Cosine weighted sampling 256 spp. 4:57

Got uniform sampling to be what I think is correct. Same params as previous but 256 spp. 4:50 render time

32-256 samples, 4 GI bounces, cosine weighted sampling, 2 GI samples, 8 bounces. ~5 minutes to render. Still slightly darker than Cems result

Much closer results to the reference when I take away energy conservation from the indirect light components, but the shadowed areas seem a little too dark still. This is done with uniform sampling, 16-64 samples, 4 bounces, 100 GI samples, and 4-16 shadow samples.

Went back to my project before I added my new code and just added attenuation. Looks like the bug is with the refactoring that I tried to do when I started the project.

Realized I was calculating the attenuation in the wrong spot. Now I have a weird issue where the edge of the reflective sphere is very bright and Im not sure why yet.

Even with energy conservation + inverse squared light attenuation, I havent been able to fix this overexposed look yet.