Light power: 1000 (W), radius used for radiance estimation: 0.6 (m). Number of photons emitted: 1 million.
Final gathering used with 32 monte carlo samples, 8-16 samples per pixel.
Rendering time: ~6 minutes.
NOTE: There were two bugs in my program (which had not been there before - they were introduced by my stupid decision to rewrite a large part of the code from the last project). The diffuse photon bounce is too biased in the normal direction, and the photons don't bounce enough times. I have fixed both of these bugs (and some more) in the code used for the Teapot Rendering Comptetition. I may re-render this scene and update this page some time in the future, though I'm too lazy for that now :)
Indirect illumination only
Original cornell box using a point light
Failure cases:
Problem: sampling for caustic from both the photon map and the backward path tracing
Problem: Fresnel reflection not working...
Problem: Color bleeding is gone due to using the incorrect material to do shading
Caustic photon map:
