mirror of
https://github.com/games-on-whales/gst-wayland-display.git
synced 2026-10-16 05:11:59 +00:00
[GH-ISSUE #25] [Nvidia] Messed up image with custom resolutions #9
Labels
No labels
enhancement
enhancement
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
games-on-whales/gst-wayland-display#9
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @ABeltramo on GitHub (Nov 21, 2025).
Original GitHub issue: https://github.com/games-on-whales/gst-wayland-display/issues/25
This seems to happen only on Nvidia, some custom resolutions like
1366x768results in the following:but if I run the same on an AMD GPU it seems to work
At first I thought it was something to do with the Nvidia encoders or Moonlight but I can reproduce the issue even locally with the following simple pipeline:
if instead I use the non-zero copy pipeline it seems to display just fine:
@Greaper88 commented on GitHub (Nov 21, 2025):
Is this related to https://github.com/games-on-whales/wolf/issues/41 ?
@ABeltramo commented on GitHub (Nov 21, 2025):
Unlikely, since this isn't related to the encoders or Gstreamer. Plus I can't reproduce it on AMD.
It's most likely something to do with our new
EGL->Cudacode to get a high performance pipeline on Nvidia.Thanks for looking that up though!
@ABeltramo commented on GitHub (Nov 23, 2025):
Note to self: I wonder if it's this: https://github.com/games-on-whales/gst-wayland-display/blob/stable/wayland-display-core%2Fsrc%2Futils%2Fallocator%2Fcuda%2Fmod.rs#L501-L508
@Pizmovc commented on GitHub (Nov 23, 2025):
Adding another data point (from Discord):
@DatCaptainHorse commented on GitHub (Nov 24, 2025):
Some encoders don't work well with resolutions that aren't to power of 2 or so, for example OBS-Studio replaces invalid resolutions that user may input with closest valid power of 2 one. (Could be something else than power of 2, but I can't remember exactly).
@ABeltramo commented on GitHub (Nov 24, 2025):
Unfortunately I can replicate this even when we aren't using any encoder, so it's not that. By the look of it it's most likely some wrong stride/offset param passed between our DMA buffer -> EGL -> CUDA. I'm trying to pin it down properly, but these things are always painful to debug.. I'll get there, just need some time!
@ABeltramo commented on GitHub (Nov 25, 2025):
Just pushed a fix for this,
wolf:stableis currently building, give it a few minutes. Feel free to re-open this if the issue is still present.