Installation:
mpv.io
Wiki:
github.com
Manual:
mpv.io
User Scripts (including opengl shaders):
github.com
High quality video output profile (goes into mpv.conf):
profile=opengl-hq
mpv.io
Config example:
>github.com
Post your system specs and config if youre asking performance related questions.
Other urls found in this thread:
0x0.st
0x0.st
lg.com
twitter.com
Jesus christ amdgpu/radeonsi/mesa is a meme. Fuck buying vega, I'll wait another few generations for them to get their shit together first.
Error parsing option tscale-clamp (option requires parameter)
What did they change again?
It was changed from a flag to a float. Set it to 1.0 to get the old behavior of “yes”
What happened? Do you mean the Mesa 17.2 issue?
No, that one is at least just a cosmetic bug. (Although it's also funny - the AMD guy thinks it's okay to violate the spec because he thinks it “feels sort of silly”). I'm talking about the 5+ other issues I have (broken vsync, vblank_mode ignored, --opengl-swapinterval does nothing, random monitor flashing after idling overnight, GPU power state flapping, no 10 bit support, error spam in dmesg, etc.)
It's like we're back to 2014-era nvidia drivers, back when everything was similarly broken.
>I'm talking about the 5+ other issues I have (broken vsync, vblank_mode ignored, --opengl-swapinterval does nothing, random monitor flashing after idling overnight, GPU power state flapping, no 10 bit support, error spam in dmesg, etc.)
Fuck... Should i be concerned as a regular user? Any chance of those things getting fixed?
I dunno. Depends on if you notice them on your end or not? And no idea
Anyone here has experience programming video/audio synchronization using PTS with ffmpeg?
I am trying to play video and audio tracks in sync by adjusting them with external timer, however am struggling with some conceptual visualization of how exactly to keep them synchronized and dranger's synchronization tutorial just makes it worse by adding bunch of things to keep track of that go unused.
If in doubt, mpv's source code could provide an example of how you could write code to implement this sort of stuff.
But I think in general, it boils down to a question of what clock source you want to be using, latency measurements, and how you want to compensate for rate mismatch?