Install mpv:

Install mpv:
mpv.io/installation/

Wiki:
github.com/mpv-player/mpv/wiki

Manual:
mpv.io/manual/stable/

User Scripts(including opengl shaders):
github.com/mpv-player/mpv/wiki/User-Scripts

input.conf:
github.com/mpv-player/mpv/blob/master/etc/input.conf

Vulkan(Linux only for now):
github.com/atomnuker/mpv

Test vulkan and post logs if it gives you any kind of problems.

Other urls found in this thread:

github.com/agiz/youtube-mpv
sourceforge.net/projects/mpv-player-windows/files/64bit/
addons.mozilla.org/en-US/firefox/addon/watch-with-mpv/
github.com/mpv-player/mpv/commit/e4bc563fd2dcff1059624efb7b948b2886a382ab
en.wikipedia.org/wiki/Lookup_table#Lookup_tables_in_image_processing
nightlies.plex.tv/public-test/plexmediaplayer/public-testing/latest/
github.com/mpv-player/mpv/pull/4575
twitter.com/SFWRedditGifs

Hobbyist player for a hobbyist OS.

...

You know I'm right.

Nope. I use Linux at work.

The only thing that keeps me on Win now is games and foobar. Dont care about madVR anymore. Thanks mpv devs!

What is vulkan and how do I use/install it

Its something mpv devs stole from Madshi!

Disgusting.

Is foozoor a good guy or a bad guy? Ive seen him praising mpv once.

Madshi is a directX shill.

i think he was joking. as in "the mpv devs stole linux from madshi!".

How do you use lavfi i'm trying to use the now depreciated pan?

I'm trying out chromium instead of firefox but I can't find an extension that lets me open mpv with just a right click on any link like in pic related.

What do you chromium bros use?

chromium based browsers cant do that
either use the terminal or drag and drop urls
or keep using firefox like a normal person

>chromium based browsers cant do that
What? Why not?

There's tons of extensions like "Open with Firefox" etc, why couldn't they open mpv similarly?

github.com/agiz/youtube-mpv

As opposed to the profesionnal anime watcher that choses the patrician player, madvr.

mpv is literally created by anime lovers. madshi doesnt give a fuck about anime.

>having to run a server for something so basic
Absolutely disgusting, I might actually vomit.

Dead.

Alive.

No, faggot
sourceforge.net/projects/mpv-player-windows/files/64bit/

lachs0r's just making a build everytime there is an official release. I think there should be one soon.

why does the stats.lua graphs show on lachs0rs build but not shinchiros newest build?

>There's tons of extensions like "Open with Firefox" etc, why couldn't they open mpv similarly?
addons.mozilla.org/en-US/firefox/addon/watch-with-mpv/

is it possible to watch .gif with mpv? i saw a frame, but they are always frozen

Works for me.

Can someone explain how 10-bit depth achieves smaller encodes with apparently better perceived quality. I'm reading that one github explanation, but can't internalize what exactly happens.
It says that LCD displays can only display up to 6-bit depth and use dithering to represent higher bit depths. What exactly happens when passing colour data to the device? Is pixel data passed to the device and the device decides on dithering? Or does dithering have to be hardcoded? Does 8-bit depth get truncated to 6-bit depth for the display?

Why would 10-bit depth not require hardcoded dithering (I assume the bitrate loss from ditching dithering leads to the smaller size here) and 8-bit releases require it? Wouldn't a 10-bit depth also be truncated to 6-bit depth? How does this magic actually work?

the actual graphs that move in realtime?
I'm using hwdec=d3d11va btw

Update stats.lua

You need to download stats.lua
To activate it you press shift+i, to see more stats you press "2".

>LCD displays can only display up to 6-bit depth
This hasn't been true for a long time. Almost all modern panels are 8bit, though some do use different color profiles which can limit their overall number of colors.

The 10bit encode being smaller than an 8bit encode is mostly bullshit, as you'd have to be retarded to apply filters at the encode point. What I mean is that a dithered 8bit gradient IS more data than a non dithered 10bit gradient, but that should never happen, as you should be dithering the 8bit encode ONLY during playback.

>colors truncated
Colors are mapped between spaces in a lookup table.

I will need to do my own comparison I suppose to figure it out, but the github article says that when both encodes are encoded with same settings, both dithered
>The video track of the 8-bit encode has 275 MiB, while the 10-bit encode has no
more than 152 MiB and doesn’t look worse at all
What's the deal here? If both are dithered and 10-bit encode requires extra bit precision, where do the space savings come from?

