aw2 & ffo - sop fix now stable

CyberXeSS/CyberXeSS.vcxproj

@@ -82,7 +82,7 @@
     <IncludePath>$(SolutionDir)external\vulkan\include;$(SolutionDir)external\nvngx_dlss_sdk;$(SolutionDir)external\xess\inc\xess;$(SolutionDir)external\simpleini;$(SolutionDir)external\unordered_dense\include;$(SolutionDir)external\spdlog\include;$(IncludePath)</IncludePath>
     <LibraryPath>$(SolutionDir)CyberXess\cas\lib;$(SolutionDir)CyberXess\fsr2\lib;$(SolutionDir)CyberXess\fsr2_212\lib;$(SolutionDir)CyberXess\vulkan;$(SolutionDir)external\xess\lib;$(SolutionDir)CyberXess\d3dx;$(LibraryPath)</LibraryPath>
     <TargetName>nvngx</TargetName>
-    <OutDir>D:\Folders\Games\Deep Rock Galactic\FSD\Binaries\Win64\</OutDir>
+    <OutDir>F:\Games\Alan Wake 2\</OutDir>
     <IntDir>.\x64\Debug</IntDir>
   </PropertyGroup>
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">

CyberXeSS/NVNGX_DLSS_Dx11.cpp

@@ -289,11 +289,9 @@ NVSDK_NGX_API NVSDK_NGX_Result NVSDK_NGX_D3D11_EvaluateFeature(ID3D11DeviceConte
 			Dx11Contexts[handleId].reset();
 			auto it = std::find_if(Dx11Contexts.begin(), Dx11Contexts.end(), [&handleId](const auto& p) { return p.first == handleId; });
 			Dx11Contexts.erase(it);
-
-			return NVSDK_NGX_Result_Success;
 		}
 
-		// next frame prepare stuff
+		// next prepare stuff
 		if (Config::Instance()->newBackend == "xess")
 		{
 			Config::Instance()->Dx11Upscaler = "xess";
@@ -317,9 +315,8 @@ NVSDK_NGX_API NVSDK_NGX_Result NVSDK_NGX_D3D11_EvaluateFeature(ID3D11DeviceConte
 
 		if (Dx11Contexts[InFeatureHandle->Id].get()->Init(D3D11Device, InDevCtx, InParameters))
 		{
-			// next frame we will start rendering again
 			Config::Instance()->changeBackend = false;
-			return NVSDK_NGX_Result_Success;
+			//return NVSDK_NGX_Result_Success;
 		}
 		else
 			return NVSDK_NGX_Result_Fail;

CyberXeSS/NVNGX_DLSS_Dx12.cpp

@@ -329,7 +329,7 @@ NVSDK_NGX_API NVSDK_NGX_Result NVSDK_NGX_D3D12_EvaluateFeature(ID3D12GraphicsCom
 			auto it = std::find_if(Dx12Contexts.begin(), Dx12Contexts.end(), [&handleId](const auto& p) { return p.first == handleId; });
 			Dx12Contexts.erase(it);
 
-			return NVSDK_NGX_Result_Success;
+			//return NVSDK_NGX_Result_Success;
 		}
 
 		// next frame prepare stuff
@@ -353,7 +353,7 @@ NVSDK_NGX_API NVSDK_NGX_Result NVSDK_NGX_D3D12_EvaluateFeature(ID3D12GraphicsCom
 		{
 			// next frame we will start rendering again
 			Config::Instance()->changeBackend = false;
-			return NVSDK_NGX_Result_Success;
+			//return NVSDK_NGX_Result_Success;
 		}
 		else
 			return NVSDK_NGX_Result_Fail;

CyberXeSS/backends/xess/XeSSFeature.cpp

@@ -13,7 +13,7 @@ RWTexture2D<float4> OutputTexture : register(u0);
 void main(uint3 DTid : SV_DispatchThreadID)
 {
 	float4 pixel = InputTexture[DTid.xy];
-	pixel.rgb *= 10.0;
+	pixel.rgb *= 125.0;
 	OutputTexture[DTid.xy] = pixel;
 })";
 
@@ -26,7 +26,7 @@ RWTexture2D<float4> OutputTexture : register(u0);
 void main(uint3 DTid : SV_DispatchThreadID)
 {
 	float4 pixel = InputTexture[DTid.xy];
-	pixel.rgb *= 0.1;
+	pixel.rgb *= 0.008;
 	OutputTexture[DTid.xy] = pixel;
 })";
 
