Niklas Haas

^ hwdec shouldn't have anything to do with ICC. The ICC profile application happens entirely inside the renderer (either `--vo=gpu-next` or `--vo=gpu`), as well as in the graphics adapter (for...

> > ^ hwdec shouldn't have anything to do with ICC. The ICC profile application happens entirely inside the renderer (either --vo=gpu-next or --vo=gpu), as well as in the graphics...

I thought drivers are already capable of doing this under the hood? Solving this generally is probably going to be rather tricky, since we don't even know the input framerate....

Actually, couldn't we do it like this? Combining the two things we *do* know: 1. The display framerate (which for adaptive sync I assume is meant to be an uppter...

Does this help? ``` video-sync=display-resample interpolation=yes tscale=oversample override-display-fps=48 ``` That'll make mpv frame double, but I have no idea if that results in usable or sane timing, since it will...

Oh, right, it doesn't change the pattern unless you combine it with `--untimed` but then the timing is fucked anyway.

LGTM Bitrates and codec quality have gone up, even on YouTube, and the recent bump in deband strength invalidates much of the preexisting arguments.

I personally agree that cycling through 'no audio' is useless because there's a more convenient 'mute' hotkey for this, and in practice you more often want to be using this...

Needs to not apply to SRT subs, which definitely shouldn't be rendered out as HDR. For this, SOME amount of internal metadata plumbing is necessary so mpv can later decide...

Seems reasonable enough.