Current nasty test scene for hobby renderer, still some SPPM corner darkening to fix. Also now quite want a donut.
Conversation
Replying to
sadly nope, only done CPU reference for SPPM so far, trying out the various vertex merging papers from this year :)
1
Replying to
I was reading the new one from hachisuka today. looks cool and a bit hairy if you dont know veach backwards...
2
Replying to
...also weights seem very sensitive to # photons per pass, but not tried bias-aware version yet, also I may have bugs :)
Replying to
how far off do you think something like SPPM is from realtime? seems so elegantly simple... do want.
1
1
Replying to
cannot predict, but 100M+ photons for smooth result is scary number, agree amazingly elegant for tricky paths tho
2
1
Show replies