Post Sun Oct 14, 2007 3:05 am

Reduced faces, reduced render time?

This may seem like an obvious question, but I want to make sure.

I know KT uses an octtree to subdivide geometry to speed up rendering. I know how oct or quad tree structures work (having used them myself). So I have 2 questions.
1) Does KT's octtree divide up objects or faces? If it's faces, then conceivably you could improve your rendering by subdividing your objects so ray collision checks could be narrowed down better. If it's only objects then the fewer the faces is almost certainly better.

2) If the octtree does subdivide faces, if I have a very large plane (like a floor) is a single face (or 2 triangles I guess) best or is there an advantage to breaking it up a little?

Conventional wisdom says the fewer faces the better, but experience tells me that it's at least possible to gain a time advantage with a little subdivision.

I don't know if anyone other than giannis can answer this question, but what the heck....