Search Issue Tracker
Won't Fix
Won't Fix in 2021.2.X
Votes
0
Found in
2020.3.13f1
2021.1.12f1
2021.2.0b1
Issue ID
UUM-3786
Regression
No
Procedural Virtual Texture will sometimes become corrupted when using InvalidateRegion
Reproduction steps:
1. Open attached Project
2. Press Play
Expected result: The terrain has no artifacts
Actual result: Some of the terrain tiles sometimes are loaded in corrupted
Reproducible with: 2020.3.13f1, 2021.1.12f1, 2021.2.0b1
Could not test with: 2019.4 (Could not downgrade)
Notes:
Additionally can be reproduced by using a Free look camera and flying around at high speeds until an Artifact is noticed
The Artifact goes away after being Invalidated
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
- Rigidbody prefab in scene is losing overrides on Linear/Angular Damping
- "ObjectDisposedException: SerializedProperty injectionPoint has disappeared!" errors are thrown in the Console and the Renderer Feature addition cannot be undone after performing Undo on a Full Screen Pass Renderer Feature
- Rendering corruption on ARM Mali GPU when using shader with FrameBuffer and DepthTexture with Vulkan API
- TextMeshPro text fails to render on macOS Standalone builds when Canvas Plane Distance is close to the Camera's Near Clip Plane
- info-log.json logs the wrong OS version when using a macOS device
Resolution Note:
Will not be fixed for now, since PVT is in an experimental, unsupported state. Please reach out if this bug needs to be reopened and we can reassess the issue.
For now, though, it looks like there's a workaround of forcing a refresh by calling an EvictRegion() (instead of InvalidateRegion()) on the entire virtual texture, or on the region that got corrupted. If this is an editor-only corruption that does not occur that often, this may be a good-enough (clunky) workaround for now?
Resolution Note (2021.2.X):
Will not be fixed for now, since PVT is in an experimental, unsupported state. Please reach out if this bug needs to be reopened and we can reassess the issue.
For now, though, it looks like there's a workaround of forcing a refresh by calling an EvictRegion() (instead of InvalidateRegion()) on the entire virtual texture, or on the region that got corrupted. If this is an editor-only corruption that does not occur that often, this may be a good-enough (clunky) workaround for now?