The flickering is because you enabled Face Orientation in Overlay. Closed as a duplication of #124013.
@Alaska yes, seems like it's scene-dependent, for the spring scene the penalty seems fine, but here a 1.68x penalty seems too large to me.
Besides quick canceling, scheduling less samples also…
Another thing I found out is that the occupancy on GPU is as low as 0.02 for the 4.2 splash screen scene. This is due to 5f11ae70ea: for quick cancel of rendering, we now schedule less samples, so…
our microdisplacement code
Which part are you refering to?
For the material in the screenshot above, the third one is relevant
This only simplifies shading, intersection will be as expensive since we need to record the intersection and shade later.…
Preliminary conclusions
For scenes with many transparent bounces, we can consider the following solutions to improve performance:
- Ask the artists to optimize their material setup
- Use…
I wonder on which device were you testing when you said you didn't see any performance difference? I would like to test on other platforms too.
At other places where fmod is occasionally…
Transparent Shadows off
I tried to change sinf/cosf -> sincosf too in the past and did not see any improvement. If we could measure the difference it's good, but if not I would not do it when unnecessary, sometimes…
Replacing dot(N, H) with local_H.z makes sense to me.
Looking at the precision ([c++ shell](https://cpp.sh/?source=%23include+%3Ccmath%3E%0A%23include+%3Ciostream%3E%0A%23include+%3Climits%3…
I guess we were discussing "does this fix #xxx" and that triggered gitea to close the report.
@Sergey I was suggesting to add something like alpha2 = max(alpha2, 5e-8f); inside bsdf_D().
I can confirm the banding artefacts with Beckmann. Not sure if we can improve the precision in this case, otherwise it's fine to clamp for Beckmann when evaluating, Beckmann's tail is not obvious…