Originally Posted by MasterZap
True. I would welcome a compelling case for adding them [would help me lobby for adding them internally]. Do you have samples where you see a clear benefit of using these?
In order for IP to be attractive enough to be implemented, two major bugs would have to be fixed:
1, Irradiance particles are not capable of interaction with specular component of material, especially in case of refraction, this is a dealbreaker as a glass planes in windows make your room completely black.
2, Irradiance particle collection is extremely slow and inefficient for some reason. With the power of modern day GPUs, i struggle to believe that collection of any irradiance data, no matter how the containing entities are implemented, would take THAT long. Especially when there are times during IP precalc phase where my CPU runs on 8-10% of it's capacity.
Now we all know that nVidia is a company selling GPUs, so their main intention is to sell as many GPUs as possible, and if they started to make a good efficient CPU renderer, they would be against themselves. They just need to find the way to make it's death look natural, so studios still using it in their pipeline do not get angry and ragequit before having chance to try their GPU based solution. But mental ray died the day MI was acquired by nVidia.
What's even worse, is that i do not see any breakthroughs in development of their GPU rendering technology (iRay) either. Every mediocre GPU renderer on the market performs better than GPU renderer from a company that makes high end computing GPUs and developed CUDA. What an irony...