Currently, the angular hit functions are the computational bottlenecks with many, many floating point operations. So the question is then, how can we reduce the number of calculations, branch predictors, and complexity without affecting the soundness of the algorithm.
What I've done so far:
Tried to reduce as many calculations as possible from per ray to per grid. See all of these in SphericalVoxelGrid.h
Currently, the angular hit functions are the computational bottlenecks with many, many floating point operations. So the question is then, how can we reduce the number of calculations, branch predictors, and complexity without affecting the soundness of the algorithm.
What I've done so far:
SphericalVoxelGrid.h