You are viewing the course site for a past offering of this course. The current offering may be found here.
Lecture 13: Global Illumination and Path Tracing (80)

Why do we care if we always have N bounces if N is sufficiently large? I would assume given some large N, we approximate the scene well enough that doing an additional bounce may not even change the pixel value.


The problem is that the value of NN depends on the scene - if you have a bunch of reflective or refractive surfaces in your scene, then you can't necessarily know a priori how many bounces you will need, so fixing a value of NN beforehand will always have some risk of missing some details in the image. The general idea that eventually additional bounces don't change the pixel value is a good one though, and the last part of the project on adaptive sampling will have you implement this idea.


Is there a way to quantify the bias we introduce have by only restricting the number of rays to fewer than infinite (like in the project where we have a max number of bounces)? I'm not really sure how to begin quantifying this bias.

You must be enrolled in the course to comment