Search Issue Tracker
Third Party Issue
Votes
0
Found in
2018.4
2019.4
2019.4.17f1
2020.3
2021.1
2021.2
Issue ID
1328737
Regression
No
Rendering results differ on Mac and Windows when both OS's are using OpenGL Core API
How to reproduce:
1. Open the attached project from "support1328737" on either Windows or Mac
2. Enter Play Mode
3. Observe different renderings of the Quad
Expected result: The rendered Quad is black on both systems
Actual result: The rendered Quad is black with white spots on Windows
Reproducible with: 2018.4.34f1, 2019.4.24f1, 2020.3.5f1, 2021.1.4f1, 2021.2.0a13
Note:
1. On Mac, in the Texture Importer of both "a.png" and "b.png", sRGB and Generate Mip Maps have to be unchecked, and the compression is set to None
Add comment
All about bugs
View bugs we have successfully reproduced, and vote for the bugs you want to see fixed most urgently.
Latest issues
- [Android] The warning "Internal: JobTempAlloc has allocations that are more than the maximum lifespan of 4 frames old - this is not allowed and likely a leak" is displayed when ShaderVariantCollection.WarmUp is called
- RectangleContainsScreenPoint returns false for some points even when the points are in the rectangle
- [iOS] Stacked camera does not render when using post effects
- TextField cannot be edited in an extended Editor window when double-clicking it
- [VisionOS] visionos build do not add plugin or frameworks
Resolution Note (2021.2.X):
OpenGL by design does not mandate pixel perfect equality between implementations. Neither does DirectX. Neither does Vulkan. There will always be some differences. The only way to guarantee exact match is to have same HW and same driver version.
If one is trying to do things like automated testing by comparing images there must be a different image per platform. Or apply tolerance when comparing, but that requires strict reading of the spec to see what's acceptable and what not.
The only way Unity could guarantee identical results is to move into a software rasterizer.