_info(s, x, y, comp))  return 1;
+   #endif
+
+   #ifndef STBI_NO_PNM
+   if (stbi__pnm_info(s, x, y, comp))  return 1;
+   #endif
+
+   #ifndef STBI_NO_HDR
+   if (stbi__hdr_info(s, x, y, comp))  return 1;
+   #endif
+
+   // test tga last because it's a crappy test!
+   #ifndef STBI_NO_TGA
+   if (stbi__tga_info(s, x, y, comp))
+       return 1;
+   #endif
+   return stbi__err("unknown image type", "Image not of any known type, or corrupt");
+}
+
+#ifndef STBI_NO_STDIO
+STBIDEF int stbi_info(char const *filename, int *x, int *y, int *comp)
+{
+    FILE *f = stbi__fopen(filename, "rb");
+    int result;
+    if (!f) return stbi__err("can't fopen", "Unable to open file");
+    result = stbi_info_from_file(f, x, y, comp);
+    fclose(f);
+    return result;
+}
+
+STBIDEF int stbi_info_from_file(FILE *f, int *x, int *y, int *comp)
+{
+   int r;
+   stbi__context s;
+   long pos = ftell(f);
+   stbi__start_file(&s, f);
+   r = stbi__info_main(&s,x,y,comp);
+   fseek(f,pos,SEEK_SET);
+   return r;
+}
+#endif // !STBI_NO_STDIO
+
+STBIDEF int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp)
+{
+   stbi__context s;
+   stbi__start_mem(&s,buffer,len);
+   return stbi__info_main(&s,x,y,comp);
+}
+
+STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *c, void *user, int *x, int *y, int *comp)
+{
+   stbi__context s;
+   stbi__start_callbacks(&s, (stbi_io_callbacks *) c, user);
+   return stbi__info_main(&s,x,y,comp);
+}
+
+#endif // STB_IMAGE_IMPLEMENTATION
+
+/*
+   revision history:
+      2.13  (2016-11-29) add 16-bit API, only supported for PNG right now
+      2.12  (2016-04-02) fix typo in 2.11 PSD fix that caused crashes
+      2.11  (2016-04-02) allocate large structures on the stack
+                         remove white matting for transparent PSD
+                         fix reported channel count for PNG & BMP
+                         re-enable SSE2 in non-gcc 64-bit
+                         support RGB-formatted JPEG
+                         read 16-bit PNGs (only as 8-bit)
+      2.10  (2016-01-22) avoid warning introduced in 2.09 by STBI_REALLOC_SIZED
+      2.09  (2016-01-16) allow comments in PNM files
+                         16-bit-per-pixel TGA (not bit-per-component)
+                         info() for TGA could break due to .hdr handling
+                         info() for BMP to shares code instead of sloppy parse
+                         can use STBI_REALLOC_SIZED if allocator doesn't support realloc
+                         code cleanup
+      2.08  (2015-09-13) fix to 2.07 cleanup, reading RGB PSD as RGBA
+      2.07  (2015-09-13) fix compiler warnings
+                         partial animated GIF support
+                         limited 16-bpc PSD support
+                         #ifdef unused functions
+                         bug with < 92 byte PIC,PNM,HDR,TGA
+      2.06  (2015-04-19) fix bug where PSD returns wrong '*comp' value
+      2.05  (2015-04-19) fix bug in progressive JPEG handling, fix warning
+      2.04  (2015-04-15) try to re-enable SIMD on MinGW 64-bit
+      2.03  (2015-04-12) extra corruption checking (mmozeiko)
+                         stbi_set_flip_vertically_on_load (nguillemot)
+                         fix NEON support; fix mingw support
+      2.02  (2015-01-19) fix incorrect assert, fix warning
+      2.01  (2015-01-17) fix various warnings; suppress SIMD on gcc 32-bit without -msse2
+      2.00b (2014-12-25) fix STBI_MALLOC in progressive JPEG
+      2.00  (2014-12-25) optimize JPG, including x86 SSE2 & NEON SIMD (ryg)
+                         progressive JPEG (stb)
+                         PGM/PPM support (Ken Miller)
+                         STBI_MALLOC,STBI_REALLOC,STBI_FREE
+                         GIF bugfix -- seemingly never worked
+                         STBI_NO_*, STBI_ONLY_*
+      1.48  (2014-12-14) fix incorrectly-named assert()
+      1.47  (2014-12-14) 1/2/4-bit PNG support, both direct and paletted (Omar Cornut & stb)
+                         optimize PNG (ryg)
+                         fix bug in interlaced PNG with user-specified channel count (stb)
+      1.46  (2014-08-26)
+              fix broken tRNS chunk (colorkey-style transparency) in non-paletted PNG
+      1.45  (2014-08-16)
+              fix MSVC-ARM internal compiler error by wrapping malloc
+      1.44  (2014-08-07)
+              various warning fixes from Ronny Chevalier
+      1.43  (2014-07-15)
+              fix MSVC-only compiler problem in code changed in 1.42
+      1.42  (2014-07-09)
+              don't define _CRT_SECURE_NO_WARNINGS (affects user code)
+              fixes to stbi__cleanup_jpeg path
+              added STBI_ASSERT to avoid requiring assert.h
+      1.41  (2014-06-25)
+              fix search&replace from 1.36 that messed up comments/error messages
+      1.40  (2014-06-22)
+              fix gcc struct-initialization warning
+      1.39  (2014-06-15)
+              fix to TGA optimization when req_comp != number of components in TGA;
+              fix to GIF loading because BMP wasn't rewinding (whoops, no GIFs in my test suite)
+              add support for BMP version 5 (more ignored fields)
+      1.38  (2014-06-06)
+              suppress MSVC warnings on integer casts truncating values
+              fix accidental rename of 'skip' field of I/O
+      1.37  (2014-06-04)
+              remove duplicate typedef
+      1.36  (2014-06-03)
+              convert to header file single-file library
+              if de-iphone isn't set, load iphone images color-swapped instead of returning NULL
+      1.35  (2014-05-27)
+              various warnings
+              fix broken STBI_SIMD path
+              fix bug where stbi_load_from_file no longer left file pointer in correct place
+              fix broken non-easy path for 32-bit BMP (possibly never used)
+              TGA optimization by Arseny Kapoulkine
+      1.34  (unknown)
+              use STBI_NOTUSED in stbi__resample_row_generic(), fix one more leak in tga failure case
+      1.33  (2011-07-14)
+              make stbi_is_hdr work in STBI_NO_HDR (as specified), minor compiler-friendly improvements
+      1.32  (2011-07-13)
+              support for "info" function for all supported filetypes (SpartanJ)
+      1.31  (2011-06-20)
+              a few more leak fixes, bug in PNG handling (SpartanJ)
+      1.30  (2011-06-11)
+              added ability to load files via callbacks to accomidate custom input streams (Ben Wenger)
+              removed deprecated format-specific test/load functions
+              removed support for installable file formats (stbi_loader) -- would have been broken for IO callbacks anyway
+              error cases in bmp and tga give messages and don't leak (Raymond Barbiero, grisha)
+              fix inefficiency in decoding 32-bit BMP (David Woo)
+      1.29  (2010-08-16)
+              various warning fixes from Aurelien Pocheville
+      1.28  (2010-08-01)
+              fix bug in GIF palette transparency (SpartanJ)
+      1.27  (2010-08-01)
+              cast-to-stbi_uc to fix warnings
+      1.26  (2010-07-24)
+              fix bug in file buffering for PNG reported by SpartanJ
+      1.25  (2010-07-17)
+              refix trans_data warning (Won Chun)
+      1.24  (2010-07-12)
+              perf improvements reading from files on platforms with lock-heavy fgetc()
+              minor perf improvements for jpeg
+              deprecated type-specific functions so we'll get feedback if they're needed
+              attempt to fix trans_data warning (Won Chun)
+      1.23    fixed bug in iPhone support
+      1.22  (2010-07-10)
+              removed image *writing* support
+              stbi_info support from Jetro Lauha
+              GIF support from Jean-Marc Lienher
+              iPhone PNG-extensions from James Brown
+              warning-fixes from Nicolas Schulz and Janez Zemva (i.stbi__err. Janez (U+017D)emva)
+      1.21    fix use of 'stbi_uc' in header (reported by jon blow)
+      1.20    added support for Softimage PIC, by Tom Seddon
+      1.19    bug in interlaced PNG corruption check (found by ryg)
+      1.18  (2008-08-02)
+              fix a threading bug (local mutable static)
+      1.17    support interlaced PNG
+      1.16    major bugfix - stbi__convert_format converted one too many pixels
+      1.15    initialize some fields for thread safety
+      1.14    fix threadsafe conversion bug
+              header-file-only version (#define STBI_HEADER_FILE_ONLY before including)
+      1.13    threadsafe
+      1.12    const qualifiers in the API
+      1.11    Support installable IDCT, colorspace conversion routines
+      1.10    Fixes for 64-bit (don't use "unsigned long")
+              optimized upsampling by Fabian "ryg" Giesen
+      1.09    Fix format-conversion for PSD code (bad global variables!)
+      1.08    Thatcher Ulrich's PSD code integrated by Nicolas Schulz
+      1.07    attempt to fix C++ warning/errors again
+      1.06    attempt to fix C++ warning/errors again
+      1.05    fix TGA loading to return correct *comp and use good luminance calc
+      1.04    default float alpha is 1, not 255; use 'void *' for stbi_image_free
+      1.03    bugfixes to STBI_NO_STDIO, STBI_NO_HDR
+      1.02    support for (subset of) HDR files, float interface for preferred access to them
+      1.01    fix bug: possible bug in handling right-side up bmps... not sure
+              fix bug: the stbi__bmp_load() and stbi__tga_load() functions didn't work at all
+      1.00    interface to zlib that skips zlib header
+      0.99    correct handling of alpha in palette
+      0.98    TGA loader by lonesock; dynamically add loaders (untested)
+      0.97    jpeg errors on too large a file; also catch another malloc failure
+      0.96    fix detection of invalid v value - particleman@mollyrocket forum
+      0.95    during header scan, seek to markers in case of padding
+      0.94    STBI_NO_STDIO to disable stdio usage; rename all #defines the same
+      0.93    handle jpegtran output; verbose errors
+      0.92    read 4,8,16,24,32-bit BMP files of several formats
+      0.91    output 24-bit Windows 3.0 BMP files
+      0.90    fix a few more warnings; bump version number to approach 1.0
+      0.61    bugfixes due to Marc LeBlanc, Christopher Lloyd
+      0.60    fix compiling as c++
+      0.59    fix warnings: merge Dave Moore's -Wall fixes
+      0.58    fix bug: zlib uncompressed mode len/nlen was wrong endian
+      0.57    fix bug: jpg last huffman symbol before marker was >9 bits but less than 16 available
+      0.56    fix bug: zlib uncompressed mode len vs. nlen
+      0.55    fix bug: restart_interval not initialized to 0
+      0.54    allow NULL for 'int *comp'
+      0.53    fix bug in png 3->4; speedup png decoding
+      0.52    png handles req_comp=3,4 directly; minor cleanup; jpeg comments
+      0.51    obey req_comp requests, 1-component jpegs return as 1-component,
+              on 'test' only check type, not whether we support this variant
+      0.50  (2006-11-19)
+              first released version
+*/
\ No newline at end of file
diff --git a/win64/include/stb_image_write.h b/win64/include/stb_image_write.h
new file mode 100644
index 0000000..018f9a8
--- /dev/null
+++ b/win64/include/stb_image_write.h
@@ -0,0 +1,1048 @@
+/* stb_image_write - v1.02 - public domain - http://nothings.org/stb/stb_image_write.h
+   writes out PNG/BMP/TGA images to C stdio - Sean Barrett 2010-2015
+                                     no warranty implied; use at your own risk
+
+   Before #including,
+
+       #define STB_IMAGE_WRITE_IMPLEMENTATION
+
+   in the file that you want to have the implementation.
+
+   Will probably not work correctly with strict-aliasing optimizations.
+
+ABOUT:
+
+   This header file is a library for writing images to C stdio. It could be
+   adapted to write to memory or a general streaming interface; let me know.
+
+   The PNG output is not optimal; it is 20-50% larger than the file
+   written by a decent optimizing implementation. This library is designed
+   for source code compactness and simplicity, not optimal image file size
+   or run-time performance.
+
+BUILDING:
+
+   You can #define STBIW_ASSERT(x) before the #include to avoid using assert.h.
+   You can #define STBIW_MALLOC(), STBIW_REALLOC(), and STBIW_FREE() to replace
+   malloc,realloc,free.
+   You can define STBIW_MEMMOVE() to replace memmove()
+
+USAGE:
+
+   There are four functions, one for each image file format:
+
+     int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes);
+     int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data);
+     int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data);
+     int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data);
+
+   There are also four equivalent functions that use an arbitrary write function. You are
+   expected to open/close your file-equivalent before and after calling these:
+
+     int stbi_write_png_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void  *data, int stride_in_bytes);
+     int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void  *data);
+     int stbi_write_tga_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void  *data);
+     int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const float *data);
+
+   where the callback is:
+      void stbi_write_func(void *context, void *data, int size);
+
+   You can define STBI_WRITE_NO_STDIO to disable the file variant of these
+   functions, so the library will not use stdio.h at all. However, this will
+   also disable HDR writing, because it requires stdio for formatted output.
+
+   Each function returns 0 on failure and non-0 on success.
+
+   The functions create an image file defined by the parameters. The image
+   is a rectangle of pixels stored from left-to-right, top-to-bottom.
+   Each pixel contains 'comp' channels of data stored interleaved with 8-bits
+   per channel, in the following order: 1=Y, 2=YA, 3=RGB, 4=RGBA. (Y is
+   monochrome color.) The rectangle is 'w' pixels wide and 'h' pixels tall.
+   The *data pointer points to the first byte of the top-left-most pixel.
+   For PNG, "stride_in_bytes" is the distance in bytes from the first byte of
+   a row of pixels to the first byte of the next row of pixels.
+
+   PNG creates output files with the same number of components as the input.
+   The BMP format expands Y to RGB in the file format and does not
+   output alpha.
+
+   PNG supports writing rectangles of data even when the bytes storing rows of
+   data are not consecutive in memory (e.g. sub-rectangles of a larger image),
+   by supplying the stride between the beginning of adjacent rows. The other
+   formats do not. (Thus you cannot write a native-format BMP through the BMP
+   writer, both because it is in BGR order and because it may have padding
+   at the end of the line.)
+
+   HDR expects linear float data. Since the format is always 32-bit rgb(e)
