- User Since
- Mar 10 2017, 7:13 PM (254 w, 3 d)
Dec 15 2021
Nov 25 2021
Nov 24 2021
The issue seems to be fixed in the actual version: 3.0.0 Alpha, branch: master, commit date: 2021-10-26 23:24, hash: rB8d8ce6443530.
The version: 3.0.0 Alpha, branch: master, commit date: 2021-10-26 23:24, hash: rB8d8ce6443530 seems to handle the splash screen scene as expected now except the hairs of the character.
Nov 23 2021
In 3.0.0 Beta, branch: master, commit date: 2021-11-22 23:55, hash: rBf749506163b7 the scene looks like this. Incomplete with teeth intersection.
Nov 22 2021
This might be on a different page, could you report this again separately?
Nov 16 2021
Nov 10 2021
In the build blender-3.0.0-beta+v30.7b530c609611-windows.amd64-release the crash is fixed but frames 2 and 3 of my attached scene will be rendered incomplete.
Nov 6 2021
Sep 22 2021
Please merge this latest bug report with Blender 3.0 Alpha and Cycles X (master branch): https://developer.blender.org/T91613
Aug 13 2021
The geo nodes create for each pixel of the texture instanced cubes but in the end the instances will be merged into a mesh with a high number of polygons. The colors for each merged cube will be generated by face groups. I‘ve revised the setup now by keeping the instances and using a texture projection for the result. This lowers the VRAM usage dramatically but slows down the GUI and viewport performance because of thousands of instances. My original scene is much more complex with over 800.000 instanced cubes and for this bug report I reduced the scene and isolated the issue to a minimum.
Aug 12 2021
UPDATE: Turning Motion Blur off will fix the OptiX Acceleration Structure issue.
Thanks for looking into this. The RTX 3080 here has 16GB VRAM and it's just an issue in OptiX mode. In CUDA mode it works. The question is: Why is the image sequence taking double VRAM? What is the difference here?
This is a pretty interesting bug worth taking a closer look at. The issue still exists and the only difference here is using a sequence or still.
Jul 25 2021
Apr 18 2021
Mar 30 2021
Mar 28 2021
@Michael bernardoff (kriptomik) EEVEE is using shadow mapping techniques and the Terminator Effect is a result of raytraced shadows.
Feb 27 2021
Feb 22 2021
@Christoph Werner (Taros) The test with the AMD 1950X 16 core and GTX 1080Ti shows the following result:
@Christoph Werner (Taros) I made an extreme test with tiles of 512x512. This gives me the suspicion that the GPU takes over the tiles but renders them from scratch because the render time in this case with different Adaptive Sampling settings for a faster result was nearly the same.
Feb 6 2021
Feb 4 2021
Feb 2 2021
Feb 1 2021
Seems to be fixed in:
2.92.0 Beta, branch: master, commit date: 2021-01-31 13:31, hash: rBa9b53daf23ba
2.93.0 Alpha, branch: master, commit date: 2021-01-31 14:53, hash: rBb10fac0432c6
Jan 31 2021
Overall 2.92/3 are rendering much faster. Fantastic. Here's the recording with the 2.92 log files. CUDA mode. Blender crashes when switching to another layer at the end.
I had also a crash while loading the scene. The scene is using a linked scene. I will provide a recording.
Jan 28 2021
Jan 21 2021
@CHET (cheteron) @Alma Talp (AlmaTalp) THIS is the same issue. Just inverted. It won't happen when you see it on the sphere like in the Blender example. ARNOLD, RS, and all other renderers claiming to have it fixed are just using another workaround because the Terminator Effect is more extreme on convex surfaces.
@CHET (cheteron) Just to make clear that the other renderers haven't solved the issue. It's just inverted. :-)
Jan 20 2021
@Brecht Van Lommel (brecht) This was fast. Good job!