It feels like there are many limitations to recursive ray tracing, such as multiple mirrors facing each other so the rays continually bounce in many directions. Is there a good limitation to how many times we recurse before we stop?
This concept reminds me of how clouds in this one video is created: https://www.youtube.com/watch?v=4QOcCGI6xOU&ab_channel=SebastianLague
So as a result, we can just use the light rays that we are more interested in rather than caring for rays that may not be interesting to us
Replying to Jerry: if you're familiar with rendering engines, the level of recursion is really up to the artist. Most rendering engines set default cap levels at about 4~8 depending on the ray type.
If there are multiple light sources, do we simply have more shadow rays? Also, when calculating the final pixel color as a "weighted sum of contributions" are we talking about primary + secondary + shadow?
The light source is characterized as a point instead of a volume in the slides. I wonder if this is true in general for raytracing since it may be hard to do the shadow ray tests on an area.
Why do we need to check for the shadow at the second point from the left? I thought that point is just a refraction point.
When a max recursion level is used, I wonder whether there will be some strange artifact due to a same max level being used for different types of surfaces. For example, if we have two pairs of surfaces (A, B) and (C, D), where A and B are facing each other, and so are C and D, so that rays bounce between surfaces in each pair forever. A and B have much greater specular coefficient than C and D. Will the fact that AB and CD are different in their specular coefficient but use the same max recursion depth cause the graphics to look weird?