+   data, alpha (if provided) is discarded, and for monochrome data it is
+   replicated across all three channels.
+
+   TGA supports RLE or non-RLE compressed data. To use non-RLE-compressed
+   data, set the global variable 'stbi_write_tga_with_rle' to 0.
+
+CREDITS:
+
+   PNG/BMP/TGA
+      Sean Barrett
+   HDR
+      Baldur Karlsson
+   TGA monochrome:
+      Jean-Sebastien Guay
+   misc enhancements:
+      Tim Kelsey
+   TGA RLE
+      Alan Hickman
+   initial file IO callback implementation
+      Emmanuel Julien
+   bugfixes:
+      github:Chribba
+      Guillaume Chereau
+      github:jry2
+      github:romigrou
+      Sergio Gonzalez
+      Jonas Karlsson
+      Filip Wasil
+      Thatcher Ulrich
+      
+LICENSE
+
+This software is dual-licensed to the public domain and under the following
+license: you are granted a perpetual, irrevocable license to copy, modify,
+publish, and distribute this file as you see fit.
+
+*/
+
+#ifndef INCLUDE_STB_IMAGE_WRITE_H
+#define INCLUDE_STB_IMAGE_WRITE_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifdef STB_IMAGE_WRITE_STATIC
+#define STBIWDEF static
+#else
+#define STBIWDEF extern
+extern int stbi_write_tga_with_rle;
+#endif
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_png(char const *filename, int w, int h, int comp, const void  *data, int stride_in_bytes);
+STBIWDEF int stbi_write_bmp(char const *filename, int w, int h, int comp, const void  *data);
+STBIWDEF int stbi_write_tga(char const *filename, int w, int h, int comp, const void  *data);
+STBIWDEF int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data);
+#endif
+
+typedef void stbi_write_func(void *context, void *data, int size);
+
+STBIWDEF int stbi_write_png_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void  *data, int stride_in_bytes);
+STBIWDEF int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void  *data);
+STBIWDEF int stbi_write_tga_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void  *data);
+STBIWDEF int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const float *data);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif//INCLUDE_STB_IMAGE_WRITE_H
+
+#ifdef STB_IMAGE_WRITE_IMPLEMENTATION
+
+#ifdef _WIN32
+   #ifndef _CRT_SECURE_NO_WARNINGS
+   #define _CRT_SECURE_NO_WARNINGS
+   #endif
+   #ifndef _CRT_NONSTDC_NO_DEPRECATE
+   #define _CRT_NONSTDC_NO_DEPRECATE
+   #endif
+#endif
+
+#ifndef STBI_WRITE_NO_STDIO
+#include <stdio.h>
+#endif // STBI_WRITE_NO_STDIO
+
+#include <stdarg.h>
+#include <stdlib.h>
+#include <string.h>
+#include <math.h>
+
+#if defined(STBIW_MALLOC) && defined(STBIW_FREE) && (defined(STBIW_REALLOC) || defined(STBIW_REALLOC_SIZED))
+// ok
+#elif !defined(STBIW_MALLOC) && !defined(STBIW_FREE) && !defined(STBIW_REALLOC) && !defined(STBIW_REALLOC_SIZED)
+// ok
+#else
+#error "Must define all or none of STBIW_MALLOC, STBIW_FREE, and STBIW_REALLOC (or STBIW_REALLOC_SIZED)."
+#endif
+
+#ifndef STBIW_MALLOC
+#define STBIW_MALLOC(sz)        malloc(sz)
+#define STBIW_REALLOC(p,newsz)  realloc(p,newsz)
+#define STBIW_FREE(p)           free(p)
+#endif
+
+#ifndef STBIW_REALLOC_SIZED
+#define STBIW_REALLOC_SIZED(p,oldsz,newsz) STBIW_REALLOC(p,newsz)
+#endif
+
+
+#ifndef STBIW_MEMMOVE
+#define STBIW_MEMMOVE(a,b,sz) memmove(a,b,sz)
+#endif
+
+
+#ifndef STBIW_ASSERT
+#include <assert.h>
+#define STBIW_ASSERT(x) assert(x)
+#endif
+
+#define STBIW_UCHAR(x) (unsigned char) ((x) & 0xff)
+
+typedef struct
+{
+   stbi_write_func *func;
+   void *context;
+} stbi__write_context;
+
+// initialize a callback-based context
+static void stbi__start_write_callbacks(stbi__write_context *s, stbi_write_func *c, void *context)
+{
+   s->func    = c;
+   s->context = context;
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+
+static void stbi__stdio_write(void *context, void *data, int size)
+{
+   fwrite(data,1,size,(FILE*) context);
+}
+
+static int stbi__start_write_file(stbi__write_context *s, const char *filename)
+{
+   FILE *f = fopen(filename, "wb");
+   stbi__start_write_callbacks(s, stbi__stdio_write, (void *) f);
+   return f != NULL;
+}
+
+static void stbi__end_write_file(stbi__write_context *s)
+{
+   fclose((FILE *)s->context);
+}
+
+#endif // !STBI_WRITE_NO_STDIO
+
+typedef unsigned int stbiw_uint32;
+typedef int stb_image_write_test[sizeof(stbiw_uint32)==4 ? 1 : -1];
+
+#ifdef STB_IMAGE_WRITE_STATIC
+static int stbi_write_tga_with_rle = 1;
+#else
+int stbi_write_tga_with_rle = 1;
+#endif
+
+static void stbiw__writefv(stbi__write_context *s, const char *fmt, va_list v)
+{
+   while (*fmt) {
+      switch (*fmt++) {
+         case ' ': break;
+         case '1': { unsigned char x = STBIW_UCHAR(va_arg(v, int));
+                     s->func(s->context,&x,1);
+                     break; }
+         case '2': { int x = va_arg(v,int);
+                     unsigned char b[2];
+                     b[0] = STBIW_UCHAR(x);
+                     b[1] = STBIW_UCHAR(x>>8);
+                     s->func(s->context,b,2);
+                     break; }
+         case '4': { stbiw_uint32 x = va_arg(v,int);
+                     unsigned char b[4];
+                     b[0]=STBIW_UCHAR(x);
+                     b[1]=STBIW_UCHAR(x>>8);
+                     b[2]=STBIW_UCHAR(x>>16);
+                     b[3]=STBIW_UCHAR(x>>24);
+                     s->func(s->context,b,4);
+                     break; }
+         default:
+            STBIW_ASSERT(0);
+            return;
+      }
+   }
+}
+
+static void stbiw__writef(stbi__write_context *s, const char *fmt, ...)
+{
+   va_list v;
+   va_start(v, fmt);
+   stbiw__writefv(s, fmt, v);
+   va_end(v);
+}
+
+static void stbiw__write3(stbi__write_context *s, unsigned char a, unsigned char b, unsigned char c)
+{
+   unsigned char arr[3];
+   arr[0] = a, arr[1] = b, arr[2] = c;
+   s->func(s->context, arr, 3);
+}
+
+static void stbiw__write_pixel(stbi__write_context *s, int rgb_dir, int comp, int write_alpha, int expand_mono, unsigned char *d)
+{
+   unsigned char bg[3] = { 255, 0, 255}, px[3];
+   int k;
+
+   if (write_alpha < 0)
+      s->func(s->context, &d[comp - 1], 1);
+
+   switch (comp) {
+      case 1:
+         s->func(s->context,d,1);
+         break;
+      case 2:
+         if (expand_mono)
+            stbiw__write3(s, d[0], d[0], d[0]); // monochrome bmp
+         else
+            s->func(s->context, d, 1);  // monochrome TGA
+         break;
+      case 4:
+         if (!write_alpha) {
+            // composite against pink background
+            for (k = 0; k < 3; ++k)
+               px[k] = bg[k] + ((d[k] - bg[k]) * d[3]) / 255;
+            stbiw__write3(s, px[1 - rgb_dir], px[1], px[1 + rgb_dir]);
+            break;
+         }
+         /* FALLTHROUGH */
+      case 3:
+         stbiw__write3(s, d[1 - rgb_dir], d[1], d[1 + rgb_dir]);
+         break;
+   }
+   if (write_alpha > 0)
+      s->func(s->context, &d[comp - 1], 1);
+}
+
+static void stbiw__write_pixels(stbi__write_context *s, int rgb_dir, int vdir, int x, int y, int comp, void *data, int write_alpha, int scanline_pad, int expand_mono)
+{
+   stbiw_uint32 zero = 0;
+   int i,j, j_end;
+
+   if (y <= 0)
+      return;
+
+   if (vdir < 0)
+      j_end = -1, j = y-1;
+   else
+      j_end =  y, j = 0;
+
+   for (; j != j_end; j += vdir) {
+      for (i=0; i < x; ++i) {
+         unsigned char *d = (unsigned char *) data + (j*x+i)*comp;
+         stbiw__write_pixel(s, rgb_dir, comp, write_alpha, expand_mono, d);
+      }
+      s->func(s->context, &zero, scanline_pad);
+   }
+}
+
+static int stbiw__outfile(stbi__write_context *s, int rgb_dir, int vdir, int x, int y, int comp, int expand_mono, void *data, int alpha, int pad, const char *fmt, ...)
+{
+   if (y < 0 || x < 0) {
+      return 0;
+   } else {
+      va_list v;
+      va_start(v, fmt);
+      stbiw__writefv(s, fmt, v);
+      va_end(v);
+      stbiw__write_pixels(s,rgb_dir,vdir,x,y,comp,data,alpha,pad, expand_mono);
+      return 1;
+   }
+}
+
+static int stbi_write_bmp_core(stbi__write_context *s, int x, int y, int comp, const void *data)
+{
+   int pad = (-x*3) & 3;
+   return stbiw__outfile(s,-1,-1,x,y,comp,1,(void *) data,0,pad,
+           "11 4 22 4" "4 44 22 444444",
+           'B', 'M', 14+40+(x*3+pad)*y, 0,0, 14+40,  // file header
+            40, x,y, 1,24, 0,0,0,0,0,0);             // bitmap header
+}
+
+STBIWDEF int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data)
+{
+   stbi__write_context s;
+   stbi__start_write_callbacks(&s, func, context);
+   return stbi_write_bmp_core(&s, x, y, comp, data);
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_bmp(char const *filename, int x, int y, int comp, const void *data)
+{
+   stbi__write_context s;
+   if (stbi__start_write_file(&s,filename)) {
+      int r = stbi_write_bmp_core(&s, x, y, comp, data);
+      stbi__end_write_file(&s);
+      return r;
+   } else
+      return 0;
+}
+#endif //!STBI_WRITE_NO_STDIO
+
+static int stbi_write_tga_core(stbi__write_context *s, int x, int y, int comp, void *data)
+{
+   int has_alpha = (comp == 2 || comp == 4);
+   int colorbytes = has_alpha ? comp-1 : comp;
+   int format = colorbytes < 2 ? 3 : 2; // 3 color channels (RGB/RGBA) = 2, 1 color channel (Y/YA) = 3
+
+   if (y < 0 || x < 0)
+      return 0;
+
+   if (!stbi_write_tga_with_rle) {
+      return stbiw__outfile(s, -1, -1, x, y, comp, 0, (void *) data, has_alpha, 0,
+         "111 221 2222 11", 0, 0, format, 0, 0, 0, 0, 0, x, y, (colorbytes + has_alpha) * 8, has_alpha * 8);
+   } else {
+      int i,j,k;
+
+      stbiw__writef(s, "111 221 2222 11", 0,0,format+8, 0,0,0, 0,0,x,y, (colorbytes + has_alpha) * 8, has_alpha * 8);
+
+      for (j = y - 1; j >= 0; --j) {
+          unsigned char *row = (unsigned char *) data + j * x * comp;
+         int len;
+
+         for (i = 0; i < x; i += len) {
+            unsigned char *begin = row + i * comp;
+            int diff = 1;
+            len = 1;
+
+            if (i < x - 1) {
+               ++len;
+               diff = memcmp(begin, row + (i + 1) * comp, comp);
+               if (diff) {
+                  const unsigned char *prev = begin;
+                  for (k = i + 2; k < x && len < 128; ++k) {
+                     if (memcmp(prev, row + k * comp, comp)) {
+                        prev += comp;
+                        ++len;
+                     } else {
+                        --len;
+                        break;
+                     }
+                  }
+               } else {
+                  for (k = i + 2; k < x && len < 128; ++k) {
+                     if (!memcmp(begin, row + k * comp, comp)) {
+                        ++len;
+                     } else {
+                        break;
+                     }
+                  }
+               }
+            }
+
+            if (diff) {
+               unsigned char header = STBIW_UCHAR(len - 1);
+               s->func(s->context, &header, 1);
+               for (k = 0; k < len; ++k) {
+                  stbiw__write_pixel(s, -1, comp, has_alpha, 0, begin + k * comp);
+               }
+            } else {
+               unsigned char header = STBIW_UCHAR(len - 129);
+               s->func(s->context, &header, 1);
+               stbiw__write_pixel(s, -1, comp, has_alpha, 0, begin);
+            }
+         }
+      }
+   }
+   return 1;
+}
+
+int stbi_write_tga_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data)
+{
+   stbi__write_context s;
+   stbi__start_write_callbacks(&s, func, context);
+   return stbi_write_tga_core(&s, x, y, comp, (void *) data);
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+int stbi_write_tga(char const *filename, int x, int y, int comp, const void *data)
+{
+   stbi__write_context s;
+   if (stbi__start_write_file(&s,filename)) {
+      int r = stbi_write_tga_core(&s, x, y, comp, (void *) data);
+      stbi__end_write_file(&s);
+      return r;
+   } else
+      return 0;
+}
+#endif
+
+// *************************************************************************************************
+// Radiance RGBE HDR writer
+// by Baldur Karlsson
+#ifndef STBI_WRITE_NO_STDIO
+
+#define stbiw__max(a, b)  ((a) > (b) ? (a) : (b))
+
+void stbiw__linear_to_rgbe(unsigned char *rgbe, float *linear)
+{
+   int exponent;
+   float maxcomp = stbiw__max(linear[0], stbiw__max(linear[1], linear[2]));
+
+   if (maxcomp < 1e-32f) {
+      rgbe[0] = rgbe[1] = rgbe[2] = rgbe[3] = 0;
+   } else {
+      float normalize = (float) frexp(maxcomp, &exponent) * 256.0f/maxcomp;
+
+      rgbe[0] = (unsigned char)(linear[0] * normalize);
+      rgbe[1] = (unsigned char)(linear[1] * normalize);
+      rgbe[2] = (unsigned char)(linear[2] * normalize);
+      rgbe[3] = (unsigned char)(exponent + 128);
+   }
+}
+
+void stbiw__write_run_data(stbi__write_context *s, int length, unsigned char databyte)
+{
+   unsigned char lengthbyte = STBIW_UCHAR(length+128);
+   STBIW_ASSERT(length+128 <= 255);
+   s->func(s->context, &lengthbyte, 1);
+   s->func(s->context, &databyte, 1);
+}
+
+void stbiw__write_dump_data(stbi__write_context *s, int length, unsigned char *data)
+{
+   unsigned char lengthbyte = STBIW_UCHAR(length);
+   STBIW_ASSERT(length <= 128); // inconsistent with spec but consistent with official code
+   s->func(s->context, &lengthbyte, 1);
+   s->func(s->context, data, length);
+}
+
+void stbiw__write_hdr_scanline(stbi__write_context *s, int width, int ncomp, unsigned char *scratch, float *scanline)
+{
+   unsigned char scanlineheader[4] = { 2, 2, 0, 0 };
+   unsigned char rgbe[4];
+   float linear[3];
+   int x;
+
+   scanlineheader[2] = (width&0xff00)>>8;
+   scanlineheader[3] = (width&0x00ff);
+
+   /* skip RLE for images too small or large */
+   if (width < 8 || width >= 32768) {
+      for (x=0; x < width; x++) {
+         switch (ncomp) {
+            case 4: /* fallthrough */
+            case 3: linear[2] = scanline[x*ncomp + 2];
+                    linear[1] = scanline[x*ncomp + 1];
+                    linear[0] = scanline[x*ncomp + 0];
+                    break;
+            default:
+                    linear[0] = linear[1] = linear[2] = scanline[x*ncomp + 0];
+                    break;
+         }
+         stbiw__linear_to_rgbe(rgbe, linear);
+         s->func(s->context, rgbe, 4);
+      }
+   } else {
+      int c,r;
+      /* encode into scratch buffer */
+      for (x=0; x < width; x++) {
+         switch(ncomp) {
+            case 4: /* fallthrough */
+            case 3: linear[2] = scanline[x*ncomp + 2];
+                    linear[1] = scanline[x*ncomp + 1];
+                    linear[0] = scanline[x*ncomp + 0];
+                    break;
+            default:
+                    linear[0] = linear[1] = linear[2] = scanline[x*ncomp + 0];
+                    break;
+         }
+         stbiw__linear_to_rgbe(rgbe, linear);
+         scratch[x + width*0] = rgbe[0];
+         scratch[x + width*1] = rgbe[1];
+         scratch[x + width*2] = rgbe[2];
+         scratch[x + width*3] = rgbe[3];
+      }
+
+      s->func(s->context, scanlineheader, 4);
+
+      /* RLE each component separately */
