Screen flashes are broken on some maps
Flecked42 opened this issue · comments
Screen flashes get broken on some maps.
The map in the video is ctsj2_bal from Coppertone Summer Jam 2 pack https://www.quaddicted.com/filebase/ctsj2_1.2.zip
It happens in other maps as well, strange bug.
QS_flashes.mp4
This also happens in QSS, but it doesn't happen in Ironwail or vkQuake.
I really have no idea about this. @andrei-drexler? (Since the op said it doesn't happen in ironwail..)
Seems to have something to do with the skybox.
Type sky "" in the console to remove the skybox and the issue goes away.
Also this issue doesn't happen with 0.94.7
Also this issue doesn't happen with 0.94.7
Now that's noteworthy. The lit water patch was the only thing that would affect such a thing??
I'm not having any luck reproducing this. Here is what my configuration looks like:
- https://www.quaddicted.com/filebase/ctsj2_1.2.zip , loading it by double clicking the .exe then running
game ctsj2followed bymap ctsj2_bal - Windows 11
- https://versaweb.dl.sourceforge.net/project/quakespasm/Windows/quakespasm-0.95.0_win64.zip
- launched
quakespasm.exe(so, the SDL2 version) - NVIDIA 517.48 driver
- I tried with a clean config (deleted config.cfg's) so QS started in windowed mode
@Flecked42 can you give details on your system configuration, or see if you spot any differences vs what I tried?
Seems to have something to do with the skybox.
Type
sky""in the console to remove the skybox and the issue goes away.
There is also this console warning, might be related:
Warning: Sky texture skybal is 128 x 128, expected 256 x 128
@ericwa I have an AMD card, RX 580, using latest drivers 22.10.1
Maybe an AMD only issue?
The lit water patch was the only thing that would affect such a thing??
Ah changing r_litwater to 0 seems to make the issue go away.
But still strange that removing the skybox also makes the issue go away.
I'm guessing its the new AMD drivers since litwater alpha in JoeQuake also broke from them.
j0zzz/JoeQuake#61
Here's a video that shows the issue in more detail
Also this issue doesn't happen with 0.94.7
I take that back. Noticed there's an issue in 0.94.7 also. I guess this is the same issue as in 0.95.0 with litwater off.
These issues are most certainly from AMDs new drivers.
QS_water.mp4
Close the ticket then?
Guess that's up to you, but i don't think AMD is gonna fix issues in a Quake sourceport.
Ironwail and vkQuake don't have the issues though, so i dunno.
@ericwa, @andrei-drexler: Do we have a workaround for this??
The issues go away when using -noglsl in the command line, maybe will help.
Here's a (hopefully zero-cost) workaround: amd-polyblend-workaround.txt
Here's a (hopefully zero-cost) workaround: amd-polyblend-workaround.txt
Huh.. Move glColor4fv between glBegin and glEnd, which is actually legal
according to the man page.
Thanks, will apply shortly.
OK, should be fixed as of f09b26b, closing.
If you can't build yourself and need win32 or win64 binaries, tell me.
Not a big deal for me as i use JoeQuake, but if you want me to test it, i would need win64 binaries.
but if you want me to test it, i would need win64 binaries.
Testing would be nice. Here are win64 builds from git master: qs-w64.zip
Looks good, fixed on my end.
Looks good, fixed on my end.
Thanks for testing.