You are viewing the course site for a past offering of this course. The current offering may be found here.
Lecture 12: Monte Carlo Integration (33)

If the viewpoint of the camera changes in this scene, is there any way to reuse the computations for the lighting here? The discussed approach we currently have is to do the lighting computation for each pixel in the camera, but it seems like that would change if we had a new camera viewpoint.


This is definitely recommended practice, even on the NVIDIA developer guide ( You can typically get away with only resampling for geometry that has just come into view or alternatively for meshes that have moved significantly. All this depends on the level of realism you require - for applications where low precision is sufficient, you can get away with a super lazy resampling scheme, but if you are going for super realistic lighting, you would likely want to recompute for the majority of vertices that have changed location relative to the camera.


How was this image sampled? I am having a bit of difficulty to try to picture it


It is interesting to see how fuzzy the image looks. I think this occurs because when you sample with solid angle, you might sample points in which there are no rays going towards the light source, creating random pixels that are darker. Is my understanding correct?

You must be enrolled in the course to comment