+      for (c=0; c < 4; c++) {
+         unsigned char *comp = &scratch[width*c];
+
+         x = 0;
+         while (x < width) {
+            // find first run
+            r = x;
+            while (r+2 < width) {
+               if (comp[r] == comp[r+1] && comp[r] == comp[r+2])
+                  break;
+               ++r;
+            }
+            if (r+2 >= width)
+               r = width;
+            // dump up to first run
+            while (x < r) {
+               int len = r-x;
+               if (len > 128) len = 128;
+               stbiw__write_dump_data(s, len, &comp[x]);
+               x += len;
+            }
+            // if there's a run, output it
+            if (r+2 < width) { // same test as what we break out of in search loop, so only true if we break'd
+               // find next byte after run
+               while (r < width && comp[r] == comp[x])
+                  ++r;
+               // output run up to r
+               while (x < r) {
+                  int len = r-x;
+                  if (len > 127) len = 127;
+                  stbiw__write_run_data(s, len, comp[x]);
+                  x += len;
+               }
+            }
+         }
+      }
+   }
+}
+
+static int stbi_write_hdr_core(stbi__write_context *s, int x, int y, int comp, float *data)
+{
+   if (y <= 0 || x <= 0 || data == NULL)
+      return 0;
+   else {
+      // Each component is stored separately. Allocate scratch space for full output scanline.
+      unsigned char *scratch = (unsigned char *) STBIW_MALLOC(x*4);
+      int i, len;
+      char buffer[128];
+      char header[] = "#?RADIANCE\n# Written by stb_image_write.h\nFORMAT=32-bit_rle_rgbe\n";
+      s->func(s->context, header, sizeof(header)-1);
+
+      len = sprintf(buffer, "EXPOSURE=          1.0000000000000\n\n-Y %d +X %d\n", y, x);
+      s->func(s->context, buffer, len);
+
+      for(i=0; i < y; i++)
+         stbiw__write_hdr_scanline(s, x, comp, scratch, data + comp*i*x);
+      STBIW_FREE(scratch);
+      return 1;
+   }
+}
+
+int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const float *data)
+{
+   stbi__write_context s;
+   stbi__start_write_callbacks(&s, func, context);
+   return stbi_write_hdr_core(&s, x, y, comp, (float *) data);
+}
+
+int stbi_write_hdr(char const *filename, int x, int y, int comp, const float *data)
+{
+   stbi__write_context s;
+   if (stbi__start_write_file(&s,filename)) {
+      int r = stbi_write_hdr_core(&s, x, y, comp, (float *) data);
+      stbi__end_write_file(&s);
+      return r;
+   } else
+      return 0;
+}
+#endif // STBI_WRITE_NO_STDIO
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// PNG writer
+//
+
+// stretchy buffer; stbiw__sbpush() == vector<>::push_back() -- stbiw__sbcount() == vector<>::size()
+#define stbiw__sbraw(a) ((int *) (a) - 2)
+#define stbiw__sbm(a)   stbiw__sbraw(a)[0]
+#define stbiw__sbn(a)   stbiw__sbraw(a)[1]
+
+#define stbiw__sbneedgrow(a,n)  ((a)==0 || stbiw__sbn(a)+n >= stbiw__sbm(a))
+#define stbiw__sbmaybegrow(a,n) (stbiw__sbneedgrow(a,(n)) ? stbiw__sbgrow(a,n) : 0)
+#define stbiw__sbgrow(a,n)  stbiw__sbgrowf((void **) &(a), (n), sizeof(*(a)))
+
+#define stbiw__sbpush(a, v)      (stbiw__sbmaybegrow(a,1), (a)[stbiw__sbn(a)++] = (v))
+#define stbiw__sbcount(a)        ((a) ? stbiw__sbn(a) : 0)
+#define stbiw__sbfree(a)         ((a) ? STBIW_FREE(stbiw__sbraw(a)),0 : 0)
+
+static void *stbiw__sbgrowf(void **arr, int increment, int itemsize)
+{
+   int m = *arr ? 2*stbiw__sbm(*arr)+increment : increment+1;
+   void *p = STBIW_REALLOC_SIZED(*arr ? stbiw__sbraw(*arr) : 0, *arr ? (stbiw__sbm(*arr)*itemsize + sizeof(int)*2) : 0, itemsize * m + sizeof(int)*2);
+   STBIW_ASSERT(p);
+   if (p) {
+      if (!*arr) ((int *) p)[1] = 0;
+      *arr = (void *) ((int *) p + 2);
+      stbiw__sbm(*arr) = m;
+   }
+   return *arr;
+}
+
+static unsigned char *stbiw__zlib_flushf(unsigned char *data, unsigned int *bitbuffer, int *bitcount)
+{
+   while (*bitcount >= 8) {
+      stbiw__sbpush(data, STBIW_UCHAR(*bitbuffer));
+      *bitbuffer >>= 8;
+      *bitcount -= 8;
+   }
+   return data;
+}
+
+static int stbiw__zlib_bitrev(int code, int codebits)
+{
+   int res=0;
+   while (codebits--) {
+      res = (res << 1) | (code & 1);
+      code >>= 1;
+   }
+   return res;
+}
+
+static unsigned int stbiw__zlib_countm(unsigned char *a, unsigned char *b, int limit)
+{
+   int i;
+   for (i=0; i < limit && i < 258; ++i)
+      if (a[i] != b[i]) break;
+   return i;
+}
+
+static unsigned int stbiw__zhash(unsigned char *data)
+{
+   stbiw_uint32 hash = data[0] + (data[1] << 8) + (data[2] << 16);
+   hash ^= hash << 3;
+   hash += hash >> 5;
+   hash ^= hash << 4;
+   hash += hash >> 17;
+   hash ^= hash << 25;
+   hash += hash >> 6;
+   return hash;
+}
+
+#define stbiw__zlib_flush() (out = stbiw__zlib_flushf(out, &bitbuf, &bitcount))
+#define stbiw__zlib_add(code,codebits) \
+      (bitbuf |= (code) << bitcount, bitcount += (codebits), stbiw__zlib_flush())
+#define stbiw__zlib_huffa(b,c)  stbiw__zlib_add(stbiw__zlib_bitrev(b,c),c)
+// default huffman tables
+#define stbiw__zlib_huff1(n)  stbiw__zlib_huffa(0x30 + (n), 8)
+#define stbiw__zlib_huff2(n)  stbiw__zlib_huffa(0x190 + (n)-144, 9)
+#define stbiw__zlib_huff3(n)  stbiw__zlib_huffa(0 + (n)-256,7)
+#define stbiw__zlib_huff4(n)  stbiw__zlib_huffa(0xc0 + (n)-280,8)
+#define stbiw__zlib_huff(n)  ((n) <= 143 ? stbiw__zlib_huff1(n) : (n) <= 255 ? stbiw__zlib_huff2(n) : (n) <= 279 ? stbiw__zlib_huff3(n) : stbiw__zlib_huff4(n))
+#define stbiw__zlib_huffb(n) ((n) <= 143 ? stbiw__zlib_huff1(n) : stbiw__zlib_huff2(n))
+
+#define stbiw__ZHASH   16384
+
+unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_len, int quality)
+{
+   static unsigned short lengthc[] = { 3,4,5,6,7,8,9,10,11,13,15,17,19,23,27,31,35,43,51,59,67,83,99,115,131,163,195,227,258, 259 };
+   static unsigned char  lengtheb[]= { 0,0,0,0,0,0,0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4,  4,  5,  5,  5,  5,  0 };
+   static unsigned short distc[]   = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577, 32768 };
+   static unsigned char  disteb[]  = { 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13 };
+   unsigned int bitbuf=0;
+   int i,j, bitcount=0;
+   unsigned char *out = NULL;
+   unsigned char ***hash_table = (unsigned char***) STBIW_MALLOC(stbiw__ZHASH * sizeof(char**));
+   if (quality < 5) quality = 5;
+
+   stbiw__sbpush(out, 0x78);   // DEFLATE 32K window
+   stbiw__sbpush(out, 0x5e);   // FLEVEL = 1
+   stbiw__zlib_add(1,1);  // BFINAL = 1
+   stbiw__zlib_add(1,2);  // BTYPE = 1 -- fixed huffman
+
+   for (i=0; i < stbiw__ZHASH; ++i)
+      hash_table[i] = NULL;
+
+   i=0;
+   while (i < data_len-3) {
+      // hash next 3 bytes of data to be compressed
+      int h = stbiw__zhash(data+i)&(stbiw__ZHASH-1), best=3;
+      unsigned char *bestloc = 0;
+      unsigned char **hlist = hash_table[h];
+      int n = stbiw__sbcount(hlist);
+      for (j=0; j < n; ++j) {
+         if (hlist[j]-data > i-32768) { // if entry lies within window
+            int d = stbiw__zlib_countm(hlist[j], data+i, data_len-i);
+            if (d >= best) best=d,bestloc=hlist[j];
+         }
+      }
+      // when hash table entry is too long, delete half the entries
+      if (hash_table[h] && stbiw__sbn(hash_table[h]) == 2*quality) {
+         STBIW_MEMMOVE(hash_table[h], hash_table[h]+quality, sizeof(hash_table[h][0])*quality);
+         stbiw__sbn(hash_table[h]) = quality;
+      }
+      stbiw__sbpush(hash_table[h],data+i);
+
+      if (bestloc) {
+         // "lazy matching" - check match at *next* byte, and if it's better, do cur byte as literal
+         h = stbiw__zhash(data+i+1)&(stbiw__ZHASH-1);
+         hlist = hash_table[h];
+         n = stbiw__sbcount(hlist);
+         for (j=0; j < n; ++j) {
+            if (hlist[j]-data > i-32767) {
+               int e = stbiw__zlib_countm(hlist[j], data+i+1, data_len-i-1);
+               if (e > best) { // if next match is better, bail on current match
+                  bestloc = NULL;
+                  break;
+               }
+            }
+         }
+      }
+
+      if (bestloc) {
+         int d = (int) (data+i - bestloc); // distance back
+         STBIW_ASSERT(d <= 32767 && best <= 258);
+         for (j=0; best > lengthc[j+1]-1; ++j);
+         stbiw__zlib_huff(j+257);
+         if (lengtheb[j]) stbiw__zlib_add(best - lengthc[j], lengtheb[j]);
+         for (j=0; d > distc[j+1]-1; ++j);
+         stbiw__zlib_add(stbiw__zlib_bitrev(j,5),5);
+         if (disteb[j]) stbiw__zlib_add(d - distc[j], disteb[j]);
+         i += best;
+      } else {
+         stbiw__zlib_huffb(data[i]);
+         ++i;
+      }
+   }
+   // write out final bytes
+   for (;i < data_len; ++i)
+      stbiw__zlib_huffb(data[i]);
+   stbiw__zlib_huff(256); // end of block
+   // pad with 0 bits to byte boundary
+   while (bitcount)
+      stbiw__zlib_add(0,1);
+
+   for (i=0; i < stbiw__ZHASH; ++i)
+      (void) stbiw__sbfree(hash_table[i]);
+   STBIW_FREE(hash_table);
+
+   {
+      // compute adler32 on input
+      unsigned int s1=1, s2=0;
+      int blocklen = (int) (data_len % 5552);
+      j=0;
+      while (j < data_len) {
+         for (i=0; i < blocklen; ++i) s1 += data[j+i], s2 += s1;
+         s1 %= 65521, s2 %= 65521;
+         j += blocklen;
+         blocklen = 5552;
+      }
+      stbiw__sbpush(out, STBIW_UCHAR(s2 >> 8));
+      stbiw__sbpush(out, STBIW_UCHAR(s2));
+      stbiw__sbpush(out, STBIW_UCHAR(s1 >> 8));
+      stbiw__sbpush(out, STBIW_UCHAR(s1));
+   }
+   *out_len = stbiw__sbn(out);
+   // make returned pointer freeable
+   STBIW_MEMMOVE(stbiw__sbraw(out), out, *out_len);
+   return (unsigned char *) stbiw__sbraw(out);
+}
+
+static unsigned int stbiw__crc32(unsigned char *buffer, int len)
+{
+   static unsigned int crc_table[256] =
+   {
+      0x00000000, 0x77073096, 0xEE0E612C, 0x990951BA, 0x076DC419, 0x706AF48F, 0xE963A535, 0x9E6495A3,
+      0x0eDB8832, 0x79DCB8A4, 0xE0D5E91E, 0x97D2D988, 0x09B64C2B, 0x7EB17CBD, 0xE7B82D07, 0x90BF1D91,
+      0x1DB71064, 0x6AB020F2, 0xF3B97148, 0x84BE41DE, 0x1ADAD47D, 0x6DDDE4EB, 0xF4D4B551, 0x83D385C7,
+      0x136C9856, 0x646BA8C0, 0xFD62F97A, 0x8A65C9EC, 0x14015C4F, 0x63066CD9, 0xFA0F3D63, 0x8D080DF5,
+      0x3B6E20C8, 0x4C69105E, 0xD56041E4, 0xA2677172, 0x3C03E4D1, 0x4B04D447, 0xD20D85FD, 0xA50AB56B,
+      0x35B5A8FA, 0x42B2986C, 0xDBBBC9D6, 0xACBCF940, 0x32D86CE3, 0x45DF5C75, 0xDCD60DCF, 0xABD13D59,
+      0x26D930AC, 0x51DE003A, 0xC8D75180, 0xBFD06116, 0x21B4F4B5, 0x56B3C423, 0xCFBA9599, 0xB8BDA50F,
+      0x2802B89E, 0x5F058808, 0xC60CD9B2, 0xB10BE924, 0x2F6F7C87, 0x58684C11, 0xC1611DAB, 0xB6662D3D,
+      0x76DC4190, 0x01DB7106, 0x98D220BC, 0xEFD5102A, 0x71B18589, 0x06B6B51F, 0x9FBFE4A5, 0xE8B8D433,
+      0x7807C9A2, 0x0F00F934, 0x9609A88E, 0xE10E9818, 0x7F6A0DBB, 0x086D3D2D, 0x91646C97, 0xE6635C01,
+      0x6B6B51F4, 0x1C6C6162, 0x856530D8, 0xF262004E, 0x6C0695ED, 0x1B01A57B, 0x8208F4C1, 0xF50FC457,
+      0x65B0D9C6, 0x12B7E950, 0x8BBEB8EA, 0xFCB9887C, 0x62DD1DDF, 0x15DA2D49, 0x8CD37CF3, 0xFBD44C65,
+      0x4DB26158, 0x3AB551CE, 0xA3BC0074, 0xD4BB30E2, 0x4ADFA541, 0x3DD895D7, 0xA4D1C46D, 0xD3D6F4FB,
+      0x4369E96A, 0x346ED9FC, 0xAD678846, 0xDA60B8D0, 0x44042D73, 0x33031DE5, 0xAA0A4C5F, 0xDD0D7CC9,
+      0x5005713C, 0x270241AA, 0xBE0B1010, 0xC90C2086, 0x5768B525, 0x206F85B3, 0xB966D409, 0xCE61E49F,
+      0x5EDEF90E, 0x29D9C998, 0xB0D09822, 0xC7D7A8B4, 0x59B33D17, 0x2EB40D81, 0xB7BD5C3B, 0xC0BA6CAD,
+      0xEDB88320, 0x9ABFB3B6, 0x03B6E20C, 0x74B1D29A, 0xEAD54739, 0x9DD277AF, 0x04DB2615, 0x73DC1683,
+      0xE3630B12, 0x94643B84, 0x0D6D6A3E, 0x7A6A5AA8, 0xE40ECF0B, 0x9309FF9D, 0x0A00AE27, 0x7D079EB1,
+      0xF00F9344, 0x8708A3D2, 0x1E01F268, 0x6906C2FE, 0xF762575D, 0x806567CB, 0x196C3671, 0x6E6B06E7,
+      0xFED41B76, 0x89D32BE0, 0x10DA7A5A, 0x67DD4ACC, 0xF9B9DF6F, 0x8EBEEFF9, 0x17B7BE43, 0x60B08ED5,
+      0xD6D6A3E8, 0xA1D1937E, 0x38D8C2C4, 0x4FDFF252, 0xD1BB67F1, 0xA6BC5767, 0x3FB506DD, 0x48B2364B,
+      0xD80D2BDA, 0xAF0A1B4C, 0x36034AF6, 0x41047A60, 0xDF60EFC3, 0xA867DF55, 0x316E8EEF, 0x4669BE79,
+      0xCB61B38C, 0xBC66831A, 0x256FD2A0, 0x5268E236, 0xCC0C7795, 0xBB0B4703, 0x220216B9, 0x5505262F,
+      0xC5BA3BBE, 0xB2BD0B28, 0x2BB45A92, 0x5CB36A04, 0xC2D7FFA7, 0xB5D0CF31, 0x2CD99E8B, 0x5BDEAE1D,
+      0x9B64C2B0, 0xEC63F226, 0x756AA39C, 0x026D930A, 0x9C0906A9, 0xEB0E363F, 0x72076785, 0x05005713,
+      0x95BF4A82, 0xE2B87A14, 0x7BB12BAE, 0x0CB61B38, 0x92D28E9B, 0xE5D5BE0D, 0x7CDCEFB7, 0x0BDBDF21,
+      0x86D3D2D4, 0xF1D4E242, 0x68DDB3F8, 0x1FDA836E, 0x81BE16CD, 0xF6B9265B, 0x6FB077E1, 0x18B74777,
+      0x88085AE6, 0xFF0F6A70, 0x66063BCA, 0x11010B5C, 0x8F659EFF, 0xF862AE69, 0x616BFFD3, 0x166CCF45,
+      0xA00AE278, 0xD70DD2EE, 0x4E048354, 0x3903B3C2, 0xA7672661, 0xD06016F7, 0x4969474D, 0x3E6E77DB,
+      0xAED16A4A, 0xD9D65ADC, 0x40DF0B66, 0x37D83BF0, 0xA9BCAE53, 0xDEBB9EC5, 0x47B2CF7F, 0x30B5FFE9,
+      0xBDBDF21C, 0xCABAC28A, 0x53B39330, 0x24B4A3A6, 0xBAD03605, 0xCDD70693, 0x54DE5729, 0x23D967BF,
+      0xB3667A2E, 0xC4614AB8, 0x5D681B02, 0x2A6F2B94, 0xB40BBE37, 0xC30C8EA1, 0x5A05DF1B, 0x2D02EF8D
+   };
+
+   unsigned int crc = ~0u;
+   int i;
+   for (i=0; i < len; ++i)
+      crc = (crc >> 8) ^ crc_table[buffer[i] ^ (crc & 0xff)];
+   return ~crc;
+}
+
+#define stbiw__wpng4(o,a,b,c,d) ((o)[0]=STBIW_UCHAR(a),(o)[1]=STBIW_UCHAR(b),(o)[2]=STBIW_UCHAR(c),(o)[3]=STBIW_UCHAR(d),(o)+=4)
+#define stbiw__wp32(data,v) stbiw__wpng4(data, (v)>>24,(v)>>16,(v)>>8,(v));
+#define stbiw__wptag(data,s) stbiw__wpng4(data, s[0],s[1],s[2],s[3])
+
+static void stbiw__wpcrc(unsigned char **data, int len)
+{
+   unsigned int crc = stbiw__crc32(*data - len - 4, len+4);
+   stbiw__wp32(*data, crc);
+}
+
+static unsigned char stbiw__paeth(int a, int b, int c)
+{
+   int p = a + b - c, pa = abs(p-a), pb = abs(p-b), pc = abs(p-c);
+   if (pa <= pb && pa <= pc) return STBIW_UCHAR(a);
+   if (pb <= pc) return STBIW_UCHAR(b);
+   return STBIW_UCHAR(c);
+}
+
+unsigned char *stbi_write_png_to_mem(unsigned char *pixels, int stride_bytes, int x, int y, int n, int *out_len)
+{
+   int ctype[5] = { -1, 0, 4, 2, 6 };
+   unsigned char sig[8] = { 137,80,78,71,13,10,26,10 };
+   unsigned char *out,*o, *filt, *zlib;
+   signed char *line_buffer;
+   int i,j,k,p,zlen;
+
+   if (stride_bytes == 0)
+      stride_bytes = x * n;
+
+   filt = (unsigned char *) STBIW_MALLOC((x*n+1) * y); if (!filt) return 0;
+   line_buffer = (signed char *) STBIW_MALLOC(x * n); if (!line_buffer) { STBIW_FREE(filt); return 0; }
+   for (j=0; j < y; ++j) {
+      static int mapping[] = { 0,1,2,3,4 };
+      static int firstmap[] = { 0,1,0,5,6 };
+      int *mymap = j ? mapping : firstmap;
+      int best = 0, bestval = 0x7fffffff;
+      for (p=0; p < 2; ++p) {
+         for (k= p?best:0; k < 5; ++k) {
+            int type = mymap[k],est=0;
+            unsigned char *z = pixels + stride_bytes*j;
+            for (i=0; i < n; ++i)
+               switch (type) {
+                  case 0: line_buffer[i] = z[i]; break;
