Closed arose closed 15 hours ago
Regarding second one, I am still able to detect it when clipping is high. In worst case scenario where clipping is at 99 (and far is enabled) I need blurBias to be 0.05 to completely get rid of the artifact.
Not sure how to handle that automatically. Since it is such an edge case and it can be manually handled, let's leave it.
An other issue I notice on (mesoscale explorer only) are some spare non-occluded pixels that are fixed on the screen when moving the camera. (pixels position depends on screen size and resolutionScale)
That is odd, not sure it is related. Best create a new issue with a state to reproduce. Thanks.
Reported in #1122.
Add
reuseOcclusion
parameter to multi-sample pass. This helps with the first artifact reported. It happens because we reuse the occlusion data in the multi-sample pass for performance. We try to correct for the jitter but apparently it dos not (fully) work. As an alternative/workaround we can disable by default in screenshots.Add
blurBias
parameter to occlusion pass. Not sure if there is an automatic way to calculate a "correct" value but 0.5 seems to work well. There was a previous issue related to that value in the synaptic gap model, @giagitom, can you check?