@@ -247,3 +247,48 @@ float XeSSFeature::GetSharpness(const NVSDK_NGX_Parameter* InParameters)
 
 	return sharpness;
 }
+
+bool XeSSFeature::CreateBufferResource(ID3D12Device* InDevice, ID3D12Resource* InSource, ID3D12Resource** OutDest, D3D12_RESOURCE_STATES InDestState)
+{
+	if (InDevice == nullptr || InSource == nullptr)
+		return false;
+
+	D3D12_RESOURCE_DESC texDesc = InSource->GetDesc();
+
+	if (*OutDest != nullptr)
+	{
+		auto bufDesc = (*OutDest)->GetDesc();
+
+		if (bufDesc.Width != texDesc.Width || bufDesc.Height != texDesc.Height || bufDesc.Format != texDesc.Format)
+		{
+			(*OutDest)->Release();
+			*OutDest = nullptr;
+		}
+		else
+			return true;
+	}
+
+	spdlog::debug("XeSSFeature::CreateCasBufferResource Start!");
+
+	D3D12_HEAP_PROPERTIES heapProperties;
+	D3D12_HEAP_FLAGS heapFlags;
+	HRESULT hr = InSource->GetHeapProperties(&heapProperties, &heapFlags);
+
+	if (hr != S_OK)
+	{
+		spdlog::error("XeSSFeature::CreateBufferResource GetHeapProperties result: {0:x}", hr);
+		return false;
+	}
+
+	texDesc.Flags |= D3D12_RESOURCE_FLAG_ALLOW_RENDER_TARGET | D3D12_RESOURCE_FLAG_ALLOW_UNORDERED_ACCESS | D3D12_RESOURCE_FLAG_ALLOW_SIMULTANEOUS_ACCESS;
+
+	hr = InDevice->CreateCommittedResource(&heapProperties, D3D12_HEAP_FLAG_NONE, &texDesc, InDestState, nullptr, IID_PPV_ARGS(OutDest));
+
+	if (hr != S_OK)
+	{
+		spdlog::error("XeSSFeature::CreateBufferResource CreateCommittedResource result: {0:x}", hr);
+		return false;
+	}
+
+	return true;
+}

CyberXeSS/backends/xess/XeSSFeature.h

@@ -42,6 +42,7 @@ class XeSSFeature : public virtual IFeature
 
 	bool InitXeSS(ID3D12Device* device, const NVSDK_NGX_Parameter* InParameters);
 	float GetSharpness(const NVSDK_NGX_Parameter* InParameters);
+	bool CreateBufferResource(ID3D12Device* InDevice, ID3D12Resource* InSource, ID3D12Resource** OutDest, D3D12_RESOURCE_STATES InDestState);
 
 public:
 

CyberXeSS/backends/xess/XeSSFeature_Dx11.cpp

@@ -119,7 +119,7 @@ bool XeSSFeatureDx11::Evaluate(ID3D11DeviceContext* InDeviceContext, const NVSDK
 
 	if (Config::Instance()->CasEnabled.value_or(true) && sharpness > 0.0f)
 	{
-		if (CAS->Buffer() == nullptr && !CAS->CreateBufferResource(Dx12on11Device, dx11Out.Dx12Resource, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
+		if (!CAS->CreateBufferResource(Dx12on11Device, dx11Out.Dx12Resource, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
 		{
 			spdlog::error("XeSSFeatureDx12::Evaluate Can't create cas buffer!");
 
@@ -138,7 +138,10 @@ bool XeSSFeatureDx11::Evaluate(ID3D11DeviceContext* InDeviceContext, const NVSDK
 	else
 	{
 		if (Config::Instance()->ColorSpaceFix.value_or(false) && OutputEncode->CreateBufferResource(Dx12on11Device, dx11Out.Dx12Resource, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
+		{
+			OutputEncode->SetBufferState(Dx12CommandList, D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
 			params.pOutputTexture = OutputEncode->Buffer();
+		}
 		else
 			params.pOutputTexture = dx11Out.Dx12Resource;
 	}
@@ -193,6 +196,8 @@ bool XeSSFeatureDx11::Evaluate(ID3D11DeviceContext* InDeviceContext, const NVSDK
 