+                  case 1: line_buffer[i] = z[i]; break;
+                  case 2: line_buffer[i] = z[i] - z[i-stride_bytes]; break;
+                  case 3: line_buffer[i] = z[i] - (z[i-stride_bytes]>>1); break;
+                  case 4: line_buffer[i] = (signed char) (z[i] - stbiw__paeth(0,z[i-stride_bytes],0)); break;
+                  case 5: line_buffer[i] = z[i]; break;
+                  case 6: line_buffer[i] = z[i]; break;
+               }
+            for (i=n; i < x*n; ++i) {
+               switch (type) {
+                  case 0: line_buffer[i] = z[i]; break;
+                  case 1: line_buffer[i] = z[i] - z[i-n]; break;
+                  case 2: line_buffer[i] = z[i] - z[i-stride_bytes]; break;
+                  case 3: line_buffer[i] = z[i] - ((z[i-n] + z[i-stride_bytes])>>1); break;
+                  case 4: line_buffer[i] = z[i] - stbiw__paeth(z[i-n], z[i-stride_bytes], z[i-stride_bytes-n]); break;
+                  case 5: line_buffer[i] = z[i] - (z[i-n]>>1); break;
+                  case 6: line_buffer[i] = z[i] - stbiw__paeth(z[i-n], 0,0); break;
+               }
+            }
+            if (p) break;
+            for (i=0; i < x*n; ++i)
+               est += abs((signed char) line_buffer[i]);
+            if (est < bestval) { bestval = est; best = k; }
+         }
+      }
+      // when we get here, best contains the filter type, and line_buffer contains the data
+      filt[j*(x*n+1)] = (unsigned char) best;
+      STBIW_MEMMOVE(filt+j*(x*n+1)+1, line_buffer, x*n);
+   }
+   STBIW_FREE(line_buffer);
+   zlib = stbi_zlib_compress(filt, y*( x*n+1), &zlen, 8); // increase 8 to get smaller but use more memory
+   STBIW_FREE(filt);
+   if (!zlib) return 0;
+
+   // each tag requires 12 bytes of overhead
+   out = (unsigned char *) STBIW_MALLOC(8 + 12+13 + 12+zlen + 12);
+   if (!out) return 0;
+   *out_len = 8 + 12+13 + 12+zlen + 12;
+
+   o=out;
+   STBIW_MEMMOVE(o,sig,8); o+= 8;
+   stbiw__wp32(o, 13); // header length
+   stbiw__wptag(o, "IHDR");
+   stbiw__wp32(o, x);
+   stbiw__wp32(o, y);
+   *o++ = 8;
+   *o++ = STBIW_UCHAR(ctype[n]);
+   *o++ = 0;
+   *o++ = 0;
+   *o++ = 0;
+   stbiw__wpcrc(&o,13);
+
+   stbiw__wp32(o, zlen);
+   stbiw__wptag(o, "IDAT");
+   STBIW_MEMMOVE(o, zlib, zlen);
+   o += zlen;
+   STBIW_FREE(zlib);
+   stbiw__wpcrc(&o, zlen);
+
+   stbiw__wp32(o,0);
+   stbiw__wptag(o, "IEND");
+   stbiw__wpcrc(&o,0);
+
+   STBIW_ASSERT(o == out + *out_len);
+
+   return out;
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_png(char const *filename, int x, int y, int comp, const void *data, int stride_bytes)
+{
+   FILE *f;
+   int len;
+   unsigned char *png = stbi_write_png_to_mem((unsigned char *) data, stride_bytes, x, y, comp, &len);
+   if (png == NULL) return 0;
+   f = fopen(filename, "wb");
+   if (!f) { STBIW_FREE(png); return 0; }
+   fwrite(png, 1, len, f);
+   fclose(f);
+   STBIW_FREE(png);
+   return 1;
+}
+#endif
+
+STBIWDEF int stbi_write_png_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int stride_bytes)
+{
+   int len;
+   unsigned char *png = stbi_write_png_to_mem((unsigned char *) data, stride_bytes, x, y, comp, &len);
+   if (png == NULL) return 0;
+   func(context, png, len);
+   STBIW_FREE(png);
+   return 1;
+}
+
+#endif // STB_IMAGE_WRITE_IMPLEMENTATION
+
+/* Revision history
+      1.02 (2016-04-02)
+             avoid allocating large structures on the stack
+      1.01 (2016-01-16)
+             STBIW_REALLOC_SIZED: support allocators with no realloc support
+             avoid race-condition in crc initialization
+             minor compile issues
+      1.00 (2015-09-14)
+             installable file IO function
+      0.99 (2015-09-13)
+             warning fixes; TGA rle support
+      0.98 (2015-04-08)
+             added STBIW_MALLOC, STBIW_ASSERT etc
+      0.97 (2015-01-18)
+             fixed HDR asserts, rewrote HDR rle logic
+      0.96 (2015-01-17)
+             add HDR output
+             fix monochrome BMP
+      0.95 (2014-08-17)
+		       add monochrome TGA output
+      0.94 (2014-05-31)
+             rename private functions to avoid conflicts with stb_image.h
+      0.93 (2014-05-27)
+             warning fixes
+      0.92 (2010-08-01)
+             casts to unsigned char to fix warnings
+      0.91 (2010-07-17)
+             first public release
+      0.90   first internal release
+*/
\ No newline at end of file
diff --git a/win64/src/main.c b/win64/src/main.c
new file mode 100644
index 0000000..52835ef
--- /dev/null
+++ b/win64/src/main.c
@@ -0,0 +1,77 @@
+#include <stdint.h>
+#include <stdlib.h>
+#include <stb_image.h>
+#include <assert.h>
+#include <mem.h>
+#include <stb_image_write.h>
+#include <normalmap.h>
+#include <stdbool.h>
+
+
+const char *get_filename_ext(const char *filename) {
+    const char *dot = strrchr(filename, '.');
+    if (!dot || dot == filename) return "";
+    return dot + 1;
+}
+
+
+int main(int argc, const char **argv) {
+    if (argc < 3) {
+        fprintf(stderr, "Usage: %s <heightmap_filename> <bumpmap_filename>", argv[0]);
+        exit(-1);
+    }
+
+    const char *infile = argv[1];
+    const char *outfile = argv[2];
+    int x, y, n;
+    uint8_t *image_in = stbi_load(infile, &x, &y, &n, 4);
+    if (!image_in) {
+        fprintf(stderr, "File %s was not loaded: %s", infile, stbi_failure_reason());
+        exit(-2);
+    }
+    uint8_t *image_out = malloc(x * y * 4);
+    struct NormalmapVals config;
+    config.filter = FILTER_NONE;
+    config.wrap = false;
+    config.conversion = CONVERT_RED;
+    config.scale = 2.0f;
+    config.dudv = DUDV_8BIT_UNSIGNED;
+
+    //// TODO: expose this via command line switches.
+    //struct NormalmapVals config = {
+    //        filter: FILTER_NONE,
+    //        wrap: false,
+    //        conversion: CONVERT_RED,
+    //        scale: 2.0f,
+    //        dudv: DUDV_8BIT_UNSIGNED
+    //};
+
+    int32_t result;
+    if ((result = normalmap(image_in, image_out, x, y, config)) != 0) {
+        fprintf(stderr, "Failed building normal map, code %d", result);
+    }
+
+    const char *ext = get_filename_ext(outfile);
+    if (strcmp("png", ext) == 0) {
+        if (!stbi_write_png(outfile, x, y, 4, image_out, 0)) {
+            fprintf(stderr, "Unable to write PNG %s", outfile);
+        }
+    } else if (strcmp("bmp", ext) == 0) {
+        if (!stbi_write_bmp(outfile, x, y, 4, image_out)) {
+            fprintf(stderr, "Unable to write BMP %s", outfile);
+        }
+    } else if (strcmp("tga", ext) == 0) {
+        if (!stbi_write_tga(outfile, x, y, 4, image_out)) {
+            fprintf(stderr, "Unable to write TGA %s", outfile);
+        }
+//    } else if (strcmp("hdr", ext) == 0) {
+//        if (!stbi_write_hdr(outfile, x, y, 4, image_out)) {
+//            fprintf(stderr, "Unable to write HDR %s", outfile);
+//        }
+    } else {
+        fprintf(stderr, "Unknown file format for save: %s", ext);
+        exit(-1);
+    }
+    stbi_image_free(image_in);
+    free(image_out);
+}
\ No newline at end of file
diff --git a/win64/src/normalmap.c b/win64/src/normalmap.c
new file mode 100644
index 0000000..173da68
--- /dev/null
+++ b/win64/src/normalmap.c
@@ -0,0 +1,971 @@
+/*
+   Copyright (C) 2002-2008 Shawn Kirst <skirst@insightbb.com>
+
+   This program is free software; you can redistribute it and/or
+   modify it under the terms of the GNU General Public
+   License as published by the Free Software Foundation; either
+   version 2 of the License, or (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program; see the file COPYING.  If not, write to
+   the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.
+*/
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <inttypes.h>
+#include <time.h>
+#include <io.h>
+#include <math.h>
+#include <normalmap.h>
+
+#define PREVIEW_SIZE 150
+
+static
+int sample_alpha_map(unsigned char *pixels, int x, int y, int w, int h, int sw, int sh);
+
+static inline int icerp(int a, int b, int c, int d, int x)
+{
+    int p = (d - c) - (a - b);
+    int q = (a - b) - p;
+    int r = c - a;
+    return((x * (x * (x * p + (q << 7)) + (r << 14)) + (b << 21)) >> 21);
+}
+
+void scale_pixels(unsigned char *dst, int dw, int dh, unsigned char *src, int sw, int sh, int bpp)
+{
+    int x, y, n, ix, iy, wx, wy, v;
+    int a, b, c, d;
+    int dstride = dw * bpp;
+    unsigned char *s;
+
+    for(y = 0; y < dh; ++y)
+    {
+        if(dh > 1)
+        {
+            iy = (((sh - 1) * y) << 7) / (dh - 1);
+            if(y == dh - 1) --iy;
+            wy = iy & 0x7f;
+            iy >>= 7;
+        }
+        else
+            iy = wy = 0;
+
+        for(x = 0; x < dw; ++x)
+        {
+            if(dw > 1)
+            {
+                ix = (((sw - 1) * x) << 7) / (dw - 1);
+                if(x == dw - 1) --ix;
+                wx = ix & 0x7f;
+                ix >>= 7;
+            }
+            else
+                ix = wx = 0;
+
+            s = src + ((iy - 1) * sw + (ix - 1)) * bpp;
+
+            for(n = 0; n < bpp; ++n)
+            {
+                b = icerp(s[(sw + 0) * bpp],
+                          s[(sw + 1) * bpp],
+                          s[(sw + 2) * bpp],
+                          s[(sw + 3) * bpp], wx);
+                if(iy > 0)
+                {
+                    a = icerp(s[      0],
+                              s[    bpp],
+                              s[2 * bpp],
+                              s[3 * bpp], wx);
+                }
+                else
+                    a = b;
+
+                c = icerp(s[(2 * sw + 0) * bpp],
+                          s[(2 * sw + 1) * bpp],
+                          s[(2 * sw + 2) * bpp],
+                          s[(2 * sw + 3) * bpp], wx);
+                if(iy < dh - 1)
+                {
+                    d = icerp(s[(3 * sw + 0) * bpp],
+                              s[(3 * sw + 1) * bpp],
+                              s[(3 * sw + 2) * bpp],
+                              s[(3 * sw + 3) * bpp], wx);
+                }
+                else
+                    d = c;
+
+                v = icerp(a, b, c, d, wy);
+                if(v < 0) v = 0;
+                if(v > 255) v = 255;
+                dst[(y * dstride) + (x * bpp) + n] = v;
+                ++s;
+            }
+        }
+    }
+}
+
+static const float oneover255 = 1.0f / 255.0f;
+
+#ifndef min
+#define min(a,b)  ((a)<(b) ? (a) : (b))
+#endif
+#ifndef max
+#define max(a,b)  ((a)>(b) ? (a) : (b))
+#endif
+
+#define SQR(x)      ((x) * (x))
+#define LERP(a,b,c) ((a) + ((b) - (a)) * (c))
+
+static inline void NORMALIZE(float *v)
+{
+    float len = sqrtf(SQR(v[0]) + SQR(v[1]) + SQR(v[2]));
+
+    if(len > 1e-04f)
+    {
+        len = 1.0f / len;
+        v[0] *= len;
+        v[1] *= len;
+        v[2] *= len;
+    }
+    else
+        v[0] = v[1] = v[2] = 0;
+}
+
+typedef struct
+{
+    int x,y;
+    float w;
+} kernel_element;
+
+static void make_kernel(kernel_element *k, float *weights, int size)
+{
+    int x, y, idx;
+
+    for(y = 0; y < size; ++y)
+    {
+        for(x = 0; x < size; ++x)
+        {
+            idx = x + y * size;
+            k[idx].x = x - (size / 2);
+            k[idx].y = (size / 2) - y;
+            k[idx].w = weights[idx];
+        }
+    }
+}
+
+static void rotate_array(float *dst, float *src, int size)
+{
+    int x, y, newx, newy;
+
+    for(y = 0; y < size; ++y)
+    {
+        for(x = 0; x < size; ++x)
+        {
+            newy = size - x - 1;
+            newx = y;
+            dst[newx + newy * size] = src[x + y * size];
+        }
+    }
+}
+
+int sample_alpha_map(unsigned char *pixels, int x, int y,
+                     int w, int h, int sw, int sh)
+{
+    int ix, iy, wx, wy, v;
+    int a, b, c, d;
+    unsigned char *s;
+
+    if(sh > 1)
+    {
+        iy = (((h - 1) * y) << 7) / (sh - 1);
+        if(y == sh - 1) --iy;
+        wy = iy & 0x7f;
+        iy >>= 7;
+    }
+    else
+        iy = wy = 0;
+
+    if(sw > 1)
+    {
+        ix = (((w - 1) * x) << 7) / (sw - 1);
+        if(x == sw - 1) --ix;
+        wx = ix & 0x7f;
+        ix >>= 7;
+    }
+    else
+        ix = wx = 0;
+
+    s = pixels + ((iy - 1) * w + (ix - 1));
+
+    b = icerp(s[w + 0],
+              s[w + 1],
+              s[w + 2],
+              s[w + 3], wx);
+    if(iy > 0)
+    {
+        a = icerp(s[0],
+                  s[1],
+                  s[2],
+                  s[3], wx);
+    }
+    else
+        a = b;
+
+    c = icerp(s[2 * w + 0],
+              s[2 * w + 1],
+              s[2 * w + 2],
+              s[2 * w + 3], wx);
+    if(iy < sh - 1)
+    {
+        d = icerp(s[3 * w + 0],
+                  s[3 * w + 1],
+                  s[3 * w + 2],
+                  s[3 * w + 3], wx);
+    }
+    else
+        d = c;
+
+    v = icerp(a, b, c, d, wy);
+
+    if(v <   0) v = 0;
+    if(v > 255) v = 255;
+
+    return((unsigned char)v);
+}
+
+static void make_heightmap(unsigned char *image, int w, int h, int bpp, struct NormalmapVals nmapvals)
+{
+    unsigned int i, num_pixels = w * h;
+    int x, y;
+    float v, hmin, hmax;
+    float *s, *r;
+
+    s = (float*)malloc(w * h * 3 * sizeof(float));
+    if(s == 0)
+    {
+        perror("Memory allocation error!");
+        return;
+    }
+    r = (float*)malloc(w * h * 4 * sizeof(float));
+    if(r == 0)
+    {
+        free(s);
+        perror("Memory allocation error!");
+        return;
+    }
+
+    /* scale into 0 to 1 range, make signed -1 to 1 */
+    for(i = 0; i < num_pixels; ++i)
+    {
+        s[3 * i + 0] = (((float)image[bpp * i + 0] / 255.0f) - 0.5) * 2.0f;
+        s[3 * i + 1] = (((float)image[bpp * i + 1] / 255.0f) - 0.5) * 2.0f;
+        s[3 * i + 2] = (((float)image[bpp * i + 2] / 255.0f) - 0.5) * 2.0f;
+    }
+
+    memset(r, 0, w * h * 4 * sizeof(float));
+
+#define S(x, y, n) s[(y) * (w * 3) + ((x) * 3) + (n)]
+#define R(x, y, n) r[(y) * (w * 4) + ((x) * 4) + (n)]
+
+    /* top-left to bottom-right */
+    for(x = 1; x < w; ++x)
+        R(x, 0, 0) = R(x - 1, 0, 0) + S(x - 1, 0, 0);
+    for(y = 1; y < h; ++y)
+        R(0, y, 0) = R(0, y - 1, 0) + S(0, y - 1, 1);
+    for(y = 1; y < h; ++y)
+    {
+        for(x = 1; x < w; ++x)
+        {
+            R(x, y, 0) = (R(x, y - 1, 0) + R(x - 1, y, 0) +
+                          S(x - 1, y, 0) + S(x, y - 1, 1)) * 0.5f;
+        }
+    }
+
+    /* top-right to bottom-left */
+    for(x = w - 2; x >= 0; --x)
+        R(x, 0, 1) = R(x + 1, 0, 1) - S(x + 1, 0, 0);
+    for(y = 1; y < h; ++y)
+        R(0, y, 1) = R(0, y - 1, 1) + S(0, y - 1, 1);
+    for(y = 1; y < h; ++y)
+    {
+        for(x = w - 2; x >= 0; --x)
+        {
+            R(x, y, 1) = (R(x, y - 1, 1) + R(x + 1, y, 1) -
+                          S(x + 1, y, 0) + S(x, y - 1, 1)) * 0.5f;
+        }
+    }
+
+    /* bottom-left to top-right */
+    for(x = 1; x < w; ++x)
+        R(x, 0, 2) = R(x - 1, 0, 2) + S(x - 1, 0, 0);
+    for(y = h - 2; y >= 0; --y)
+        R(0, y, 2) = R(0, y + 1, 2) - S(0, y + 1, 1);
+    for(y = h - 2; y >= 0; --y)
+    {
