> latest stable windows builds for mpv:
> latest windows build for youtube-dl:
> lowest vo configuration:vo=opengl
> low vo configuration:vo=opengl-hq:deband=no
> medium vo configuration:vo=opengl-hq:interpolation
> high vo configuration:vo=opengl-hq:interpolation:scale=ewa_lanczossharp:cscale=ewa_lanczossoft
> insane vo configuration:vo=opengl-hq:interpolation:scale=ewa_lanczossharp:cscale=ewa_lanczossoft:prescale=nnedi3:nnedi3-neurons=64:prescale-downscaling-threshold=1.5
> medium ytdl-format config example:ytdl-format=bestvideo[height<=?1080]+bestaudio/best
> display statistics for the currently played file:
> sharpening post-shaders:
The latest (experimental) plex media player release here:
Should I enable ANGLE now that its in the windows builds?
Also why might I seem to be getting judder and stuttering when I have a 144hz display and I don't see any dropped frames
I see delayed and mistimed frames in stats, but I shouldn't have to enable interpolation when I'm running at 144 to get rid of stutter, but I do.# Video settings
# Audio settings
# Subtitle settings
sub-text-font="Source Sans Pro Semibold"
# On-Screen-Display Settings
osd-font="Source Sans Pro Semibold"
# OSC Options Via Script-Opts
# Screenshot settings
# Window settings
# Miscellaneous settings
# ytdl - Youtube Downloader
# Extensions - Do not add custom settings that are not extension related below this line.
I want to use vapoursynth-mvtools for motion interpolation.
How THE FUCK do I do that? Do I need to compile this myself?
Since directshow subtitle engines are shits, I have to use mpv.
However I fall in love of crispen edge and super-res algorithms from madvr.
Unfortunately they are no equivalent feature inside mpv, that's so pathetic.
Off the top of my head:
- tscale now does nothing when video FPS == display FPS
- ANGLE now supports 10-bit
- there is now a delay between decoding a video frame on the GPU and reading it back, so the CPU doesn't wait on the GPU
- wm4 is working on A-V filters for ffplay-style visualisations (spectrogram, etc.)
- wm4 is working on 10-bit HEVC decoding with DXVA
The effect is pretty subtle, especially if you have accurate VSync and mpv is presenting the frames with a proper 3:2 pattern. It's only really obvious with certain panning speeds.
Not at all. You're just use to seeing 24fps video everywhere. You'll start noticing it when you start watching a lot more 60fps videos.
I've only just recently started watching videos at 60fps, and can notice when frames go down. Sometimes it's subtle and I don't notice and then there's times when it's extremely noticeable.
Windows 10 is arguably the best Windows version for mpv because you can finally resize console windows with the mouse. It's almost as good as using it on an OS with real pseudoterminals.
>especially if you have accurate VSync and mpv is presenting the frames with a proper 3:2 pattern
How can I determine if this is properly occurring? I'm using the "high vo config" from the OP.
If you're using the :interpolation suboption, there should be no judder (it will be replaced with slight blurriness instead.) If you remove :interpolation, but keep video-sync=display-*, you should get a proper 3:2 VSync pattern. Either way, you can confirm it's working by looking at the thing on the end of the status line. It's in the format <vsync ratio>/<total dropped frames>, so you want to see 2.500/0 for 24fps material on a 60fps monitor.
How do I know which config in supposed to be running? I use an amd r7 370 for example. What is the default config? Is it good enough?
And what difference is one supposed to notice exactly?
seems like this is the best place to ask:
what's some software I can use audio/video sync? basically I'm trying to take a raw release and re-mux the audio from an older release into it so it has the japanese and english audio. last time I tried this I just adjusted the offset in mkvtoolnix and in the player until I got it right, but there must be a better way that the "pros" use, right?
it's the best I can do considering this will probably never be picked up by a proper release group because it's old, there are only a few known versions online (all bad), and the only sources are VHS and LD, and the only existing sub is an old hardsub with terrible font. so really it needs a new script and new sub, probably a new rip altogether, but that's beyond me.
I have a 7870 and I use the "high quality" config and it runs well for me even on massive 15GB bluray encodes. Your card is a tad weaker than a 7870 so should still do.
I've tried the "insane" quality preset which uses nnedi3, and the 7870 definitely can't handle it. You need a 980Ti or some shit for that.
r9 390 8GB
Tested on Kernel 4.3, 4.2 and Debian Stable's Kernel
Last week I posted saying I was having problems with mpv playback with any added shaders past the default configuration where it would cause white artifacting when a shader was applied or simply the sane default of opengl-hq was applied.
Finally found the fix today after some related issues popped up, the r9 390 has issues with dpm enabled (this is done by default). Setting the kernel parameter radeon.dpm=0 (disabling it) fixes the issues and allows vo to be modified again without problems. The issue can also cause gpu lockup during general use, this is also fixed by using this workaround.
Obligatory sorry for the blogspot, hope it helps anyone having issues.
Someone on github seems to have done more extensive research but I agree with his findings.
ANGLE/dxinterop produces much less jittery output when using interpolation.
Also potentially improved performance.
Thanks, I've been fucking myself in the ass with msys2 for like a week glad this is over.
I messaged him on BB a few days ago and he replied back saying he didn't use it for a while now and someone (obv from here) kept messaging him stupid shit about being a shill and he had no idea what they were on about so decided to completely remove it.
>I've been fucking myself in the ass with msys2 for like a week glad this is over.
You must be doing something wrong. I got compilation with msys2 running in like 10 mins and
now I just run a script to update source and compile, takes like 2 mins. It's way faster this way than downloading binaries.
The place I work at records conferences to multiple sources. If there are problems with the audio from one source I remux it from another source.
I sync the audio from an external source to the original one by simply importing the videos into Audacity and drag the external until it matches the original. Then I cut any leftovers so the lengths match and export the prepared external audio which I then mux with the original video.
If you don't want to recode the audio, then get the length of the initial offset and the total length and use those as input options in ffmpeg let's say.
what about this?
this won't work on youtube tho.
something like this will:
When will mpv integrate SSimDownscaler?
I have used Potplayer for a long time and wonder if mpv is able to do the following:
Resume last played file,
Keep video position saved,
60fps or use svp 4.
Also from what I remember it might have been less straightforward to change processing range to (0-255).
>Subjectivity aside, which is objectively superior? I guess I mean to ask, which scaling method is the least "destructive".
There is no objectively superior as long as you don't give the goals you have for a scaler. Is the picture supposed to be as sharp as possible? The least amount of ringing? Alasing? You have to define what a good upscaled picture is for you. Some of the traits I just mentioned (and the ones I left out) unfortunately also cancel each other out to some extend, so it really depends on what you like subjectively the best.
The trait "smooth" usually does better in hiding artifacts created by upscaling.
I use MadVR with MpcHC and some mods or whatever, am i doing good?