At 39:30 , true you need to consider the second box for possible intersections but ONLY if you care about location of intersection or other details. For example , it you’re just triangle to detect if the shadow ray reaches the light source, you can stop at the first intersection discovered.
At 1:05:50 , just keep a list of triangle ids for triangles you’ve already checked. This could be fixed size too and “circular buffer” so you don’t need alloc while marching. The circular buffer will naturally always hold triangles for a few neighboring bins if the spatial hash is tuned properly (e.g. bins hold number of triangles equal to 1/5 circular buffer size.
I recently heard about volumetric meshes. Does anyone know how that is related to the bounding volume hierarchy (BVH) technique? Is volume mesh an alternative?
Thank you very much for the free knowledge !
At 39:30 , true you need to consider the second box for possible intersections but ONLY if you care about location of intersection or other details. For example , it you’re just triangle to detect if the shadow ray reaches the light source, you can stop at the first intersection discovered.
At 1:05:50 , just keep a list of triangle ids for triangles you’ve already checked. This could be fixed size too and “circular buffer” so you don’t need alloc while marching. The circular buffer will naturally always hold triangles for a few neighboring bins if the spatial hash is tuned properly (e.g. bins hold number of triangles equal to 1/5 circular buffer size.
awesome!
I recently heard about volumetric meshes. Does anyone know how that is related to the bounding volume hierarchy (BVH) technique? Is volume mesh an alternative?
I think volume mesh is about subdividing within the surface of objects not around them. Maybe you mean voxel grid?
Nice 👍
This is sooo good!! thank you!