You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Get everything working with the latest Mitsuba master branch. Fully correct implementation of integrators, volpath, volpathmis and prbvolpath. Comparison with PBRT-V4 in a homogeneous medium bounded by the sphere shape with emission, absorption and scattering gives identical results.
TODO:
* Test mesh sampling and directional PDFs are correct
* Validate `heterogeneous` media results
* Update reference images in `resources` submodule
Caveats:
* PRBVolpath gives incorrect gradients (correct sign, incorrect magnitude) for the medium density `sigma_t` and the phase function `phase`, but this exists in the original implementation as well.
0 commit comments