Visual glitches with shadows with large blur radius #4238
Assignees
Milestone
Comments
emilk
added a commit
that referenced
this issue
Mar 27, 2024
* Part of #4238 When one side of a rectangle is all rounding we need to take care not to produce duplicated vertices in the rectangle path generator. The old code only handled three sides, but forgot the last side (the right side). The new code handles the right side, and also handles the other sides more robustly (with a floating point eps) and efficiently (in a single pass). The glitch was most notable in shadows with a high blur width. Examples of the glitch: <img width="203" alt="Screenshot 2024-03-26 at 20 15 38" src="https://github.com/emilk/egui/assets/1148717/dc1c0a06-35f0-4fda-a011-0e37d18454a0"> <img width="220" alt="Screenshot 2024-03-27 at 09 48 48" src="https://github.com/emilk/egui/assets/1148717/c278b28e-c3f9-4c82-ba20-0480621efd2f"> <img width="33" alt="Screenshot 2024-03-27 at 09 49 21" src="https://github.com/emilk/egui/assets/1148717/379ddf77-6590-4444-9c2e-67ab1e071f0f">
emilk
added a commit
that referenced
this issue
Mar 27, 2024
* Closes #4238 The comment in the code explains it well, but the short of it is this: we can't handle a shadow blur width larger than the shadow rectangle, so we need to clamp the blur. This means smaller things will cast shadows with a smaller blur width, but that's better than having visual glitches.
hacknus
pushed a commit
to hacknus/egui
that referenced
this issue
Oct 30, 2024
…lk#4244) * Part of emilk#4238 When one side of a rectangle is all rounding we need to take care not to produce duplicated vertices in the rectangle path generator. The old code only handled three sides, but forgot the last side (the right side). The new code handles the right side, and also handles the other sides more robustly (with a floating point eps) and efficiently (in a single pass). The glitch was most notable in shadows with a high blur width. Examples of the glitch: <img width="203" alt="Screenshot 2024-03-26 at 20 15 38" src="https://github.com/emilk/egui/assets/1148717/dc1c0a06-35f0-4fda-a011-0e37d18454a0"> <img width="220" alt="Screenshot 2024-03-27 at 09 48 48" src="https://github.com/emilk/egui/assets/1148717/c278b28e-c3f9-4c82-ba20-0480621efd2f"> <img width="33" alt="Screenshot 2024-03-27 at 09 49 21" src="https://github.com/emilk/egui/assets/1148717/379ddf77-6590-4444-9c2e-67ab1e071f0f">
hacknus
pushed a commit
to hacknus/egui
that referenced
this issue
Oct 30, 2024
* Closes emilk#4238 The comment in the code explains it well, but the short of it is this: we can't handle a shadow blur width larger than the shadow rectangle, so we need to clamp the blur. This means smaller things will cast shadows with a smaller blur width, but that's better than having visual glitches.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
There are two separate issues.
One is that a blur radius larger than the width/height of a rect looks borked:
The other is a strange bug in the rect tessellator:
The text was updated successfully, but these errors were encountered: