Hello, I am currently playing a bit with texture splatting and it works quite well with 2 textures. However I would like to use 8 or 16 possible textures - and I wonder what impact this will have on performance. Is there overhead "in general", or only if I do texture2D lookups? Is it better to branch unescessary texture2D-calls away on PS3-capable hardware or is the branch even more expensive than texture fetches? Thanks, lg Clemens

Benchmark! :D

Seriously though, it varies between types of hardware. In general, more recent hardware will handle the dynamic branching better, but it's impossible to say that one approach or another will definitely be faster.

From my benchmarks, I've found that more texture lookups are slower than dynamic branching. (Soft shadow mapping, multiple samples)