Similarly, if 10-bit encodes do not reduce the file size, what is the point of using them? Reduced banding? Does the encoder somehow deduce/simulate the extra color information required to represent color gradients better?

>Colors are mapped between spaces in a lookup table.
Any technical term for this so that I can read up more? Does the device color buffer (or whatever you call it) store color values as 10-bit, even when only 8-bit depth can be represented? Do the values in lookup table not get truncated to make them suitable for display?

TN panels are limited to 6-bit.
Modern TN panels are 6+2-bit(6-bit panel with 8-bit dither).
IPS panels can be 8 and 10-bit, some 4k monitors that advertise 10-bit color depth are just 8+2-bit(8-bit ips panels with 10-bit dither).

>Does the encoder somehow deduce/simulate the extra color information required to represent color gradients better?
The encoder can apply filters (debanding, etc) in 16-bit before dithering to 10-bit.
But even without filters a 10-bit encode from an 8-bit source is preferable because 8-bit encodes are prone to worsening source banding or outright introducing it, especially at low bitrates.

i want high quality and smooth experience, or just smooth experience

give me the best you've got

For a "smooth experience"
interpolation
video-sync=display-resample
tscale=oversample
tscale-radius=3.0

>tscale-radius=3.0
?

It provides better interpolation quality, I forgot to mention that I have my monitor overclocked to 72hz.
tscale is limited to a radius of 3

how many hours a day do you dwellers spend on configuring mpv?

What do you mean by "better" ? Legit curious.

My config is always open.

oversample is too blurry.
tscale-radius=3.0 makes it a bit sharper.

mine too

>tscale-radius=3.0 makes it a bit sharper.
Any tradeoffs?
Almost 20.

opengl-shader=1.glsl
opengl-shader=2.glsl
opengl-shader=3.glsl
Who the fuck though this was a good idea?

github.com/mpv-player/mpv/commit/e4bc563fd2dcff1059624efb7b948b2886a382ab

I think its good but it seems that they reverted this already.

None.

This is my full scaling conf:
# Scaling
scale=ewa_lanczossharp
cscale=ewa_lanczossharp
dscale=mitchell
tscale=oversample
tscale-radius=3.0
correct-downscaling
sigmoid-upscaling
sigmoid-slope=10.0
linear-scaling

Where is video-sync=display-resample ?

dscale=mitchell
correct-downscaling
sigmoid-upscaling
linear-scaling
If you use profile=opengl-hq you can delete these.

>Test vulkan and post logs if it gives you any kind of problems.
I did and resizing crashes on x11 and memeland is not ready for the desktop.

>no build for 2+ months
Did they at least fix stuttering on macOS?

In the # Other section

How do I bind seeking to start of the file?
Tried set start 0 but it returns an error.

Meant for

Where does this extra color information come from? In case of anime BDMV, aren't most of them 8-bit?

seek 0 absolute

>sigmoid-slope=10.0
cargo cult meme

All the anime BDMV are 8-bit, and color information come from the filters.

post everything related you've got to video quality

Test it you dumb fuck.

>cargo cult meme
What's wrong with it?

Test for what? You think it improves quality? Post proof then. sigmoid-slope=10.0 is a meme.

Doesn't work. But fuck it, it is easier to seek minus 99 hours so I just did that.

what about set playback-time 0

# Video
vo=opengl
profile=opengl-hq
deband=yes
deband-iterations=2
deband-range=12
deband-grain=0

# Scaling
scale=ewa_lanczossharp
cscale=ewa_lanczossharp
dscale=mitchell
tscale=oversample
tscale-radius=3.0
correct-downscaling
sigmoid-upscaling
sigmoid-slope=10.0
linear-scaling

# Other
opengl-backend=angle
angle-max-frame-latency=1 #Lower values should make VSync timing more accurate.
hwdec=dxva2-copy
vd-lavc-threads=16
ad-lavc-threads=16
interpolation
video-sync=display-resample

That one works. Thanks.

*I don't use any opengl-shaders and I use "vd-lavc-threads=16" because I have a Ryzen 7 1700

I just tried n seek 0 absolute and it works for me.

Any way to compare 8-bit and 10-bit encodes using avisynth with something like Interleave? Would the depth of one encode get converted to another depth or can avisynth keep them separate?

