Closed germolinal closed 2 years ago
Everything just works; IEEE floating point has both positive and negative 'infinite' values that are well-defined; they do sensible things like positive infinity is greater than any finite float in a comparison, etc. There's some further discussion of infinities with ray-bounds intersection tests here.
So, I was wondering about this line... what happens when the vector has an element == 0? Then one of this values is infinite. Any suggested handling? is it necessary?
https://github.com/mmp/pbrt-v3/blob/aaa552a4b9cbf9dccb71450f47b268e0ed6370e2/src/accelerators/bvh.cpp#L666