I wonder what the performance differences of these two sampling methods would be, and if it differs depending on the objects of the scene.
JefferyYC
@bernardmc8 I am also wondering the tradeoff here. I think there are two main overheads for importance sampling. First we need to efficiently access all the light sources in the scene (which should be constant time if we pre-store it). Then we need to perform change of variable and calculate a new luminance function, as described in the later slides. However, it seems that it also takes constant time to calculate this new function for each light source from the formula.
I wonder what the performance differences of these two sampling methods would be, and if it differs depending on the objects of the scene.
@bernardmc8 I am also wondering the tradeoff here. I think there are two main overheads for importance sampling. First we need to efficiently access all the light sources in the scene (which should be constant time if we pre-store it). Then we need to perform change of variable and calculate a new luminance function, as described in the later slides. However, it seems that it also takes constant time to calculate this new function for each light source from the formula.