The above mentioned paper compensates the tilted normal with an additional tangent facet, it requires random walk between facets to compute…
Actually I do not know if this condition always holds, but seems to be so in the test scenes.
Hey! This is not a problem of "more processing time per sample", it's the pre-processing (building the light tree) that takes too long because there are too many emissive triangles, should have…
Uploading some files here, as it seems one can only upload 5 files per comment
@mod_moder There are different roughness types, specified for half angle or outgoing direction for example, they are not quite the same thing. I'm actually more interested in another field called…
Jacques was suggesting comparing the identifier instead of name. identifier is default to name if not specified: https://projects.blender.org/blender/blender/src/commit/3b900048f1c5ec098ade…