[rtextures] LoadImageDataNormalized() broken for some pixel formats due to missing increment

[henrikglass]

• [ YES] I checked there is no similar issue already reported
• [ YES] I checked the documentation on the wiki
• [ YES] My code has no errors or misuse of raylib
• [ TECNICALLY YES] I tested it on latest raylib version from master branch

Issue description

I was experimenting a bit with GenMeshHeightmap and noticed that it simply refused to work for Images with the pixelformat PIXELFORMAT_UNCOMPRESSED_R32. Long story short, I tracked it down:

LoadImageDataNormalized in rtextures.c is broken for both PIXELFORMAT_UNCOMPRESSED_R32 and PIXELFORMAT_UNCOMPRESSED_R16 due to k not being incremented for each pixel. See snippet below:

case PIXELFORMAT_UNCOMPRESSED_R32:
{
    pixels[i].x = ((float *)image.data)[k];
    pixels[i].y = 0.0f;
    pixels[i].z = 0.0f;
    pixels[i].w = 1.0f;
} break;

     /* ... */

case PIXELFORMAT_UNCOMPRESSED_R16:
{
    pixels[i].x = HalfToFloat(((unsigned short *)image.data)[k]);
    pixels[i].y = 0.0f;
    pixels[i].z = 0.0f;
    pixels[i].w = 1.0f;
} break;

This should be:

case PIXELFORMAT_UNCOMPRESSED_R32:
{
    pixels[i].x = ((float *)image.data)[k];
    pixels[i].y = 0.0f;
    pixels[i].z = 0.0f;
    pixels[i].w = 1.0f;
    k += 1;
} break;

     /* ... */

case PIXELFORMAT_UNCOMPRESSED_R16:
{
    pixels[i].x = HalfToFloat(((unsigned short *)image.data)[k]);
    pixels[i].y = 0.0f;
    pixels[i].z = 0.0f;
    pixels[i].w = 1.0f;
    k += 1;
} break;

I made these changes to the source of raylib 5.0* and tested it with my own little raylib project. It seemed to produce the result I expected. However, with the latest version of raylib (master), I was unable to get my program to run even without the changes. My program would segfault when calling DrawMesh. I don't know what that is about. I'll leave that one for someone else to figure out.

I might make a pull request if I can make the latest version of raylib work, but if anyone else is tempted to do so before me please go ahead.

* I also had fix LoadImageColors, which had the same problem. It seems to be fixed in the latest version (master).

[henrikglass]

I did a very quick test. 53ea275 still causes my program to crash. I'm about to leave my computer, but I'll see if I can reproduce the problem later on a different machine, and, if so, produce a minimal example.

[raysan5]

@henrikglass Please, could you send a PR with the review? Thanks!

[henrikglass]

Good news, I couldn't replicate the crash on my home computer, which probably means I originally did something silly.

I'll post a PR later today with the fixes.