Texture & pixel format

Hi,

If my OpenGL application is running in a 16bpp pixel format, I would create my textures in a 16bpp format, e.g. GL_RGBA4, to not hog the bandwidth by moving texture data I won’t be using.

Similarly for a 32bpp pixel format, I would create my textures in a 32bpp format.

If I know in advance that the image I’m reading from contains all opaque pixels, is it in the interests of performance to pad out the alpha channel to keep the same bit-depth, or to reduce the bit-depth of my textures, say to 24bpp?

So, for example, in a 32-bpp pixel display:

opaque texture -> GL_RGB8 or GL_RGBA8?

Cheers,

Mark

Dont specify an alpha’d format if you have no alpha channel. RGB8 should be fine.

Also, RGB5 is likely to look better than RGBA4.