Thought maybe someone here could shed light on this.
I recall reading that graphics cards are better at working with one large image (256x256) rather than say 16 64x64 images. So if I were making a model I’d be better off using one larger tex than several smaller ones.
Reason being, in my particular instance it would be easier to have numerous smaller texs but if there could be a noticable reduction in performance I will use the alternative.
Definitely use one large texture rather than several small ones in this case. The renderer batches stuff up by texture/shader to minimize the number of GL calls.
If the entire model is under 1000 verts (2000 tris), then it’ll render in one batch. If there were 16 different textures, then it would take 16 batches to render.
y
Sort of a tangent, but doesn’t the size of a texture, such as 1024x1024 vs. 512x512 only make a difference in performance if you don’t have enough memory to load the big ones? I recall having to lower texture detail on my old TNT2 to raise framerates, but on my GF3 Ti 200, texture detail is almost meaningless in most of the games I play.
On a purely rendering point - the smaller the texture, the faster it will be. If performance really means that much to you for your map, then you could use lower resolution textures, but people with high end graphics cards won’t appreciate it so much. Persoanlly, I use high res textures - after all, that’s what the low quality texture option is there for.