I sometimes can not help the feeling that when using universal settings in 3.4, in some cases, especially interior scenarios, DomeLight is just not sampled enough. In some scenes where there is an interior with one (but not very small) window, using portal instead of domelight can give me around 3-4x faster rendering.
The problem with Portals is that they are not importance sampled though, so as soon as I use true HDRI environment, they will project highlights on HDRI map as inaccurate and noisy lighting, therefore I am forced to revert back to generally slower dome light. It has usually a lot noisier illumination than pretty much everything else in the scene.
I would like to avoid ever having to go back to legacy local subdivs workflow.
Maybe, DomeLight could benefit from having doubled the amount of secondary rays per pixel it gets sampled with? That would probably not break or slow down much in simple scenes, yet it could help a lot in more complex ones, where paths to reach DomeLight are not so simple.
Increasing Min. Shading Rate to something like 16 usually helps a little bit, but not that much.
The problem with Portals is that they are not importance sampled though, so as soon as I use true HDRI environment, they will project highlights on HDRI map as inaccurate and noisy lighting, therefore I am forced to revert back to generally slower dome light. It has usually a lot noisier illumination than pretty much everything else in the scene.
I would like to avoid ever having to go back to legacy local subdivs workflow.
Maybe, DomeLight could benefit from having doubled the amount of secondary rays per pixel it gets sampled with? That would probably not break or slow down much in simple scenes, yet it could help a lot in more complex ones, where paths to reach DomeLight are not so simple.
Increasing Min. Shading Rate to something like 16 usually helps a little bit, but not that much.
Comment