Search Issue Tracker
Fixed in 8.x
Votes
2
Found in [Package]
7.5.1
Issue ID
1222576
Regression
No
[Shaderlibrary] Inverse View Projection transformation matrix is not the inverse of View Projection transformation matrix
Universals Inverse View Projection matrix is wrong.
Repro steps:
1. Open project attatched (1222576-InverseVP.zip)
2. Open SampleScene
Expected Results: All objects have a 1 unit RGB 'grid' aligned with the world
Actual Results: All objects have an RGB thing that changes wildly upon chamera movement
Regression happend sometime between 7.3.1 and 7.5.1 URP
Happens still in latest on 2020.2b5
Notes:
Tested on DX11 and metal, there may be more issues on OpenGL APIs becasue fo the inverse depth.
Sample project contains hacks in both shadergraph and custom shader that can be used to check the expected results(the hack is inversing the 2nd column of the matrix)
This is used in the Boat Attack demo project for caustics rendering and we have seeen many users learn from this project and use the same technique and either run into the same issue or use the hack, both are not great options and make this a higher urgency issue IMO
All about bugs
View bugs we have successfully reproduced, and vote for the bugs you want to see fixed most urgently.
Latest issues
- Crash on JobQueue::HasJobGroupIDCompleted when closing the Editor while in Play mode on a specific project
- In "Preferences" section the “SpriteShape” menu item, the details page title “SpriteShape”, and “ControlPoint” entries are displayed as code strings rather than formatted UI strings
- Errors thrown constantly when Virtual Offset Debug is enabled and lighting was baked on AMD machine
- Persistent Memory Leak when reloading domain and using Distance-based Ghost Importance
- HDRP project doesn't render in standalone player when using High stripping
Add comment