It's not quite that easy because of a) the noise (you're fitting distributions not values) and b) the fact that multiple light sources affect a given pixel, so you have to be a bit clever. But yes, you can get parallelism by spatial partition.
No comments yet.