+        for(x = 1; x < w; ++x)
+        {
+            R(x, y, 2) = (R(x, y + 1, 2) + R(x - 1, y, 2) +
+                          S(x - 1, y, 0) - S(x, y + 1, 1)) * 0.5f;
+        }
+    }
+
+    /* bottom-right to top-left */
+    for(x = w - 2; x >= 0; --x)
+        R(x, 0, 3) = R(x + 1, 0, 3) - S(x + 1, 0, 0);
+    for(y = h - 2; y >= 0; --y)
+        R(0, y, 3) = R(0, y + 1, 3) - S(0, y + 1, 1);
+    for(y = h - 2; y >= 0; --y)
+    {
+        for(x = w - 2; x >= 0; --x)
+        {
+            R(x, y, 3) = (R(x, y + 1, 3) + R(x + 1, y, 3) -
+                          S(x + 1, y, 0) - S(x, y + 1, 1)) * 0.5f;
+        }
+    }
+
+#undef S
+#undef R
+
+    /* accumulate, find min/max */
+    hmin =  1e10f;
+    hmax = -1e10f;
+    for(i = 0; i < num_pixels; ++i)
+    {
+        r[4 * i] += r[4 * i + 1] + r[4 * i + 2] + r[4 * i + 3];
+        if(r[4 * i] < hmin) hmin = r[4 * i];
+        if(r[4 * i] > hmax) hmax = r[4 * i];
+    }
+
+    /* scale into 0 - 1 range */
+    for(i = 0; i < num_pixels; ++i)
+    {
+        v = (r[4 * i] - hmin) / (hmax - hmin);
+        /* adjust contrast */
+        v = (v - 0.5f) * nmapvals.contrast + v;
+        if(v < 0) v = 0;
+        if(v > 1) v = 1;
+        r[4 * i] = v;
+    }
+
+    /* write out results */
+    for(i = 0; i < num_pixels; ++i)
+    {
+        v = r[4 * i] * 255.0f;
+        image[bpp * i + 0] = (unsigned char)v;
+        image[bpp * i + 1] = (unsigned char)v;
+        image[bpp * i + 2] = (unsigned char)v;
+    }
+
+    free(s);
+    free(r);
+}
+
+int32_t normalmap(uint8_t * img, uint8_t * nm_img, size_t width, size_t height, struct NormalmapVals nmapvals)
+{
+    const size_t bpp = 4;
+    size_t rowbytes;
+    int x, y;
+    int pw, ph, amap_w = 0, amap_h = 0;
+    unsigned char *d, *dst, *s, *src, *tmp, *amap = 0;
+    float *heights;
+    float val, du, dv, n[3], weight;
+    float rgb_bias[3];
+    int i, num_elements = 0;
+    kernel_element *kernel_du = 0;
+    kernel_element *kernel_dv = 0;
+
+    if(nmapvals.filter < 0 || nmapvals.filter >= MAX_FILTER_TYPE)
+        nmapvals.filter = FILTER_NONE;
+
+    rowbytes = width * bpp;
+
+    dst = nm_img;
+    src = img;
+
+    heights = calloc(width * height, sizeof(float));
+    if(heights == 0)
+    {
+        perror("Memory allocation error!");
+        return(-1);
+    }
+
+    switch(nmapvals.filter)
+    {
+        default:
+        case FILTER_NONE:
+            num_elements = 2;
+            kernel_du = (kernel_element*)malloc(2 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(2 * sizeof(kernel_element));
+
+            kernel_du[0].x = -1; kernel_du[0].y = 0; kernel_du[0].w = -0.5f;
+            kernel_du[1].x =  1; kernel_du[1].y = 0; kernel_du[1].w =  0.5f;
+
+            kernel_dv[0].x = 0; kernel_dv[0].y =  1; kernel_dv[0].w =  0.5f;
+            kernel_dv[1].x = 0; kernel_dv[1].y = -1; kernel_dv[1].w = -0.5f;
+
+            break;
+        case FILTER_SOBEL_3x3:
+            num_elements = 6;
+            kernel_du = (kernel_element*)malloc(6 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(6 * sizeof(kernel_element));
+
+            kernel_du[0].x = -1; kernel_du[0].y =  1; kernel_du[0].w = -1.0f;
+            kernel_du[1].x = -1; kernel_du[1].y =  0; kernel_du[1].w = -2.0f;
+            kernel_du[2].x = -1; kernel_du[2].y = -1; kernel_du[2].w = -1.0f;
+            kernel_du[3].x =  1; kernel_du[3].y =  1; kernel_du[3].w =  1.0f;
+            kernel_du[4].x =  1; kernel_du[4].y =  0; kernel_du[4].w =  2.0f;
+            kernel_du[5].x =  1; kernel_du[5].y = -1; kernel_du[5].w =  1.0f;
+
+            kernel_dv[0].x = -1; kernel_dv[0].y =  1; kernel_dv[0].w =  1.0f;
+            kernel_dv[1].x =  0; kernel_dv[1].y =  1; kernel_dv[1].w =  2.0f;
+            kernel_dv[2].x =  1; kernel_dv[2].y =  1; kernel_dv[2].w =  1.0f;
+            kernel_dv[3].x = -1; kernel_dv[3].y = -1; kernel_dv[3].w = -1.0f;
+            kernel_dv[4].x =  0; kernel_dv[4].y = -1; kernel_dv[4].w = -2.0f;
+            kernel_dv[5].x =  1; kernel_dv[5].y = -1; kernel_dv[5].w = -1.0f;
+
+            break;
+        case FILTER_SOBEL_5x5:
+            num_elements = 20;
+            kernel_du = (kernel_element*)malloc(20 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(20 * sizeof(kernel_element));
+
+            kernel_du[ 0].x = -2; kernel_du[ 0].y =  2; kernel_du[ 0].w =  -1.0f;
+            kernel_du[ 1].x = -2; kernel_du[ 1].y =  1; kernel_du[ 1].w =  -4.0f;
+            kernel_du[ 2].x = -2; kernel_du[ 2].y =  0; kernel_du[ 2].w =  -6.0f;
+            kernel_du[ 3].x = -2; kernel_du[ 3].y = -1; kernel_du[ 3].w =  -4.0f;
+            kernel_du[ 4].x = -2; kernel_du[ 4].y = -2; kernel_du[ 4].w =  -1.0f;
+            kernel_du[ 5].x = -1; kernel_du[ 5].y =  2; kernel_du[ 5].w =  -2.0f;
+            kernel_du[ 6].x = -1; kernel_du[ 6].y =  1; kernel_du[ 6].w =  -8.0f;
+            kernel_du[ 7].x = -1; kernel_du[ 7].y =  0; kernel_du[ 7].w = -12.0f;
+            kernel_du[ 8].x = -1; kernel_du[ 8].y = -1; kernel_du[ 8].w =  -8.0f;
+            kernel_du[ 9].x = -1; kernel_du[ 9].y = -2; kernel_du[ 9].w =  -2.0f;
+            kernel_du[10].x =  1; kernel_du[10].y =  2; kernel_du[10].w =   2.0f;
+            kernel_du[11].x =  1; kernel_du[11].y =  1; kernel_du[11].w =   8.0f;
+            kernel_du[12].x =  1; kernel_du[12].y =  0; kernel_du[12].w =  12.0f;
+            kernel_du[13].x =  1; kernel_du[13].y = -1; kernel_du[13].w =   8.0f;
+            kernel_du[14].x =  1; kernel_du[14].y = -2; kernel_du[14].w =   2.0f;
+            kernel_du[15].x =  2; kernel_du[15].y =  2; kernel_du[15].w =   1.0f;
+            kernel_du[16].x =  2; kernel_du[16].y =  1; kernel_du[16].w =   4.0f;
+            kernel_du[17].x =  2; kernel_du[17].y =  0; kernel_du[17].w =   6.0f;
+            kernel_du[18].x =  2; kernel_du[18].y = -1; kernel_du[18].w =   4.0f;
+            kernel_du[19].x =  2; kernel_du[19].y = -2; kernel_du[19].w =   1.0f;
+
+            kernel_dv[ 0].x = -2; kernel_dv[ 0].y =  2; kernel_dv[ 0].w =   1.0f;
+            kernel_dv[ 1].x = -1; kernel_dv[ 1].y =  2; kernel_dv[ 1].w =   4.0f;
+            kernel_dv[ 2].x =  0; kernel_dv[ 2].y =  2; kernel_dv[ 2].w =   6.0f;
+            kernel_dv[ 3].x =  1; kernel_dv[ 3].y =  2; kernel_dv[ 3].w =   4.0f;
+            kernel_dv[ 4].x =  2; kernel_dv[ 4].y =  2; kernel_dv[ 4].w =   1.0f;
+            kernel_dv[ 5].x = -2; kernel_dv[ 5].y =  1; kernel_dv[ 5].w =   2.0f;
+            kernel_dv[ 6].x = -1; kernel_dv[ 6].y =  1; kernel_dv[ 6].w =   8.0f;
+            kernel_dv[ 7].x =  0; kernel_dv[ 7].y =  1; kernel_dv[ 7].w =  12.0f;
+            kernel_dv[ 8].x =  1; kernel_dv[ 8].y =  1; kernel_dv[ 8].w =   8.0f;
+            kernel_dv[ 9].x =  2; kernel_dv[ 9].y =  1; kernel_dv[ 9].w =   2.0f;
+            kernel_dv[10].x = -2; kernel_dv[10].y = -1; kernel_dv[10].w =  -2.0f;
+            kernel_dv[11].x = -1; kernel_dv[11].y = -1; kernel_dv[11].w =  -8.0f;
+            kernel_dv[12].x =  0; kernel_dv[12].y = -1; kernel_dv[12].w = -12.0f;
+            kernel_dv[13].x =  1; kernel_dv[13].y = -1; kernel_dv[13].w =  -8.0f;
+            kernel_dv[14].x =  2; kernel_dv[14].y = -1; kernel_dv[14].w =  -2.0f;
+            kernel_dv[15].x = -2; kernel_dv[15].y = -2; kernel_dv[15].w =  -1.0f;
+            kernel_dv[16].x = -1; kernel_dv[16].y = -2; kernel_dv[16].w =  -4.0f;
+            kernel_dv[17].x =  0; kernel_dv[17].y = -2; kernel_dv[17].w =  -6.0f;
+            kernel_dv[18].x =  1; kernel_dv[18].y = -2; kernel_dv[18].w =  -4.0f;
+            kernel_dv[19].x =  2; kernel_dv[19].y = -2; kernel_dv[19].w =  -1.0f;
+
+            break;
+        case FILTER_PREWITT_3x3:
+            num_elements = 6;
+            kernel_du = (kernel_element*)malloc(6 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(6 * sizeof(kernel_element));
+
+            kernel_du[0].x = -1; kernel_du[0].y =  1; kernel_du[0].w = -1.0f;
+            kernel_du[1].x = -1; kernel_du[1].y =  0; kernel_du[1].w = -1.0f;
+            kernel_du[2].x = -1; kernel_du[2].y = -1; kernel_du[2].w = -1.0f;
+            kernel_du[3].x =  1; kernel_du[3].y =  1; kernel_du[3].w =  1.0f;
+            kernel_du[4].x =  1; kernel_du[4].y =  0; kernel_du[4].w =  1.0f;
+            kernel_du[5].x =  1; kernel_du[5].y = -1; kernel_du[5].w =  1.0f;
+
+            kernel_dv[0].x = -1; kernel_dv[0].y =  1; kernel_dv[0].w =  1.0f;
+            kernel_dv[1].x =  0; kernel_dv[1].y =  1; kernel_dv[1].w =  1.0f;
+            kernel_dv[2].x =  1; kernel_dv[2].y =  1; kernel_dv[2].w =  1.0f;
+            kernel_dv[3].x = -1; kernel_dv[3].y = -1; kernel_dv[3].w = -1.0f;
+            kernel_dv[4].x =  0; kernel_dv[4].y = -1; kernel_dv[4].w = -1.0f;
+            kernel_dv[5].x =  1; kernel_dv[5].y = -1; kernel_dv[5].w = -1.0f;
+
+            break;
+        case FILTER_PREWITT_5x5:
+            num_elements = 20;
+            kernel_du = (kernel_element*)malloc(20 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(20 * sizeof(kernel_element));
+
+            kernel_du[ 0].x = -2; kernel_du[ 0].y =  2; kernel_du[ 0].w = -1.0f;
+            kernel_du[ 1].x = -2; kernel_du[ 1].y =  1; kernel_du[ 1].w = -1.0f;
+            kernel_du[ 2].x = -2; kernel_du[ 2].y =  0; kernel_du[ 2].w = -1.0f;
+            kernel_du[ 3].x = -2; kernel_du[ 3].y = -1; kernel_du[ 3].w = -1.0f;
+            kernel_du[ 4].x = -2; kernel_du[ 4].y = -2; kernel_du[ 4].w = -1.0f;
+            kernel_du[ 5].x = -1; kernel_du[ 5].y =  2; kernel_du[ 5].w = -2.0f;
+            kernel_du[ 6].x = -1; kernel_du[ 6].y =  1; kernel_du[ 6].w = -2.0f;
+            kernel_du[ 7].x = -1; kernel_du[ 7].y =  0; kernel_du[ 7].w = -2.0f;
+            kernel_du[ 8].x = -1; kernel_du[ 8].y = -1; kernel_du[ 8].w = -2.0f;
+            kernel_du[ 9].x = -1; kernel_du[ 9].y = -2; kernel_du[ 9].w = -2.0f;
+            kernel_du[10].x =  1; kernel_du[10].y =  2; kernel_du[10].w =  2.0f;
+            kernel_du[11].x =  1; kernel_du[11].y =  1; kernel_du[11].w =  2.0f;
+            kernel_du[12].x =  1; kernel_du[12].y =  0; kernel_du[12].w =  2.0f;
+            kernel_du[13].x =  1; kernel_du[13].y = -1; kernel_du[13].w =  2.0f;
+            kernel_du[14].x =  1; kernel_du[14].y = -2; kernel_du[14].w =  2.0f;
+            kernel_du[15].x =  2; kernel_du[15].y =  2; kernel_du[15].w =  1.0f;
+            kernel_du[16].x =  2; kernel_du[16].y =  1; kernel_du[16].w =  1.0f;
+            kernel_du[17].x =  2; kernel_du[17].y =  0; kernel_du[17].w =  1.0f;
+            kernel_du[18].x =  2; kernel_du[18].y = -1; kernel_du[18].w =  1.0f;
+            kernel_du[19].x =  2; kernel_du[19].y = -2; kernel_du[19].w =  1.0f;
+
+            kernel_dv[ 0].x = -2; kernel_dv[ 0].y =  2; kernel_dv[ 0].w =  1.0f;
+            kernel_dv[ 1].x = -1; kernel_dv[ 1].y =  2; kernel_dv[ 1].w =  1.0f;
+            kernel_dv[ 2].x =  0; kernel_dv[ 2].y =  2; kernel_dv[ 2].w =  1.0f;
+            kernel_dv[ 3].x =  1; kernel_dv[ 3].y =  2; kernel_dv[ 3].w =  1.0f;
+            kernel_dv[ 4].x =  2; kernel_dv[ 4].y =  2; kernel_dv[ 4].w =  1.0f;
+            kernel_dv[ 5].x = -2; kernel_dv[ 5].y =  1; kernel_dv[ 5].w =  2.0f;
+            kernel_dv[ 6].x = -1; kernel_dv[ 6].y =  1; kernel_dv[ 6].w =  2.0f;
+            kernel_dv[ 7].x =  0; kernel_dv[ 7].y =  1; kernel_dv[ 7].w =  2.0f;
+            kernel_dv[ 8].x =  1; kernel_dv[ 8].y =  1; kernel_dv[ 8].w =  2.0f;
+            kernel_dv[ 9].x =  2; kernel_dv[ 9].y =  1; kernel_dv[ 9].w =  2.0f;
+            kernel_dv[10].x = -2; kernel_dv[10].y = -1; kernel_dv[10].w = -2.0f;
+            kernel_dv[11].x = -1; kernel_dv[11].y = -1; kernel_dv[11].w = -2.0f;
+            kernel_dv[12].x =  0; kernel_dv[12].y = -1; kernel_dv[12].w = -2.0f;
+            kernel_dv[13].x =  1; kernel_dv[13].y = -1; kernel_dv[13].w = -2.0f;
+            kernel_dv[14].x =  2; kernel_dv[14].y = -1; kernel_dv[14].w = -2.0f;
+            kernel_dv[15].x = -2; kernel_dv[15].y = -2; kernel_dv[15].w = -1.0f;
+            kernel_dv[16].x = -1; kernel_dv[16].y = -2; kernel_dv[16].w = -1.0f;
+            kernel_dv[17].x =  0; kernel_dv[17].y = -2; kernel_dv[17].w = -1.0f;
+            kernel_dv[18].x =  1; kernel_dv[18].y = -2; kernel_dv[18].w = -1.0f;
+            kernel_dv[19].x =  2; kernel_dv[19].y = -2; kernel_dv[19].w = -1.0f;
+
+            break;
+        case FILTER_3x3:
+            num_elements = 6;
+            kernel_du = (kernel_element*)malloc(6 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(6 * sizeof(kernel_element));
+
+            weight = 1.0f / 6.0f;
+
+            kernel_du[0].x = -1; kernel_du[0].y =  1; kernel_du[0].w = -weight;
+            kernel_du[1].x = -1; kernel_du[1].y =  0; kernel_du[1].w = -weight;
+            kernel_du[2].x = -1; kernel_du[2].y = -1; kernel_du[2].w = -weight;
+            kernel_du[3].x =  1; kernel_du[3].y =  1; kernel_du[3].w =  weight;
+            kernel_du[4].x =  1; kernel_du[4].y =  0; kernel_du[4].w =  weight;
+            kernel_du[5].x =  1; kernel_du[5].y = -1; kernel_du[5].w =  weight;
+
+            kernel_dv[0].x = -1; kernel_dv[0].y =  1; kernel_dv[0].w =  weight;
+            kernel_dv[1].x =  0; kernel_dv[1].y =  1; kernel_dv[1].w =  weight;
+            kernel_dv[2].x =  1; kernel_dv[2].y =  1; kernel_dv[2].w =  weight;
+            kernel_dv[3].x = -1; kernel_dv[3].y = -1; kernel_dv[3].w = -weight;
+            kernel_dv[4].x =  0; kernel_dv[4].y = -1; kernel_dv[4].w = -weight;
+            kernel_dv[5].x =  1; kernel_dv[5].y = -1; kernel_dv[5].w = -weight;
+            break;
+        case FILTER_5x5:
+        {
+            size_t j;
+            float usum = 0, vsum = 0;
+            float wt22 = 1.0f / 16.0f;
+            float wt12 = 1.0f / 10.0f;
+            float wt02 = 1.0f / 8.0f;
+            float wt11 = 1.0f / 2.8f;
+            num_elements = 20;
+            kernel_du = (kernel_element*)malloc(20 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(20 * sizeof(kernel_element));
+
+            kernel_du[0 ].x = -2; kernel_du[0 ].y =  2; kernel_du[0 ].w = -wt22;
+            kernel_du[1 ].x = -1; kernel_du[1 ].y =  2; kernel_du[1 ].w = -wt12;
+            kernel_du[2 ].x =  1; kernel_du[2 ].y =  2; kernel_du[2 ].w =  wt12;
+            kernel_du[3 ].x =  2; kernel_du[3 ].y =  2; kernel_du[3 ].w =  wt22;
+            kernel_du[4 ].x = -2; kernel_du[4 ].y =  1; kernel_du[4 ].w = -wt12;
+            kernel_du[5 ].x = -1; kernel_du[5 ].y =  1; kernel_du[5 ].w = -wt11;
+            kernel_du[6 ].x =  1; kernel_du[6 ].y =  1; kernel_du[6 ].w =  wt11;
+            kernel_du[7 ].x =  2; kernel_du[7 ].y =  1; kernel_du[7 ].w =  wt12;
+            kernel_du[8 ].x = -2; kernel_du[8 ].y =  0; kernel_du[8 ].w = -wt02;
+            kernel_du[9 ].x = -1; kernel_du[9 ].y =  0; kernel_du[9 ].w = -0.5f;
+            kernel_du[10].x =  1; kernel_du[10].y =  0; kernel_du[10].w =  0.5f;
+            kernel_du[11].x =  2; kernel_du[11].y =  0; kernel_du[11].w =  wt02;
+            kernel_du[12].x = -2; kernel_du[12].y = -1; kernel_du[12].w = -wt12;
