AdamYuan
commented
2 years ago Yes, these features are in my TODO list, but may not be implemented in the near future. Currently I'm working on parallel SBVH construction.
Open tigrazone123 opened 2 years ago
AdamYuan
commented
2 years ago Yes, these features are in my TODO list, but may not be implemented in the near future. Currently I'm working on parallel SBVH construction.
tigrazone123
commented
2 years ago Did you seen https://github.com/jbikker/bvh_article/? This article series have fast bvh methods
AdamYuan
commented
2 years ago It seems the article only described a naive single-threaded recursive binned SAH BVH constructor. The method didn't make use of multi-core CPU power, and the naive SAH BVH is inferior to SBVH in quality (ray tracing performance). My parallel SBVH construction system aims to build high-quality SBVH as well as fully utilize CPU power, which is far more complex.
These are some of the references: https://www.nvidia.in/docs/IO/77714/sbvh.pdf https://diglib.eg.org/bitstream/handle/10.2312/pgv20161179/021-030.pdf http://www.sci.utah.edu/~wald/Publications/2010///LRB_BVH/download//LRB-BVH.pdf
tigrazone
commented
2 years ago Good thing is Blas, Tlas for instancing
Hello! Did you plan to implement: