Wiki:
github.com
Manual:
Stable: mpv.io
Git: mpv.io
User Scripts & Shaders:
github.com
High quality video output profile (goes into mpv.conf):
profile=gpu-hq
Configuration Files:
mpv.io
mpv.io
Input.conf:
github.com
Post your system specs and config if you're asking performance related questions.
Windows Builds:
sourceforge.net
Evaluating mpv's upscaling algorithms:
artoriuz.github.io
Other urls found in this thread:
github.com
youtube.com
0x0.st
files.catbox.moe
github.com
youtube.com
ffmpeg.org
twitter.com
>When used with a luma prescaler you need to multiply the first offset (the one that is -0.5 by default) by a prescaling factor (due to mpv shader API limitations, offsets can't be calculated automatically).
That did igv meant by that?
When trying to use that mvtools script, launching mpv just results in
>[vapoursynth] using 4 concurrent requests.
and mpv closes
does anyone know why? using latest shinchiro
With g-sync is it a good idea to set the fps to 48, as to avoid the monitor frame doubling, and enable interpolation? Is there a way to automatically double the frames in place of using a fixed limit?
How much time do you guys spend messing around with your config vs actually watching content?
How do I move slim box upwards so it doesn't block subtitles?
Is this an old version of MPV?
Why does your seekbar look like that?
There are different seekbar styles and osc layouts to choose from
osc options are here
github.com
I think:
osd-bar-h
Look in the documentation.
you need an iq of at least 40 to use mpv
Can somebody try watching this video youtube.com
fuck off with this shit. klite codecs with mpc hc is all you'll ever need.
Exact same behavior for me.
Would if I could, friend. I'd have to drown my monitor in holy water afterwards, and that's too much of a hassle.
JEEEEEEEEEEEEEB
>klite codecs
Fucking lol. Is this 2006?
Posted this in the last thread but no help.
Left is mpv & right is mpc-hc.
Both with default settings.
What is wrong with mpv?
>no information
Who knows? If mpv is different, chances are mpv is more accurate.
Thanks, good to know it's not just my hardware being retarded.
IGGGGGGGGGGGGVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVV
I already told you last thread, you can clearly see more quantization errors on the right so the left image is more accurate tho both are useless without a color managed monitor and icc profile.
i configured mine 2 years ago. spent around 2 weeks.
now, i dont configure it. i saved my mpv.conf.
/thread
The only question that matters here.
why is this better than VideoLan?
Bump?
What's his endgame bros?
Taking hostage all the pull request till the end of time?
It seems Vulkan doesn't have that problem. I wasn't getting those red dots until I tried it with gpu-api=d3d11.
Just fork mpv and merge all the pull request yourself.
1000IQ
It might be a d3d11/SPIRV-Cross bug then.
I don't think so. I'm having the same problem with opengl/dxinterop and I don't even have d3d11 backend, shaderc and crossc in my build.
Is there a way for mpv to automatically manage subtitles based on the audio?
I want something kind of like this:
1. Audio languages that I know, for instance en, fr. If there's a forced subtitle in the same language, use that, otherwise no subtitles.
2. Audio languages that I'm learning, for instance ja. Use subtitle language in the same language. If not available use en or fr.
3. Audio language in any other language. Use either en or fr subtitles.
You can make mpv prefer certain languages among audio and subtitle tracks, but nothing more complicated. You can probably do it with a script, though.
I'm seeing it with d3d11, vulkan, and opengl.
I'm on gentoo and don't see an issue with vulkan.
I can't watch things for most of the day because people will barge in and interrupt me/look at what I'm doing. In those times I configure mpv.
Probably more time configuring it than using it since they take most of my time.
>2 minutes every 6 months
>no more than 5 if some incompetent dev messed up all the code
>klite codecs
HAHAHAHAHAHAHAHAHA
Should I just use bilinear for downscaling a video that is an integer multiple of my resolution or is there an advantage in using something else?
Btw, why not force disable correct-downscaling with dscale=bilinear? It doesn't do anything, only slows down rendering.
lads just match the source resolution to your displays and stop worried so much about scaling
How do you propose to do that without scaling?
I didn't even use hardware decoding though.
what kind of wonky screen res do you have
why should i use this instead of vlc?
1366x768
Not 810p at least.
because its better
I can't believe those hdr autists are still arguing on github
So you can be cool, "fit in with Jow Forums", "autistic", "elitist" and many other things that will cause many non-mpv users to foam at the mouth. It's a lifestyle thing, apparently. Give it a go.
Highly and easily customized.
Neither did I.
panels from your ancient thinkpad will be shit no matter what settings you apply
what contrast does it have? 10:1?
can it stream to Chromecast?
No idea. Isn't that just sending a file though? I wouldn't say that that's a media player's job.
I'm not familiar with the details put it's streaming a video file to a tv from PC for example. VLC can do it.
And VLC is a mess of useless options, wouldn't be surprised if that had some more "features" that mpv doesn't have.
it works really well actually
How does it connect, anyway? You can for example, send your webcam input to mpv via ffmpeg.
As I said I'm not sure of the technical side of things but it plugs into an HDMI port on your TV and connects to vlc (on your pc) through your home wifi
what a bunch of nerds
frick off h8r
Here are comparison pictures of the 4k mmd linked earlier in this thread, downscaled for displaying on a 1080p screen. It's a 4k video with 4:2:0 chroma subsampling, played at 1080p. The screenshots are "Ctrl s" screenshots, i.e what is actually being displayed by mpv. Because the vulkan backend can't take such screenshots, I'm using opengl.
The screenshots use the exact same mpv.conf, except for the scalers (there's a readme explaining everything). Tested are the bilinear, mitchell and ssimdownscaler scalers.
bump
>Because the vulkan backend can't take such screenshots
what
Use diff.pics
Check the comments on github, he gives some more details to a fellow who asked about it.
>webshit
I want valid comparisons.
github.com
That issue doesn't mention lignux but I'm on lignux and it's present here too.
>Because the vulkan backend can't take such screenshots
I'm pretty sure libplacebo fixed that
>actually
I didn't say it didn't, no need to say this like you're contradicting me.
Reminder that fbo-format=rgba32f is not a meme!
can you stop posting weeaboo asses over my fav media player
>mmd
It's the shitty idolmaster gacha game
>can you stop posting weeaboo asses over my fav media player
NO!
How so
youtube.com
can you play 360 videos using mpv?
fuck off, 'rgba32f' has zero benefits over the default 'auto'.
It has 2^16 higher bit depth.
Thank you
What's the best upscaler to watch old live action DVDs (mostly sitcoms) on a 1080p TV?
And by best I actually mean most reasonable. Nothing super meme-y that uses 1000% cpu for a 0.0001% improvement when you compare a still frame side-by-side.
In order words, what gives me the best bang for buck?
sharpen=2
As you wish.
Scalers are GPU bound. The last link in the OP compares their speed and quality. It doesn't,, however, compare bang for your buck. That's mostly for you to decide, and depends on how noticeable the improvements are for you on your setup.
FSRCNNX is really good for upscaling dvds.
so when did you all finally make the switch from VLC to MPV? for me it was 2017
This sounds smart. I'll try it.
Thanks for the explanation. I actually don't know anything about scalers since I mostly watch 1080p content. What I meant mostly was what would be better for non animated content, since I know most of Jow Forums optimize their mpvs for anime. I'll check the link.
I have a feeling this is a ruse.
>that ringing
>that weird wobbling on the lines
Whatever you used there is worse than bilinear.
>switch from VLC
If you mean switch from MPC-HC, it was somewhere around late 2014/early 2015 for me.
rate .conf
# VIDEO
profile=gpu-hq
gpu-api=vulkan
vulkan-async-compute
vulkan-queue-count=3
spirv-compiler=shaderc
hdr-compute-peak=yes
scale=ewa_lanczossharp
cscale=ewa_lanczossharp
video-sync=display-resample
interpolation
tscale=oversample
# AUDIO
audio-file-auto=fuzzy
volume-max=150
# SUBS
alang=eng,en,hr
slang=eng,en
sub-auto=fuzzy
# MISC
stop-screensaver=yes
screenshot-format=jpg
screenshot-jpeg-quality=100
screenshot-tag-colorspace=yes
screenshot-high-bit-depth=yes
screenshot-jpeg-source-chroma=yes
Not enough memes.
I used FSRCNNX_16, that " weird wobbling" is because the DVD is interlaced.
Pic related is a deinterlaced DVD also using FSRCNNX_16
Using "vf=bwdif:parity=tff" instead of mpv's default deinterlacer improves thins a little.
*things a little
Not an mpv question but an ffmpeg question.
ffmpeg -h encoder=libx264 returns
Supported pixel formats: yuv420p yuvj420p yuv422p yuvj422p yuv444p yuvj444p nv12 nv16 nv21 yuv420p10le yuv422p10le yuv444p10le nv20le
So it seems both 8 bit and 10 bit are supported. How do I specify when encoding which bit depth it should use?
--pix_fmt yuv420p10le -c:v libx264 --profile:v high10
This should be enough for 10bit I think.
I think you have to configure for x264's bit depth when you build ffmpeg
ffmpeg.org
That seems to be outdated though, because it's showing both 8 and 10bit formats... e.g. yuv420p AND yuv420p10le; yuv422p AND yuv422p10le etc…
There's no need to specify a profile.
>watching shitty interlaced dvds for Eva
Why not just just download the BDs?
God terminals are awesome.
Because those DVDs are the Archives of Evangelion, this release contains the originally-aired footage of NGE.
install konsole
install source code pro
uninstall KDE