+            kernel_du[13].x = -1; kernel_du[13].y = -1; kernel_du[13].w = -wt11;
+            kernel_du[14].x =  1; kernel_du[14].y = -1; kernel_du[14].w =  wt11;
+            kernel_du[15].x =  2; kernel_du[15].y = -1; kernel_du[15].w =  wt12;
+            kernel_du[16].x = -2; kernel_du[16].y = -2; kernel_du[16].w = -wt22;
+            kernel_du[17].x = -1; kernel_du[17].y = -2; kernel_du[17].w = -wt12;
+            kernel_du[18].x =  1; kernel_du[18].y = -2; kernel_du[18].w =  wt12;
+            kernel_du[19].x =  2; kernel_du[19].y = -2; kernel_du[19].w =  wt22;
+
+            kernel_dv[0 ].x = -2; kernel_dv[0 ].y =  2; kernel_dv[0 ].w =  wt22;
+            kernel_dv[1 ].x = -1; kernel_dv[1 ].y =  2; kernel_dv[1 ].w =  wt12;
+            kernel_dv[2 ].x =  0; kernel_dv[2 ].y =  2; kernel_dv[2 ].w =  0.25f;
+            kernel_dv[3 ].x =  1; kernel_dv[3 ].y =  2; kernel_dv[3 ].w =  wt12;
+            kernel_dv[4 ].x =  2; kernel_dv[4 ].y =  2; kernel_dv[4 ].w =  wt22;
+            kernel_dv[5 ].x = -2; kernel_dv[5 ].y =  1; kernel_dv[5 ].w =  wt12;
+            kernel_dv[6 ].x = -1; kernel_dv[6 ].y =  1; kernel_dv[6 ].w =  wt11;
+            kernel_dv[7 ].x =  0; kernel_dv[7 ].y =  1; kernel_dv[7 ].w =  0.5f;
+            kernel_dv[8 ].x =  1; kernel_dv[8 ].y =  1; kernel_dv[8 ].w =  wt11;
+            kernel_dv[9 ].x =  2; kernel_dv[9 ].y =  1; kernel_dv[9 ].w =  wt22;
+            kernel_dv[10].x = -2; kernel_dv[10].y = -1; kernel_dv[10].w = -wt22;
+            kernel_dv[11].x = -1; kernel_dv[11].y = -1; kernel_dv[11].w = -wt11;
+            kernel_dv[12].x =  0; kernel_dv[12].y = -1; kernel_dv[12].w = -0.5f;
+            kernel_dv[13].x =  1; kernel_dv[13].y = -1; kernel_dv[13].w = -wt11;
+            kernel_dv[14].x =  2; kernel_dv[14].y = -1; kernel_dv[14].w = -wt12;
+            kernel_dv[15].x = -2; kernel_dv[15].y = -2; kernel_dv[15].w = -wt22;
+            kernel_dv[16].x = -1; kernel_dv[16].y = -2; kernel_dv[16].w = -wt12;
+            kernel_dv[17].x =  0; kernel_dv[17].y = -2; kernel_dv[17].w = -0.25f;
+            kernel_dv[18].x =  1; kernel_dv[18].y = -2; kernel_dv[18].w = -wt12;
+            kernel_dv[19].x =  2; kernel_dv[19].y = -2; kernel_dv[19].w = -wt22;
+
+            for(j = 0; j < 20; ++j)
+            {
+                usum += fabsf(kernel_du[j].w);
+                vsum += fabsf(kernel_dv[j].w);
+            }
+            for(j = 0; j < 20; ++j)
+            {
+                kernel_du[j].w /= usum;
+                kernel_dv[j].w /= vsum;
+            }
+
+            break;
+        }
+        case FILTER_7x7:
+        {
+            float du_weights[]=
+                    {
+                            -1, -2, -3, 0, 3, 2, 1,
+                            -2, -3, -4, 0, 4, 3, 2,
+                            -3, -4, -5, 0, 5, 4, 3,
+                            -4, -5, -6, 0, 6, 5, 4,
+                            -3, -4, -5, 0, 5, 4, 3,
+                            -2, -3, -4, 0, 4, 3, 2,
+                            -1, -2, -3, 0, 3, 2, 1
+                    };
+            float dv_weights[49];
+            int n;
+            float usum = 0, vsum = 0;
+
+            num_elements = 49;
+            kernel_du = (kernel_element*)malloc(49 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(49 * sizeof(kernel_element));
+
+            make_kernel(kernel_du, du_weights, 7);
+            rotate_array(dv_weights, du_weights, 7);
+            make_kernel(kernel_dv, dv_weights, 7);
+
+            for(n = 0; n < 49; ++n)
+            {
+                usum += fabsf(kernel_du[n].w);
+                vsum += fabsf(kernel_dv[n].w);
+            }
+            for(n = 0; n < 49; ++n)
+            {
+                kernel_du[n].w /= usum;
+                kernel_dv[n].w /= vsum;
+            }
+
+            break;
+        }
+        case FILTER_9x9:
+        {
+            float du_weights[]=
+                    {
+                            -1, -2, -3, -4, 0, 4, 3, 2, 1,
+                            -2, -3, -4, -5, 0, 5, 4, 3, 2,
+                            -3, -4, -5, -6, 0, 6, 5, 4, 3,
+                            -4, -5, -6, -7, 0, 7, 6, 5, 4,
+                            -5, -6, -7, -8, 0, 8, 7, 6, 5,
+                            -4, -5, -6, -7, 0, 7, 6, 5, 4,
+                            -3, -4, -5, -6, 0, 6, 5, 4, 3,
+                            -2, -3, -4, -5, 0, 5, 4, 3, 2,
+                            -1, -2, -3, -4, 0, 4, 3, 2, 1
+                    };
+            float dv_weights[81];
+            int n;
+            float usum = 0, vsum = 0;
+
+            num_elements = 81;
+            kernel_du = (kernel_element*)malloc(81 * sizeof(kernel_element));
+            kernel_dv = (kernel_element*)malloc(81 * sizeof(kernel_element));
+
+            make_kernel(kernel_du, du_weights, 9);
+            rotate_array(dv_weights, du_weights, 9);
+            make_kernel(kernel_dv, dv_weights, 9);
+
+            for(n = 0; n < 81; ++n)
+            {
+                usum += fabsf(kernel_du[n].w);
+                vsum += fabsf(kernel_dv[n].w);
+            }
+            for(n = 0; n < 81; ++n)
+            {
+                kernel_du[n].w /= usum;
+                kernel_dv[n].w /= vsum;
+            }
+
+            break;
+        }
+    }
+
+    if(nmapvals.conversion == CONVERT_BIASED_RGB)
+    {
+        /* approximated average color of the image
+         * scale to 16x16, accumulate the pixels and average */
+        unsigned int sum[3];
+
+        tmp = malloc(16 * 16 * bpp);
+        scale_pixels(tmp, 16, 16, src, width, height, bpp);
+
+        sum[0] = sum[1] = sum[2] = 0;
+
+        s = src;
+        for(y = 0; y < 16; ++y)
+        {
+            for(x = 0; x < 16; ++x)
+            {
+                sum[0] += *s++;
+                sum[1] += *s++;
+                sum[2] += *s++;
+                s++; // Ignore alpha
+            }
+        }
+
+        rgb_bias[0] = (float)sum[0] / 256.0f;
+        rgb_bias[1] = (float)sum[1] / 256.0f;
+        rgb_bias[2] = (float)sum[2] / 256.0f;
+
+        free(tmp);
+    }
+    else if (nmapvals.conversion == CONVERT_KEY_RGB)
+    {
+        // Pull from key pixel.
+        rgb_bias[0] = src[0];
+        rgb_bias[1] = src[1];
+        rgb_bias[2] = src[2];
+    }
+    else
+    {
+        rgb_bias[0] = 0;
+        rgb_bias[1] = 0;
+        rgb_bias[2] = 0;
+    }
+    printf("RGB bias: R%f G%f B%f", rgb_bias[0], rgb_bias[1], rgb_bias[2]);
+
+    if(nmapvals.conversion != CONVERT_NORMALIZE_ONLY &&
+       nmapvals.conversion != CONVERT_DUDV_TO_NORMAL &&
+       nmapvals.conversion != CONVERT_HEIGHTMAP)
+    {
+        s = src;
+        for(y = 0; y < height; ++y)
+        {
+            for(x = 0; x < width; ++x)
+            {
+                if(!nmapvals.height_source)
+                {
+                    switch(nmapvals.conversion)
+                    {
+                        case CONVERT_NONE:
+                            val = (float)s[0] * 0.3f +
+                                  (float)s[1] * 0.59f +
+                                  (float)s[2] * 0.11f;
+                            break;
+                        case CONVERT_BIASED_RGB:
+                            val = (((float)max(0, s[0] - rgb_bias[0])) * 0.3f ) +
+                                  (((float)max(0, s[1] - rgb_bias[1])) * 0.59f) +
+                                  (((float)max(0, s[2] - rgb_bias[2])) * 0.11f);
+                            break;
+                        case CONVERT_RED:
+                            val = (float)s[0];
+                            break;
+                        case CONVERT_GREEN:
+                            val = (float)s[1];
+                            break;
+                        case CONVERT_BLUE:
+                            val = (float)s[2];
+                            break;
+                        case CONVERT_MAX_RGB:
+                            val = (float)max(s[0], max(s[1], s[2]));
+                            break;
+                        case CONVERT_MIN_RGB:
+                            val = (float)min(s[0], min(s[1], s[2]));
+                            break;
+                        case CONVERT_COLORSPACE:
+                            val = (1.0f - ((1.0f - ((float)s[0] / 255.0f)) *
+                                           (1.0f - ((float)s[1] / 255.0f)) *
+                                           (1.0f - ((float)s[2] / 255.0f)))) * 255.0f;
+                            break;
+                        default:
+                            val = 255.0f;
+                            break;
+                    }
+                }
+                else
+                    val = (float)s[3];
+
+                heights[x + y * width] = val * oneover255;
+//                if (val < 0) printf("%d %d: %f\n", x, y, val);
+
+                s += bpp;
+            }
+        }
+    }
+
+#define HEIGHT(x,y) \
+   (heights[(max(0, min(width - 1, (x)))) + (max(0, min(height - 1, (y)))) * width])
+#define HEIGHT_WRAP(x,y) \
+   (heights[((x) < 0 ? (width + (x)) : ((x) >= width ? ((x) - width) : (x)))+ \
+            (((y) < 0 ? (height + (y)) : ((y) >= height ? ((y) - height) : (y))) * width)])
+
+    for(y = 0; y < height; ++y)
+    {
+        for(x = 0; x < width; ++x)
+        {
+            d = dst + ((y * rowbytes) + (x * bpp));
+            s = src + ((y * rowbytes) + (x * bpp));
+
+            if(nmapvals.conversion == CONVERT_NORMALIZE_ONLY ||
+               nmapvals.conversion == CONVERT_HEIGHTMAP)
+            {
+                n[0] = (((float)s[0] * oneover255) - 0.5f) * 2.0f;
+                n[1] = (((float)s[1] * oneover255) - 0.5f) * 2.0f;
+                n[2] = (((float)s[2] * oneover255) - 0.5f) * 2.0f;
+                n[0] *= nmapvals.scale;
+                n[1] *= nmapvals.scale;
+            }
+            else if(nmapvals.conversion == CONVERT_DUDV_TO_NORMAL)
+            {
+                n[0] = (((float)s[0] * oneover255) - 0.5f) * 2.0f;
+                n[1] = (((float)s[1] * oneover255) - 0.5f) * 2.0f;
+                n[2] = sqrtf(1.0f - (n[0] * n[0] - n[1] * n[1]));
+                n[0] *= nmapvals.scale;
+                n[1] *= nmapvals.scale;
+            }
+            else
+            {
+                du = 0; dv = 0;
+                if(!nmapvals.wrap)
+                {
+                    for(i = 0; i < num_elements; ++i)
+                        du += HEIGHT(x + kernel_du[i].x,
+                                     y + kernel_du[i].y) * kernel_du[i].w;
+                    for(i = 0; i < num_elements; ++i)
+                        dv += HEIGHT(x + kernel_dv[i].x,
+                                     y + kernel_dv[i].y) * kernel_dv[i].w;
+                }
+                else
+                {
+                    for(i = 0; i < num_elements; ++i)
+                        du += HEIGHT_WRAP(x + kernel_du[i].x,
+                                          y + kernel_du[i].y) * kernel_du[i].w;
+                    for(i = 0; i < num_elements; ++i)
+                        dv += HEIGHT_WRAP(x + kernel_dv[i].x,
+                                          y + kernel_dv[i].y) * kernel_dv[i].w;
+                }
+
+                n[0] = -du * nmapvals.scale;
+                n[1] = -dv * nmapvals.scale;
+                n[2] = 1.0f;
+            }
+
+            NORMALIZE(n);
+
+            if(n[2] < nmapvals.minz)
+            {
+                n[2] = nmapvals.minz;
+                NORMALIZE(n);
+            }
+
+            if(nmapvals.xinvert) n[0] = -n[0];
+            if(nmapvals.yinvert) n[1] = -n[1];
+            if(nmapvals.swapRGB)
+            {
+                val = n[0];
+                n[0] = n[2];
+                n[2] = val;
+            }
+
+            if(!nmapvals.dudv)
+            {
+                *d++ = (unsigned char)((n[0] + 1.0f) * 127.5f);
+                *d++ = (unsigned char)((n[1] + 1.0f) * 127.5f);
+                *d++ = (unsigned char)((n[2] + 1.0f) * 127.5f);
+
+                switch(nmapvals.alpha) {
+                    case ALPHA_NONE:
+                        *d++ = s[3];
+                        break;
+                    case ALPHA_HEIGHT:
+                        *d++ = (unsigned char) (heights[x + y * width] * 255.0f);
+                        break;
+                    case ALPHA_INVERSE_HEIGHT:
+                        *d++ = 255 - (unsigned char) (heights[x + y * width] * 255.0f);
+                        break;
+                    case ALPHA_ZERO:
+                        *d++ = 0;
+                        break;
+                    case ALPHA_ONE:
+                        *d++ = 255;
+                        break;
+                    case ALPHA_INVERT:
+                        *d++ = 255 - s[3];
+                        break;
+                    case ALPHA_MAP:
+                        *d++ = sample_alpha_map(amap, x, y, amap_w, amap_h,
+                                                width, height);
+                        break;
+                    default:
+                        *d++ = s[3];
+                        break;
+                }
+            }
+            else
+            {
+                if(nmapvals.dudv == DUDV_8BIT_SIGNED ||
+                   nmapvals.dudv == DUDV_8BIT_UNSIGNED)
+                {
+                    if(nmapvals.dudv == DUDV_8BIT_UNSIGNED)
+                    {
+                        n[0] += 1.0f;
+                        n[1] += 1.0f;
+                    }
+                    *d++ = (unsigned char)(n[0] * 127.5f);
+                    *d++ = (unsigned char)(n[1] * 127.5f);
+                    *d++ = 255;
+                    *d++ = 255;
+                }
+                else if(nmapvals.dudv == DUDV_16BIT_SIGNED ||
+                        nmapvals.dudv == DUDV_16BIT_UNSIGNED)
+                {
+                    unsigned short *d16 = (unsigned short*)d;
+                    if(nmapvals.dudv == DUDV_16BIT_UNSIGNED)
+                    {
+                        n[0] += 1.0f;
+                        n[1] += 1.0f;
+                    }
+                    *d16++ = (unsigned short)(n[0] * 32767.5f);
+                    *d16++ = (unsigned short)(n[1] * 32767.5f);
+                }
+            }
+        }
+
+    }
+
+    if(nmapvals.conversion == CONVERT_HEIGHTMAP)
+        make_heightmap(dst, width, height, bpp, nmapvals);
+
+#undef HEIGHT
+#undef HEIGHT_WRAP
+
+    free(heights);
+    free(kernel_du);
+    free(kernel_dv);
+
+    return(0);
+}
\ No newline at end of file
diff --git a/win64/src/stb_image.c b/win64/src/stb_image.c
new file mode 100644
index 0000000..148bce1
--- /dev/null
+++ b/win64/src/stb_image.c
@@ -0,0 +1,2 @@
+#define STB_IMAGE_IMPLEMENTATION
+#include <stb_image.h>
\ No newline at end of file
diff --git a/win64/src/stb_image_write.c b/win64/src/stb_image_write.c
new file mode 100644
index 0000000..2f540c3
--- /dev/null
+++ b/win64/src/stb_image_write.c
@@ -0,0 +1,2 @@
+#define STB_IMAGE_WRITE_IMPLEMENTATION
+#include <stb_image_write.h>
\ No newline at end of file

