Why do we choose to Monte Carlo integration on the BRDF instead of the surface integral?
Does the Monte Carlo algorithm have a high risk of missing powerful light sources coming from a very narrow area, i.e. a laser pointer? It seems unlikely that a random direction would happen to be pointing in the same direction as the laser pointer. I suppose the solution is just to take more samples?
This might be discussed later, but what part of this process is accelerated in RT chips?
I am missed... I hope we can talk about it in the lecture again.
How do we determine the sampling distribution?