Currenly Im working on PVR texture image codec module and there is a little problem.
PVR may contain 16bit and 32bit textures with different sub-pixel formats: 5551 4444 565 and others.
However, render modules can "eat" only RGBA8888 \ RGB888 pixel formats, there is simply no apropritate argument in OpenGLTexture::loadFromMemory for sub-pixel format. Cross-render-API unification??
Possible solution is to do conversion from "any sub-pixel format" to RBGA8888 in place. But it's not the best way, since there also will be compressed texture formats which must stay compressed (GPU supported), and as im going to write my own OpenGL ES render module, i want my OpenGLES_Texture::loadFromMemory to know if a pointed block of memory is actually compressed texture or not. (And simply i dont know how to do that conversion
What im trying to say, maybe if you could make texture loading more flexible somehow. What i need is sub-pixel format definition and more about pixel-format (compression format).
Donot know about DX9\10 but OpenGL can make textures with all kinds of sub-pixel formats.
PS How DX fans is feeling without their DXTC? =)