 		if (Config::Instance()->ColorSpaceFix.value_or(false) && OutputEncode->CreateBufferResource(Dx12on11Device, params.pOutputTexture, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
 		{
+			OutputEncode->SetBufferState(Dx12CommandList, D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
+
 			if (!CAS->Dispatch(Dx12CommandList, sharpness, params.pOutputTexture, OutputEncode->Buffer()))
 			{
 				Config::Instance()->CasEnabled = false;
@@ -209,6 +214,8 @@ bool XeSSFeatureDx11::Evaluate(ID3D11DeviceContext* InDeviceContext, const NVSDK
 		}
 		else
 		{
+			CAS->SetBufferState(Dx12CommandList, D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE);
+
 			if (!CAS->Dispatch(Dx12CommandList, sharpness, CAS->Buffer(), dx11Out.Dx12Resource))
 			{
 				Config::Instance()->CasEnabled = false;
@@ -228,18 +235,26 @@ bool XeSSFeatureDx11::Evaluate(ID3D11DeviceContext* InDeviceContext, const NVSDK
 	if (OutputEncode->Buffer() && Config::Instance()->ColorSpaceFix.value_or(false))
 	{
 		OutputEncode->SetBufferState(Dx12CommandList, D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE);
-		OutputEncode->Dispatch(Dx12on11Device, Dx12CommandList, OutputEncode->Buffer(), dx11Out.Dx12Resource, 16, 16);
 
-#if _DEBUG
-		D3DX11SaveTextureToFile(InDeviceContext, dx11Out.SharedTexture, D3DX11_IFF_JPG, L"D:\\aaaa.jpg");
-#endif // _DEBUG
+		if (CreateBufferResource(Dx12on11Device, dx11Out.Dx12Resource, &_outBuffer, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
+		{
+			OutputEncode->Dispatch(Dx12on11Device, Dx12CommandList, OutputEncode->Buffer(), _outBuffer, 16, 16);
 
-	}
+			ResourceBarrier(Dx12CommandList, _outBuffer, D3D12_RESOURCE_STATE_UNORDERED_ACCESS, D3D12_RESOURCE_STATE_COPY_SOURCE);
+			ResourceBarrier(Dx12CommandList, dx11Out.Dx12Resource, D3D12_RESOURCE_STATE_UNORDERED_ACCESS, D3D12_RESOURCE_STATE_COPY_DEST);
 
-	// Execute dx12 commands to process xess
-	Dx12CommandList->Close();
-	ID3D12CommandList* ppCommandLists[] = { Dx12CommandList };
-	Dx12CommandQueue->ExecuteCommandLists(1, ppCommandLists);
+			Dx12CommandList->CopyResource(dx11Out.Dx12Resource, _outBuffer);
+		}
+		else
+		{
+			OutputEncode->SetBufferState(Dx12CommandList, D3D12_RESOURCE_STATE_COPY_SOURCE);
+			ResourceBarrier(Dx12CommandList, dx11Out.Dx12Resource, D3D12_RESOURCE_STATE_UNORDERED_ACCESS, D3D12_RESOURCE_STATE_COPY_DEST);
+
+			Dx12CommandList->CopyResource(dx11Out.Dx12Resource, OutputEncode->Buffer());
+		}
+
+		ResourceBarrier(Dx12CommandList, dx11Out.Dx12Resource, D3D12_RESOURCE_STATE_COPY_DEST, D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
+	}
 
 	if (!CopyBackOutput())
 	{
@@ -251,6 +266,11 @@ bool XeSSFeatureDx11::Evaluate(ID3D11DeviceContext* InDeviceContext, const NVSDK
 		return false;
 	}
 
+	// Execute dx12 commands to process xess
+	Dx12CommandList->Close();
+	ID3D12CommandList* ppCommandLists[] = { Dx12CommandList };
+	Dx12CommandQueue->ExecuteCommandLists(1, ppCommandLists);
+
 	// imgui
 	if (Imgui)
 	{
@@ -270,6 +290,10 @@ bool XeSSFeatureDx11::Evaluate(ID3D11DeviceContext* InDeviceContext, const NVSDK
 
 XeSSFeatureDx11::~XeSSFeatureDx11()
 {
-	spdlog::debug("XeSSFeatureDx11::XeSSFeatureDx11");
+	if (_outBuffer != nullptr)
+	{
+		_outBuffer->Release();
+		_outBuffer = nullptr;
+	}
 }
 

CyberXeSS/backends/xess/XeSSFeature_Dx11.h

@@ -5,6 +5,9 @@
 
 class XeSSFeatureDx11 : public XeSSFeature, public IFeature_Dx11wDx12
 {
+private:
+	ID3D12Resource* _outBuffer;
+
 protected:
 
 public:

