Lecture 13: Global Illumination & Path Tracing (81)
yinxudeng
I understand there is an infinite number of bounces, but why does the integral have infinite dimensions?
o0WeiyuFeng0o
You need to integrate all paths of each bounce, then integrate all bounces. You could think this from the perspective of linear algebra. You could think each bounce as a vector of path. Then, path will be the rows of a matrix, and the times of bounce will be the columns. (Vice versa, path could be columns and times of bounce could be rows). This leads to infinite dimensions.
shreyaskompalli
One question I had about this process is how this changes based on the surfaces that we are reflecting off of. For instance, for a highly reflective object the light rays that bounce off of that object will be far more significant and visible than rays that bounce off of a more diffuse or dull object. So, do we change our N cutoff value based on the objects in our scene?
I understand there is an infinite number of bounces, but why does the integral have infinite dimensions?
You need to integrate all paths of each bounce, then integrate all bounces. You could think this from the perspective of linear algebra. You could think each bounce as a vector of path. Then, path will be the rows of a matrix, and the times of bounce will be the columns. (Vice versa, path could be columns and times of bounce could be rows). This leads to infinite dimensions.
One question I had about this process is how this changes based on the surfaces that we are reflecting off of. For instance, for a highly reflective object the light rays that bounce off of that object will be far more significant and visible than rays that bounce off of a more diffuse or dull object. So, do we change our N cutoff value based on the objects in our scene?