>avisynth
avisynth doesn't have real 10-bit support, use avisynth+ or vapoursynth

Post logs and try again with --vulkan-direct=no

>Any technical term for this so that I can read up more? Does the device color buffer (or whatever you call it) store color values as 10-bit, even when only 8-bit depth can be represented? Do the values in lookup table not get truncated to make them suitable for display?

A color LUT / colormap; en.wikipedia.org/wiki/Lookup_table#Lookup_tables_in_image_processing

>What's the deal here?

mostly explained it, the caveat is that this is only valid when you have low bitrates and dither at the encode point. Arguably, the CORRECT way to do video encode / playback is to apply filters* (such as debanding) only at playback, thus avoiding this situation entirely and making 10bit color mostly a meme.

*some filtering and alterations are better applied at / before encode, such as color correction, degrain, deghost or other alterations to "correct" the source material. To be clear though, this doesn't include visual enhancements like debanding or antiringing.

2nd ref should point to

Can someone explain ign's SSIM to me please? In madvr you could chose SSIM as a downscaler but in mpv SSIM is used after downscaler? So ign's version is not a downscaler but something like enhancer to downscaler?

Are igv's shaders working with this plexmediaplayer macosx release?
nightlies.plex.tv/public-test/plexmediaplayer/public-testing/latest/

OMG! It seems haasn found a way to make mpv really faster!

> and now I've gotten scale and cscale efficient enough to the point where I can stop using auto-profiles and apply my full on meme placebo scalers to all resolutions \o/

How? Vulkan?

github.com/mpv-player/mpv/pull/4575

Nah, but it probably means: vulkan is even more useless now.

...

>Can someone explain how 10-bit depth achieves smaller encodes with apparently better perceived quality.
More efficient encoding of gradients, because you can just encode a 10-bit gradient without noise whereas for 8-bit you need to dither

>Why would 10-bit depth not require hardcoded dithering
Because you can dither during playback

So, he'll just use haansblur for everything now?

Why do you need dither 8-bit but not 10-bit if the display itself cannot display the bit depth? Wouldn't you need to dither both at playback to simulate the extra bit depth that display cannot handle?

The display handles dithering from 8 to 6 bit (assuming it needs to)

Sure, if you don't care about cpu usage, gpu usage, you bank account, not wanting desync when you alt+tab, whether you run free software, are a slave of the gpu mafia and whether or not throwing more power at an issue doesn't solve your problems anymore. That's a lot to not care about.

Why do you need to dither the 8-bit encode then as stated if it is implicitly handled by the display? In fact, wouldn't you need to dither the 10-bit instead of the 8-bit encode to simulate the 10-bit depth?

Are there any drag and drop based playlist plugins. mpv is literally shit for watching a series because of lack of manual playlist support.

click the user scripts link in the OP

>Test for what? You think it improves quality? Post proof then. sigmoid-slope=10.0 is a meme.
It had very slight improvements or more like less artifacts as shown in various comparisons in many old threats.
Granted, I don't really see the difference without eye balling an image. However, it doesn't cost anything, so why not?

because 8 bits are insufficient to cover the range of human vision, we need at least 10 bits (standard gamut) or 12 bits (HDR / wide gamut).

if you try encoding a gradient 8 bits undithered, it will look like shit (banding), so you need to compensate by adding more dithering/noise (expensive to encode)

but you can just encode a gradient as 10 bits for free because the rounding artifacts are within the limits of human perception

What about Intel?

>not wanting desync when you alt+tab
works fine for me with opengl

honestly the most important improvement vulkan would bring with it is the ability to be cross-platform, which opengl can't really satisfy (because microsoft hates it)

Try it and find out?

because if its slower on intel or amd who gives a fuck, these poor subhuman bastards can't even run the supreme upscalers out there
but hey, memcpys are free, right, lets have more of 'em
some guy wrote some amazing assembly for in-place intraprediction? bah, gotta find more places to memcpy, my cpu is getting cold and my electricity bill is down a percent
atomnuker: shitposting is free, writing code is not
either contribute improvements or shut the fuck up
I'm sick of yo
you
well I'm sick of you too
fyi I wrote over 2000 lines of code for vo_vulkan and I do not enjoy constant reminders of how I should just give it up
What's this beef about? Is the new code really slower for AMD/Intel?

EVERYONE STOP FIGHTING!!!!!!!!!

Nah that's always good for foss, atomnuker should create his own optimized vulkan/linux fork to beat the madvr supremacy.