Lecture 13: Global Illumination & Path Tracing (5)
NicholasDrian
Global Illumination ray tracing seems really computationally expensive for real-time graphics. It makes me wonder if global illumination can instead be calculated per-vertex and textured on.
Staffziyaointl
Great observation! Some effects of global illumination can be pre-computed (e.g. shadows), while others are very hard/impossible (e.g. reflections and refractions) because they change depending on the position of the camera.
somaniarushi
I'm wondering if designing inter-reflections is just an extension of standard ray tracing for shadows — except that we follow the ray through more stages of reflections, and the material the rays are interacting with are more reflective. Is this understanding correct?
melodysifry
To answer the above comment, I believe that understanding does sound correct- we're recursively tracing more and more reflections
NKJEW
Something I always find really interesting about any scene with a mirror is that its evidence that in addition to being able to have complex rendering in your main line of sight, you also need to be able to model how those complex artifacts appear when bounced off of another surface. A less physically accurate model might miss the intricacies of having inter-reflections and the like in the mirrored surface on the left.
Global Illumination ray tracing seems really computationally expensive for real-time graphics. It makes me wonder if global illumination can instead be calculated per-vertex and textured on.
Great observation! Some effects of global illumination can be pre-computed (e.g. shadows), while others are very hard/impossible (e.g. reflections and refractions) because they change depending on the position of the camera.
I'm wondering if designing inter-reflections is just an extension of standard ray tracing for shadows — except that we follow the ray through more stages of reflections, and the material the rays are interacting with are more reflective. Is this understanding correct?
To answer the above comment, I believe that understanding does sound correct- we're recursively tracing more and more reflections
Something I always find really interesting about any scene with a mirror is that its evidence that in addition to being able to have complex rendering in your main line of sight, you also need to be able to model how those complex artifacts appear when bounced off of another surface. A less physically accurate model might miss the intricacies of having inter-reflections and the like in the mirrored surface on the left.