From 8b5009112f4b6e3bee9ae62d3ae0ceb19126875c Mon Sep 17 00:00:00 2001
From: "info@vltmedia.com" <info@vltmedia.com>
Date: Mon, 7 Jun 2021 17:20:39 -0400
Subject: [PATCH 37/66] Fixed Linux Github Action building

---
 CMakeLists.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index eb84658..1b5bab1 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -15,4 +15,4 @@ set(SOURCE_FILES
         include/normalmap.h include/stb_image_write.h src/stb_image_write.c)
 
 add_executable(normalmap ${SOURCE_FILES})
-# target_link_libraries(normalmap PRIVATE m)
+target_link_libraries(normalmap PRIVATE m)

From 59e48b279e9109cd49c5dce09a7c53747b246c39 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 17:27:38 -0400
Subject: [PATCH 38/66] updated Windows to build windows 64folder

---
 .github/workflows/cmake_windows.yml | 19 +++++++------------
 1 file changed, 7 insertions(+), 12 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 25c90d4..fe7180a 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -22,21 +22,16 @@ jobs:
     - uses: actions/checkout@v2
 
     - name: Configure CMake
+      working-directory: ${{github.workspace}}/win64
       # Configure CMake in a 'build' subdirectory. `CMAKE_BUILD_TYPE` is only required if you are using a single-configuration generator such as make.
       # See https://cmake.org/cmake/help/latest/variable/CMAKE_BUILD_TYPE.html?highlight=cmake_build_type
-      run: cmake -B ${{github.workspace}}/build -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}}
+      run: cmake -B ${{github.workspace}}/win64/build -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}}
 
     - name: Build
+      working-directory: ${{github.workspace}}/win64
       # Build your program with the given configuration
-      run: cmake --build ${{github.workspace}}/build --config ${{env.BUILD_TYPE}}
-
-
-    - name: CheckBuildFolder
-      working-directory: ${{github.workspace}}/build
-      # Execute tests defined by the CMake configuration.  
-      # See https://cmake.org/cmake/help/latest/manual/ctest.1.html for more detail
-      run: ls
-          
+      run: cmake --build ${{github.workspace}}/win64/build --config ${{env.BUILD_TYPE}}
+   
     - name: Get branch names
       id: branch-name
       uses: tj-actions/branch-names@v4.3
@@ -44,7 +39,7 @@ jobs:
     - name: Set env
       id: envname
       run: echo ::set-output name=pr_title::"Windows build - v0.2.0 - ${{ steps.branch-name.outputs.current_branch }}"
-        
+      
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
@@ -53,6 +48,6 @@ jobs:
         prerelease: true
         title: ${{steps.envname.outputs.pr_title}}
         files: |
-          ${{github.workspace}}/build/normalmap
+          ${{github.workspace}}/win64/build/Release/normalmap.exe
           ${{github.workspace}}/README.md
          

From 866f381952e7cae3718fec57d5671674efac417f Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 17:32:52 -0400
Subject: [PATCH 39/66] updated  path slashes

---
 .github/workflows/cmake_windows.yml | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index fe7180a..563f01f 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -38,16 +38,16 @@ jobs:
       
     - name: Set env
       id: envname
-      run: echo ::set-output name=pr_title::"Windows build - v0.2.0 - ${{ steps.branch-name.outputs.current_branch }}"
+      run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }}"
       
-    - name: Release Linux
+    - name: Release Windows
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
-        automatic_release_tag: "v0.2.0"
+        automatic_release_tag: "v0.2.0-win64"
         prerelease: true
         title: ${{steps.envname.outputs.pr_title}}
         files: |
-          ${{github.workspace}}/win64/build/Release/normalmap.exe
-          ${{github.workspace}}/README.md
+          ${{github.workspace}}\win64\build\Release\normalmap.exe
+          ${{github.workspace}}\README.md
          

From d8042cc52021736400d133d13f3cb52910e60bff Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 17:41:59 -0400
Subject: [PATCH 40/66] added hardcoded release exe

---
 .github/workflows/cmake_windows.yml | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 563f01f..d2df106 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -41,13 +41,12 @@ jobs:
       run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }}"
       
     - name: Release Windows
+    
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
         automatic_release_tag: "v0.2.0-win64"
         prerelease: true
         title: ${{steps.envname.outputs.pr_title}}
-        files: |
-          ${{github.workspace}}\win64\build\Release\normalmap.exe
-          ${{github.workspace}}\README.md
-         
+        files: D:\a\normalmap\normalmap\win64\build\Release\normalmap.exe
+          

From 435154f2bc8fe62fc26d81164138b19718cc13b3 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 17:47:04 -0400
Subject: [PATCH 41/66] added new windows build release

---
 .github/workflows/cmake_windows.yml | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index d2df106..d6e2b14 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -40,13 +40,10 @@ jobs:
       id: envname
       run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }}"
       
-    - name: Release Windows
-    
-      uses: "marvinpinto/action-automatic-releases@latest"
+    - uses: actions/checkout@v1
+    - uses: ncipollo/release-action@v1
       with:
-        repo_token: "${{ secrets.GITHUB_TOKEN }}"
-        automatic_release_tag: "v0.2.0-win64"
-        prerelease: true
-        title: ${{steps.envname.outputs.pr_title}}
-        files: D:\a\normalmap\normalmap\win64\build\Release\normalmap.exe
-          
+        artifacts: "D:/a/normalmap/normalmap/win64/build/Release/normalmap.exe"
+        bodyFile: "${{github.workspace}}/README.md"
+        token: ${{ secrets.GITHUB_TOKEN }}
+   

From 1182f08419c004640000a42232c967878a248063 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 17:49:24 -0400
Subject: [PATCH 42/66] added tag to windows release

---
 .github/workflows/cmake_windows.yml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index d6e2b14..8c64585 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -44,6 +44,7 @@ jobs:
     - uses: ncipollo/release-action@v1
       with:
         artifacts: "D:/a/normalmap/normalmap/win64/build/Release/normalmap.exe"
+        tag: "v0.2.0-win64"
         bodyFile: "${{github.workspace}}/README.md"
         token: ${{ secrets.GITHUB_TOKEN }}
    

From e7073fb0ac3aa0597fb80cdd9c5b769e0c40dcbe Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:00:14 -0400
Subject: [PATCH 43/66] updated the release with date

---
 .github/workflows/cmake_windows.yml | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 8c64585..e58afcf 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -31,7 +31,15 @@ jobs:
       working-directory: ${{github.workspace}}/win64
       # Build your program with the given configuration
       run: cmake --build ${{github.workspace}}/win64/build --config ${{env.BUILD_TYPE}}
-   
+      
+    
+    - name: Set current date as env variable
+      id: datee
+      run: echo ::set-output name=pr_title::$(date +'%Y-%m-%dT%H:%M:%S')
+      
+    - name: Echo current release
+      run: echo "v0.2.0-win64-${{steps.datee.outputs.datee}}"
+        
     - name: Get branch names
       id: branch-name
       uses: tj-actions/branch-names@v4.3
@@ -44,7 +52,7 @@ jobs:
     - uses: ncipollo/release-action@v1
       with:
         artifacts: "D:/a/normalmap/normalmap/win64/build/Release/normalmap.exe"
-        tag: "v0.2.0-win64"
+        tag: "v0.2.0-win64-${{steps.datee.outputs.datee}}"
         bodyFile: "${{github.workspace}}/README.md"
         token: ${{ secrets.GITHUB_TOKEN }}
    

From 6f7d23789fc8c3833c3d0251fc0edda593b467b3 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:04:32 -0400
Subject: [PATCH 44/66] updated current date

---
 .github/workflows/cmake_windows.yml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index e58afcf..752e4b0 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -35,7 +35,7 @@ jobs:
     
     - name: Set current date as env variable
       id: datee
-      run: echo ::set-output name=pr_title::$(date +'%Y-%m-%dT%H:%M:%S')
+      run: echo ::set-output name=currentdate::$(date +'%Y-%m-%d')
       
     - name: Echo current release
       run: echo "v0.2.0-win64-${{steps.datee.outputs.datee}}"
@@ -46,13 +46,13 @@ jobs:
       
     - name: Set env
       id: envname
-      run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }}"
+      run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }} - ${{steps.datee.outputs.currentdate}}"
       
     - uses: actions/checkout@v1
     - uses: ncipollo/release-action@v1
       with:
         artifacts: "D:/a/normalmap/normalmap/win64/build/Release/normalmap.exe"
-        tag: "v0.2.0-win64-${{steps.datee.outputs.datee}}"
+        tag: "v0.2.0-win64-${{steps.datee.outputs.currentdate}}"
         bodyFile: "${{github.workspace}}/README.md"
         token: ${{ secrets.GITHUB_TOKEN }}
    

From 6d6b895ef3d604a0f222ec25f16a52fb7c4580fc Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:19:48 -0400
Subject: [PATCH 45/66] updated windows artifact uploading

---
 .github/workflows/cmake_windows.yml | 33 +++++++++++++++++++++--------
 1 file changed, 24 insertions(+), 9 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 752e4b0..4cbc02e 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -2,9 +2,9 @@ name: CMake_Windows
 
 on:
   push:
-    branches: [ master ]
-  pull_request:
-    branches: [ master ]
+    tags:
+      - '*'
+
 
 env:
   # Customize the CMake build type here (Release, Debug, RelWithDebInfo, etc.)
@@ -36,6 +36,17 @@ jobs:
     - name: Set current date as env variable
       id: datee
       run: echo ::set-output name=currentdate::$(date +'%Y-%m-%d')
+    - name: release
+      uses: actions/create-release@v1
+      id: create_release
+      with:
+        draft: false
+        prerelease: true
+        release_name: "v0.2.0-win64-${{steps.datee.outputs.currentdate}}"
+        tag_name: ${{ github.ref }}
+        body_path: README.md
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
       
     - name: Echo current release
       run: echo "v0.2.0-win64-${{steps.datee.outputs.datee}}"
@@ -48,11 +59,15 @@ jobs:
       id: envname
       run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }} - ${{steps.datee.outputs.currentdate}}"
       
-    - uses: actions/checkout@v1
-    - uses: ncipollo/release-action@v1
+    - name: Upload Windows Artifact
+      uses: actions/upload-release-asset@v1
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
       with:
-        artifacts: "D:/a/normalmap/normalmap/win64/build/Release/normalmap.exe"
-        tag: "v0.2.0-win64-${{steps.datee.outputs.currentdate}}"
-        bodyFile: "${{github.workspace}}/README.md"
-        token: ${{ secrets.GITHUB_TOKEN }}
+        upload_url: ${{ steps.create_release.outputs.upload_url }}
+        asset_path: ./win64/build/Release/normalmap.exe
+        asset_name: normalmap.exe
+        asset_content_type: application/vnd.microsoft.portable-executable
+
+
    

From 6530338136af200e88dfad6020e5eecc241cde52 Mon Sep 17 00:00:00 2001
From: "info@vltmedia.com" <info@vltmedia.com>
Date: Mon, 7 Jun 2021 18:21:21 -0400
Subject: [PATCH 46/66] fixed formatting on cmake linx

---
 CMakeLists.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 1b5bab1..53304a6 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -15,4 +15,5 @@ set(SOURCE_FILES
         include/normalmap.h include/stb_image_write.h src/stb_image_write.c)
 
 add_executable(normalmap ${SOURCE_FILES})
+
 target_link_libraries(normalmap PRIVATE m)

From 178ae0d429e75453f58172e3d11e425691c4dd64 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:23:34 -0400
Subject: [PATCH 47/66] updated conditional push

---
 .github/workflows/cmake_windows.yml | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 4cbc02e..3719bf2 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -2,8 +2,7 @@ name: CMake_Windows
 
 on:
   push:
-    tags:
-      - '*'
+    branches: [ master ]
 
 
 env:

From aa4923d687fef84f7a5abdb6ab94f04283d68201 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:27:24 -0400
Subject: [PATCH 48/66] attempting to fix tag naming

---
 .github/workflows/cmake_windows.yml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 3719bf2..4d75821 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -41,14 +41,14 @@ jobs:
       with:
         draft: false
         prerelease: true
-        release_name: "v0.2.0-win64-${{steps.datee.outputs.currentdate}}"
+        release_name: "v0.2.0-win64-$(date +'%Y-%m-%d')"
         tag_name: ${{ github.ref }}
         body_path: README.md
       env:
         GITHUB_TOKEN: ${{ github.token }}
       
     - name: Echo current release
-      run: echo "v0.2.0-win64-${{steps.datee.outputs.datee}}"
+      run: echo "v0.2.0-win64-$(date +'%Y-%m-%d')"
         
     - name: Get branch names
       id: branch-name
@@ -56,7 +56,7 @@ jobs:
       
     - name: Set env
       id: envname
-      run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }} - ${{steps.datee.outputs.currentdate}}"
+      run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }} - $(date +'%Y-%m-%d')"
       
     - name: Upload Windows Artifact
       uses: actions/upload-release-asset@v1

From f3f89f07b49e72016c5af3b0901059f50a63b2e5 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:33:42 -0400
Subject: [PATCH 49/66] updated envname

---
 .github/workflows/cmake_windows.yml | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 4d75821..342f2bd 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -35,13 +35,23 @@ jobs:
     - name: Set current date as env variable
       id: datee
       run: echo ::set-output name=currentdate::$(date +'%Y-%m-%d')
+
+    - name: Set env
+      id: envname
+      run: echo ::set-output name=outname::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }} - ${{ steps.datee.outputs.currentdate }}"
+      
+          
+    - name: Echo current env
+      run: echo ${{ steps.envname.outputs.outname }}
+        
+      
     - name: release
       uses: actions/create-release@v1
       id: create_release
       with:
         draft: false
         prerelease: true
-        release_name: "v0.2.0-win64-$(date +'%Y-%m-%d')"
+        release_name: ${{ steps.envname.outputs.outname }}
         tag_name: ${{ github.ref }}
         body_path: README.md
       env:
@@ -54,10 +64,7 @@ jobs:
       id: branch-name
       uses: tj-actions/branch-names@v4.3
       
-    - name: Set env
-      id: envname
-      run: echo ::set-output name=pr_title::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }} - $(date +'%Y-%m-%d')"
-      
+
     - name: Upload Windows Artifact
       uses: actions/upload-release-asset@v1
       env:

From cabb3e4a65c953a206224cce984886bb6c9e2758 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:36:25 -0400
Subject: [PATCH 50/66] updated env name

---
 .github/workflows/cmake_windows.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 342f2bd..90a7b4f 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -38,7 +38,7 @@ jobs:
 
