check out my latest render HD Buddha:
I now have my three reference images which I will use to compare against and validate the upcoming Bidirectional Path tracer. They are:
- The standard Cornell Box with a glass sphere, only to test the correctness of the BDPT algorithm.
- The Buddha above, which is a pure acceleration structure test, as the buddha has over a million triangles.
- The "table" scene, which will test both the BDPT algorithm and the acceleration structure.
I don't anticipate any issues getting the bidirectional algorithm up and running, even though it requires some rewriting, as it's a fairly obvious extension of the naive technique (although multiple importance sampling will probably take me a while to get right). My next immediate goal will be replacing my ugly octree with a proper BVH. This should keep me busy for a while, so I'll see what to do next when I get there (probably improve the primitive interface and work a bit on the scene file format)