CyberXeSS/backends/xess/XeSSFeature_Dx12.cpp

@@ -111,26 +111,23 @@ bool XeSSFeatureDx12::Evaluate(ID3D12GraphicsCommandList* InCommandList, const N
 		spdlog::debug("XeSSFeatureDx12::Evaluate Output exist..");
 
 		if (Config::Instance()->OutputResourceBarrier.has_value())
-			ResourceBarrier(InCommandList, paramOutput,
-				(D3D12_RESOURCE_STATES)Config::Instance()->OutputResourceBarrier.value(),
-				D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
-
-		if (Config::Instance()->CasEnabled.value_or(true) && sharpness > 0.0f)
 		{
-			if (CAS->Buffer() == nullptr && !CAS->CreateBufferResource(Device, paramOutput, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
-			{
-				spdlog::error("XeSSFeatureDx12::Evaluate Can't create cas buffer!");
-				return false;
-			}
+			ResourceBarrier(InCommandList, paramOutput, (D3D12_RESOURCE_STATES)Config::Instance()->OutputResourceBarrier.value(), D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
+		}
 
+		if (Config::Instance()->CasEnabled.value_or(true) && sharpness > 0.0f && CAS->CreateBufferResource(Device, paramOutput, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
+		{
+			CAS->SetBufferState(InCommandList, D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
 			params.pOutputTexture = CAS->Buffer();
 		}
+		else if (Config::Instance()->ColorSpaceFix.value_or(false) && OutputEncode->CreateBufferResource(Device, paramOutput, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
+		{
+			OutputEncode->SetBufferState(InCommandList, D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
+			params.pOutputTexture = OutputEncode->Buffer();
+		}
 		else
 		{
-			if (Config::Instance()->ColorSpaceFix.value_or(false) && OutputEncode->CreateBufferResource(Device, paramOutput, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
-				params.pOutputTexture = OutputEncode->Buffer();
-			else
-				params.pOutputTexture = paramOutput;
+			params.pOutputTexture = paramOutput;
 		}
 	}
 	else
@@ -229,12 +226,14 @@ bool XeSSFeatureDx12::Evaluate(ID3D12GraphicsCommandList* InCommandList, const N
 	}
 
 	//apply cas
-	if (Config::Instance()->CasEnabled.value_or(true) && sharpness > 0.0f)
+	if (Config::Instance()->CasEnabled.value_or(true) && sharpness > 0.0f && CAS->Buffer() != nullptr)
 	{
 		ResourceBarrier(InCommandList, params.pOutputTexture, D3D12_RESOURCE_STATE_UNORDERED_ACCESS, D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE);
 
 		if (Config::Instance()->ColorSpaceFix.value_or(false) && OutputEncode->CreateBufferResource(Device, params.pOutputTexture, D3D12_RESOURCE_STATE_UNORDERED_ACCESS))
 		{
+			OutputEncode->SetBufferState(InCommandList, D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
+
 			if (!CAS->Dispatch(InCommandList, sharpness, params.pOutputTexture, OutputEncode->Buffer()))
 			{
 				Config::Instance()->CasEnabled = false;
@@ -243,6 +242,8 @@ bool XeSSFeatureDx12::Evaluate(ID3D12GraphicsCommandList* InCommandList, const N
 		}
 		else
 		{
+			CAS->SetBufferState(InCommandList, D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE);
+
 			if (!CAS->Dispatch(InCommandList, sharpness, params.pOutputTexture, paramOutput))
 			{
 				Config::Instance()->CasEnabled = false;

CyberXeSS/cas/CAS_Dx12.cpp

@@ -100,7 +100,6 @@ inline static FfxCas::FfxSurfaceFormat ffxGetSurfaceFormatDX12(DXGI_FORMAT forma
 		return FfxCas::FFX_SURFACE_FORMAT_R16_UNORM;
 	case (DXGI_FORMAT_R16_SNORM):
 		return FfxCas::FFX_SURFACE_FORMAT_R16_SNORM;
-		//case DXGI_FORMAT_R16_SINT:
 
 	case DXGI_FORMAT_R8_TYPELESS:
 	case DXGI_FORMAT_R8_UNORM:
@@ -225,6 +224,9 @@ bool CAS_Dx12::Dispatch(ID3D12CommandList* InCommandList, float InSharpness, ID3
 
 void CAS_Dx12::SetBufferState(ID3D12GraphicsCommandList* InCommandList, D3D12_RESOURCE_STATES InState)
 {
+	if (_bufferState == InState)
+		return;
+
 	D3D12_RESOURCE_BARRIER barrier = {};
 	barrier.Type = D3D12_RESOURCE_BARRIER_TYPE_TRANSITION;
 	barrier.Transition.pResource = _buffer;