     - name: Set env
       id: envname
-      run: echo ::set-output name=outname::"Windows build - v0.2.0-win64 - ${{ steps.branch-name.outputs.current_branch }} - ${{ steps.datee.outputs.currentdate }}"
+      run: echo ::set-output name=outname::${{ steps.branch-name.outputs.current_branch }}${{ steps.datee.outputs.currentdate }}
       
           
     - name: Echo current env

From ed9128dc228448e453c423a35c32efffe7d883ca Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:40:20 -0400
Subject: [PATCH 51/66] added quotes

---
 .github/workflows/cmake_windows.yml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 90a7b4f..2c00a8f 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -34,11 +34,11 @@ jobs:
     
     - name: Set current date as env variable
       id: datee
-      run: echo ::set-output name=currentdate::$(date +'%Y-%m-%d')
+      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d')"
 
     - name: Set env
       id: envname
-      run: echo ::set-output name=outname::${{ steps.branch-name.outputs.current_branch }}${{ steps.datee.outputs.currentdate }}
+      run: echo "::set-output name=outname::${{ steps.branch-name.outputs.current_branch }}${{ steps.datee.outputs.currentdate }}"
       
           
     - name: Echo current env

From 6571e11d72b97fe29e0829a3f3399323302d0d7c Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:43:46 -0400
Subject: [PATCH 52/66] moved branch-name up

---
 .github/workflows/cmake_windows.yml | 33 ++++++++++++++---------------
 1 file changed, 16 insertions(+), 17 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 2c00a8f..4bba721 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -19,7 +19,22 @@ jobs:
 
     steps:
     - uses: actions/checkout@v2
+    - name: Get branch names
+      id: branch-name
+      uses: tj-actions/branch-names@v4.3
+       
+    - name: Set current date as env variable
+      id: datee
+      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d')"
 
+    - name: Set env
+      id: envname
+      run: echo "::set-output name=outname::${{ steps.branch-name.outputs.current_branch }}${{ steps.datee.outputs.currentdate }}"
+          
+    - name: Echo current env
+      run: echo ${{ steps.envname.outputs.outname }}
+        
+        
     - name: Configure CMake
       working-directory: ${{github.workspace}}/win64
       # Configure CMake in a 'build' subdirectory. `CMAKE_BUILD_TYPE` is only required if you are using a single-configuration generator such as make.
@@ -30,20 +45,7 @@ jobs:
       working-directory: ${{github.workspace}}/win64
       # Build your program with the given configuration
       run: cmake --build ${{github.workspace}}/win64/build --config ${{env.BUILD_TYPE}}
-      
-    
-    - name: Set current date as env variable
-      id: datee
-      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d')"
 
-    - name: Set env
-      id: envname
-      run: echo "::set-output name=outname::${{ steps.branch-name.outputs.current_branch }}${{ steps.datee.outputs.currentdate }}"
-      
-          
-    - name: Echo current env
-      run: echo ${{ steps.envname.outputs.outname }}
-        
       
     - name: release
       uses: actions/create-release@v1
@@ -60,10 +62,7 @@ jobs:
     - name: Echo current release
       run: echo "v0.2.0-win64-$(date +'%Y-%m-%d')"
         
-    - name: Get branch names
-      id: branch-name
-      uses: tj-actions/branch-names@v4.3
-      
+
 
     - name: Upload Windows Artifact
       uses: actions/upload-release-asset@v1

From f6b3e1ba9981f344ad88cf42f968d14271dca53d Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:47:14 -0400
Subject: [PATCH 53/66] updated naming for release

---
 .github/workflows/cmake_windows.yml | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 4bba721..839dd48 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -23,13 +23,21 @@ jobs:
       id: branch-name
       uses: tj-actions/branch-names@v4.3
        
+    - name: Set current version as env variable
+      id: versionnumb
+      run: echo "::set-output name=currentversion::v0.2.0"
+             
+    - name: Set current os as env variable
+      id: oschoice
+      run: echo "::set-output name=choice::windows64"
+      
     - name: Set current date as env variable
       id: datee
       run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d')"
 
     - name: Set env
       id: envname
-      run: echo "::set-output name=outname::${{ steps.branch-name.outputs.current_branch }}${{ steps.datee.outputs.currentdate }}"
+      run: echo "::set-output name=outname::${{ steps.versionnumb.outputs.currentversion }}-${{ steps.oschoice.outputs.choice }}-${{ steps.branch-name.outputs.current_branch }}-${{ steps.datee.outputs.currentdate }}"
           
     - name: Echo current env
       run: echo ${{ steps.envname.outputs.outname }}
@@ -54,7 +62,7 @@ jobs:
         draft: false
         prerelease: true
         release_name: ${{ steps.envname.outputs.outname }}
-        tag_name: ${{ github.ref }}
+        tag_name: ${{ steps.envname.outputs.outname }}
         body_path: README.md
       env:
         GITHUB_TOKEN: ${{ github.token }}

From d5941860e013308fb6bb23a0292d7956591bfe3f Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:57:53 -0400
Subject: [PATCH 54/66] set conditional to push tags only

---
 .github/workflows/cmake_windows.yml | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 839dd48..b53d3b5 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -2,7 +2,8 @@ name: CMake_Windows
 
 on:
   push:
-    branches: [ master ]
+    tags:
+      - "v*"
 
 
 env:

From aebc86af5d02f5e657adf0e031fa47e3936cb5e4 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 18:58:28 -0400
Subject: [PATCH 55/66] made conditional on tags

---
 .github/workflows/cmake.yml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index e4df81b..e0a145d 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -2,9 +2,9 @@ name: CMake
 
 on:
   push:
-    branches: [ master ]
-  pull_request:
-    branches: [ master ]
+    # Sequence of patterns matched against refs/tags
+    tags:
+      - 'v*' # Push events to matching v*, i.e. v1.0, v20.15.10
 
 env:
   # Customize the CMake build type here (Release, Debug, RelWithDebInfo, etc.)

From 9bd1a2aac5722088aee8cd07a4cfda452ff2d123 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 19:06:45 -0400
Subject: [PATCH 56/66] Updated naming convention

---
 .github/workflows/cmake.yml | 37 ++++++++++++++++++++++++++-----------
 1 file changed, 26 insertions(+), 11 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index e0a145d..9d42251 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -20,7 +20,30 @@ jobs:
 
     steps:
     - uses: actions/checkout@v2
+    - name: Set current version as env variable
+      id: versionnumb
+      run: echo "::set-output name=currentversion::v0.2.0"
+                  
+    - name: Get branch names
+      id: branch-name
+      uses: tj-actions/branch-names@v4.3
+     
+    - name: Set current os as env variable
+      id: oschoice
+      run: echo "::set-output name=choice::windows64"
+      
+    - name: Set current date as env variable
+      id: datee
+      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d')"
 
+    - name: Set env
+      id: envname
+      run: echo "::set-output name=outname::${{ steps.branch-name.outputs.current_branch }}-${{ steps.oschoice.outputs.choice }}-${{ steps.datee.outputs.currentdate }}"
+          
+    - name: Echo current env
+      run: echo ${{ steps.envname.outputs.outname }}
+        
+        
     - name: Configure CMake
       # Configure CMake in a 'build' subdirectory. `CMAKE_BUILD_TYPE` is only required if you are using a single-configuration generator such as make.
       # See https://cmake.org/cmake/help/latest/variable/CMAKE_BUILD_TYPE.html?highlight=cmake_build_type
@@ -42,22 +65,14 @@ jobs:
       with:
         tagRegex: "v*"  # Optional. Returns specified group text as tag name. Full tag string is returned if regex is not defined.
         tagRegexGroup: 1 # Optional. Default is 1.
-          
-    - name: Get branch names
-      id: branch-name
-      uses: tj-actions/branch-names@v4.3
-      
-    - name: Set env
-      id: envname
-      run: echo ::set-output name=pr_title::"Linux build - v0.2.0 - ${{ steps.branch-name.outputs.current_branch }}"
-        
+
     - name: Release Linux
       uses: "marvinpinto/action-automatic-releases@latest"
       with:
         repo_token: "${{ secrets.GITHUB_TOKEN }}"
-        automatic_release_tag: "v0.2.0"
+        automatic_release_tag: ${{ steps.envname.outputs.outname }}
         prerelease: true
-        title: ${{steps.envname.outputs.pr_title}}
+        title: ${{ steps.envname.outputs.outname }}
         files: |
           ${{github.workspace}}/build/normalmap
           ${{github.workspace}}/README.md

From fa872ba1921910812ac89921064f1de027646089 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 19:06:59 -0400
Subject: [PATCH 57/66] Updated naming convention

---
 .github/workflows/cmake_windows.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index b53d3b5..24005ab 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -38,7 +38,7 @@ jobs:
 
     - name: Set env
       id: envname
-      run: echo "::set-output name=outname::${{ steps.versionnumb.outputs.currentversion }}-${{ steps.oschoice.outputs.choice }}-${{ steps.branch-name.outputs.current_branch }}-${{ steps.datee.outputs.currentdate }}"
+      run: echo "::set-output name=outname::${{ steps.branch-name.outputs.current_branch }}-${{ steps.oschoice.outputs.choice }}-${{ steps.datee.outputs.currentdate }}"
           
     - name: Echo current env
       run: echo ${{ steps.envname.outputs.outname }}

From 25e2b07e8a49ad37af2d5d09e5d3c5c4f0e72c46 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 19:18:23 -0400
Subject: [PATCH 58/66] updated naming to add seconds and remove windows

---
 .github/workflows/cmake.yml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index 9d42251..b5fb649 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -30,11 +30,11 @@ jobs:
      
     - name: Set current os as env variable
       id: oschoice
-      run: echo "::set-output name=choice::windows64"
+      run: echo "::set-output name=choice::linux"
       
     - name: Set current date as env variable
       id: datee
-      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d')"
+      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d%_%H%M%S')"
 
     - name: Set env
       id: envname

From 1eccb752cc530188e5335279e70a2e3824fc6add Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 19:18:51 -0400
Subject: [PATCH 59/66] updated time naming convention

---
 .github/workflows/cmake_windows.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 24005ab..cf91717 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -34,7 +34,7 @@ jobs:
       
     - name: Set current date as env variable
       id: datee
-      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d')"
+      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d%_%H%M%S')"
 
     - name: Set env
       id: envname

From 67d5815487dda1ba33dac60e089c21be0349bf92 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 19:24:52 -0400
Subject: [PATCH 60/66] fixed datetime bug

---
 .github/workflows/cmake_windows.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index cf91717..8700e22 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -34,7 +34,7 @@ jobs:
       
     - name: Set current date as env variable
       id: datee
-      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d%_%H%M%S')"
+      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d_%H%M%S')"
 
     - name: Set env
       id: envname

From 91646fca34c2a3debc0c3feb8a2d4b72b734a2b0 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 19:25:05 -0400
Subject: [PATCH 61/66] fixed datetime bug

---
 .github/workflows/cmake.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index b5fb649..a4d0e6e 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -34,7 +34,7 @@ jobs:
       
     - name: Set current date as env variable
       id: datee
-      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d%_%H%M%S')"
+      run: echo "::set-output name=currentdate::$(date +'%Y-%m-%d_%H%M%S')"
 
     - name: Set env
       id: envname

From 0f4b6b65662d2669a3b9194bad7ba480efa47637 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 19:47:21 -0400
Subject: [PATCH 62/66] Added Changelog and added that to releases

---
 .github/workflows/cmake.yml | 41 ++++++++++++++++++++++++++++---------
 1 file changed, 31 insertions(+), 10 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index a4d0e6e..fc605b0 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -65,15 +65,36 @@ jobs:
       with:
         tagRegex: "v*"  # Optional. Returns specified group text as tag name. Full tag string is returned if regex is not defined.
         tagRegexGroup: 1 # Optional. Default is 1.
-
-    - name: Release Linux
-      uses: "marvinpinto/action-automatic-releases@latest"
+        
+    - name: Generate Changelog
+      id: changeloggen
+      uses: heinrichreimer/github-changelog-generator-action@v2.1.1
+      with:
+        token: ${{ secrets.GITHUB_TOKEN }}
+        
+        
+        
+    - name: release
+      uses: actions/create-release@v1
+      id: create_release
       with:
-        repo_token: "${{ secrets.GITHUB_TOKEN }}"
-        automatic_release_tag: ${{ steps.envname.outputs.outname }}
+        draft: false
         prerelease: true
-        title: ${{ steps.envname.outputs.outname }}
-        files: |
-          ${{github.workspace}}/build/normalmap
-          ${{github.workspace}}/README.md
-         
+        release_name: ${{ steps.envname.outputs.outname }}
+        tag_name: ${{ steps.envname.outputs.outname }}
+        body: ${{ steps.changeloggen.outputs.changelog }}
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+
+
+    - name: upload linux artifact
+      uses: actions/upload-release-asset@v1
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+      with:
+        upload_url: ${{ steps.create_release.outputs.upload_url }}
+        asset_path: ./build/normalmap
+        asset_name: normalmap
+        asset_content_type: application/x-executable
+      
+      

From 07200dc6881492c472e36c424f8d62feca4c6a8b Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 19:47:35 -0400
Subject: [PATCH 63/66] Added Changelog and added that to releases

---
 .github/workflows/cmake_windows.yml | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 8700e22..78b314a 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -56,6 +56,14 @@ jobs:
       run: cmake --build ${{github.workspace}}/win64/build --config ${{env.BUILD_TYPE}}
 
       
+    - name: Generate Changelog
+      id: changeloggen
+      uses: heinrichreimer/github-changelog-generator-action@v2.1.1
+      with:
+        token: ${{ secrets.GITHUB_TOKEN }}
+        
+        
+        
     - name: release
       uses: actions/create-release@v1
       id: create_release
@@ -64,7 +72,7 @@ jobs:
         prerelease: true
         release_name: ${{ steps.envname.outputs.outname }}
         tag_name: ${{ steps.envname.outputs.outname }}
-        body_path: README.md
+        body: ${{ steps.changeloggen.outputs.changelog }}
       env:
         GITHUB_TOKEN: ${{ github.token }}
       

From 274d5881e00a6517c7a6714275e80f6299711cb9 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 20:08:12 -0400
Subject: [PATCH 64/66] removed Generate changelog

---
 .github/workflows/cmake_windows.yml | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/.github/workflows/cmake_windows.yml b/.github/workflows/cmake_windows.yml
index 78b314a..e744225 100644
--- a/.github/workflows/cmake_windows.yml
+++ b/.github/workflows/cmake_windows.yml
@@ -56,13 +56,7 @@ jobs:
       run: cmake --build ${{github.workspace}}/win64/build --config ${{env.BUILD_TYPE}}
 
       
-    - name: Generate Changelog
-      id: changeloggen
-      uses: heinrichreimer/github-changelog-generator-action@v2.1.1
-      with:
-        token: ${{ secrets.GITHUB_TOKEN }}
-        
-        
+
         
     - name: release
       uses: actions/create-release@v1
@@ -72,7 +66,7 @@ jobs:
         prerelease: true
         release_name: ${{ steps.envname.outputs.outname }}
         tag_name: ${{ steps.envname.outputs.outname }}
-        body: ${{ steps.changeloggen.outputs.changelog }}
+        body_path: ${{github.workspace}}/CHANGELOG.md
       env:
         GITHUB_TOKEN: ${{ github.token }}
       

From 4267cd7f6edfb01a7989d80e11940484ebf64fb3 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 20:08:45 -0400
Subject: [PATCH 65/66] removed Changelog

---
 .github/workflows/cmake.yml | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/.github/workflows/cmake.yml b/.github/workflows/cmake.yml
index fc605b0..660b2cf 100644
--- a/.github/workflows/cmake.yml
+++ b/.github/workflows/cmake.yml
@@ -66,13 +66,7 @@ jobs:
         tagRegex: "v*"  # Optional. Returns specified group text as tag name. Full tag string is returned if regex is not defined.
         tagRegexGroup: 1 # Optional. Default is 1.
         
-    - name: Generate Changelog
-      id: changeloggen
-      uses: heinrichreimer/github-changelog-generator-action@v2.1.1
-      with:
-        token: ${{ secrets.GITHUB_TOKEN }}
-        
-        
+
         
     - name: release
       uses: actions/create-release@v1
@@ -82,7 +76,7 @@ jobs:
         prerelease: true
         release_name: ${{ steps.envname.outputs.outname }}
         tag_name: ${{ steps.envname.outputs.outname }}
-        body: ${{ steps.changeloggen.outputs.changelog }}
+        body_path: ${{github.workspace}}/CHANGELOG.md
       env:
         GITHUB_TOKEN: ${{ github.token }}
 

From 92bc2d3e1535fcd0d7f78efa8e400b084930ccd9 Mon Sep 17 00:00:00 2001
From: VLT Media <39071455+vltmedia@users.noreply.github.com>
Date: Mon, 7 Jun 2021 20:09:20 -0400
Subject: [PATCH 66/66] added changelog

---
 CHANGELOG.md | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 CHANGELOG.md

diff --git a/CHANGELOG.md b/CHANGELOG.md
new file mode 100644
index 0000000..f17833c
--- /dev/null
+++ b/CHANGELOG.md
@@ -0,0 +1,22 @@
+# Changelog
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Released]
+
+## [0.1.0a] - 2017-01-07
+- Initial original version
+
+## [0.2.0] - 2021-06-07
+### Added
+- win64 source folder for building win64 separately.
+- Github Actions for automatically building Windows and Linux versions every new tag
+- This Changelog
+
+### Fixed
+
+- linux CMake building
+- win64 CMake building
+