/mpv/

Why is madshi using mpv?

--hwdec=nvdec-copy
--profile=gpu-hq
--scale=ewa_lanczossharp
--cscale=ewa_lanczossharp
--video-sync=display-resample
--interpolation
--tscale=oversample
--gpu-api=vulkan
--gpu-debug
#--gpu-context=d3d11

log it
mpv.io/manual/master/#options-log-file

Why are you still using legacy sharpeners when there are better alternatives?

So I just installed MPV on my PC for the first time because I kept having an issue with MPC-HC where AAC audio would consistently cut out for anywhere from 3-10 seconds every so often.
MPV plays the audio perfectly but the video is slow and laggy and constantly drops frames in high speed sequences, while MPC had perfect video. Why is this?

it's hard to say without any information, do you have an old PC?

Doesn't work.
Everything looks blurry without it.

Yeah, it's an old piece of crap laptop from 2012 but I never had these problems until about a month ago. At first I thought it was an issue with my files so I redownloaded but it didn't do anything of course.
I'm in the middle of downloading a torrent right now so granted I haven't tried running MPV and nothing else but I wouldn't think downloading at 40kbps is really going to cause MPV to chug like this.
Seems like the skipping audio only happens with 6.1 AAC mixes, but I doubt it's a codec issue as MPV works perfectly for the audio aspect.

if log doesnt work, you are doing it wrong.

do you have a config file? are you using gpu-hq? if you're not sure, press shift+i and then 2 and check if mpv is doing spline36 upscaling and debanding.

Here's what I came up with, I don't know much about video decoding so it's all Greek to me.

Attached: mpv.png (1105x534, 968K)

SSSR is better
Potato PC, remove profile=gpu-hq

I used both at the same time and everything is blurry without adaptive sharpen.

no wonder you're dropping frames. do you have a config file? if not
>create a folder named portable_config in the same folder as mpv.exe
>create a text file
>rename it to mpv.conf
>open it up in a text editor
>paste this and save
scale=bilinear
cscale=bilinear
dscale=bilinear

>remove profile=gpu-hq
Done, and while it isn't chopping up nearly as badly as before it's still doing it a lot. Is my PC really that shitty?

Attached: specs.png (678x469, 94K)

I compared them and SSSR is slightly sharper at default strength. If default strength is not sharp enough just increase it.

It should work fine with default settings, they are the same as default MPC-HC.

I compared everything so much that my head hurts. I know it's subjective but can you please share your config? Maybe it will lead me somewhere.

--linear-downscaling
Scale in linear light when downscaling. It should only be used with a --fbo-format that has at least 16 bit precision. This option has no effect on HDR content.
--linear-upscaling
Scale in linear light when upscaling. Like --linear-downscaling, it should only be used with a --fbo-format that has at least 16 bits precisions. This is not usually recommended except for testing/specific purposes. Users are advised to either enable --sigmoid-upscaling or keep both options disabled (i.e. scaling in gamma light).
Let it on or off? I don't trust haasn.

>can't open multiple files in a playlist with nautilus
why are gnome devs so retarded?
gitlab.gnome.org/GNOME/nautilus/issues/117

profile=gpu-hq, latest SSSR with locality set to 10 and simoid-upscaling=no

wouldn't bash alias work?
alias mpvpp='mpv --pause --reset-on-next-file=pause'
pp as in pause pause

Sorry for the stupidity but which parameter do I need to put on 10?
#define locality 1000.0
#define spread 1.0 / locality

>#define locality 1000.0

>ewa_ginseng
is this good for any of the scale options? man mpv says to use ewa_lanzossharp by default, but i'm playing with some of the others. which would you recommend:
Available scalers:
bilinear
bicubic_fast
oversample
spline16
spline36
spline64
sinc
lanczos
ginseng
jinc
ewa_lanczos
ewa_hanning
ewa_ginseng
ewa_lanczossharp
ewa_lanczossoft
haasnsoft
bicubic
bcspline
catmull_rom
mitchell
robidoux
robidouxsharp
ewa_robidoux
ewa_robidouxsharp
box
nearest
triangle
gaussian

ginseng seems to use the least cpu so far

>ginseng seems to use the least cpu so far
none of them use cpu...

Thank you very much.

well whatever frame timings info screen shows as a percentage

On for downscaling, off for upscaling. Don't use sigmoid too.

ewa_* scalers are the heaviest and pretty much placebo, just use default scaler from gpu-hq

Any osc that can match the title bar in mac os mojave?

Attached: mpv.png (3104x1844, 3.72M)

github.com/422658476/MPV-EASY-Player/blob/master/mpv-easy-data/osc-style/osc-iina-box-knob-or-bar-15.lua

you have to change the layout to box or it's the same as the default mpv bar

Is there a way to apply configs to certain videos? I want default options applied when I play youtube links, or everything else, but I want gpu-hq, scaling options, debanding etc applied when I'm playing anime
All those options make a difference watching anime stuff, but I'd rather not waste power or cycles on live action things.

auto-profiles.lua
You can set up a profile that applies to every video where "anime" is in the file path.

I've been speedwatching through various lectures, and there's probably a way to watch (or rather, listen) to them faster. I've played the same video on mpv and one on native youtube player, and there at 2x is still easy to understand, but ] at 1.61 on mpv gets hard to comprehend.
Is there something special youtube is doing, and can this be applied to mpv too?

Maybe trying messing with the librubberband options? github.com/mpv-player/mpv/blob/master/DOCS/man/af.rst that's what mpv uses for pitch correction. I dunno what youtube does differently.

gpu-api=vulkan
vulkan-swap-mode=fifo
vulkan-queue-count=4
vulkan-async-compute=yes
gpu-shader-cache-dir="~~/shader-cache"

fbo-format=rgba32f
icc-profile-auto=yes
icc-cache-dir="~~/icc-cache"
hdr-compute-peak=yes

#Scale
scale=ewa_lanczossoft
cscale=haasnsoft
dscale=catmull_rom
correct-downscaling=yes

#Deband
deband=yes
deband-grain=0
deband-iterations=2
deband-range=12
deband-threshold=64

#Dither
dither-depth=auto
temporal-dither=yes

#Interpolation
interpolation=yes
tscale=linear
blend-subtitles=yes
video-sync=display-resample

Madshi is a winfag and since windows mobile is dead he has to use an android phone. Nothing can't beat mpv on android and madvr is too windows related to ported. :(

Anyone have any idea why the new autoload.lua in the scripts list isn't working?

No unit tests or code quality checker like a lot of C projects. :(

how can I disable moving/resizing the window? I want it for my ytdl profile pip mode.

works for me in linux

I'm on windows.
An older version of it worked for me but it did some weird things when the numbers weren't properly formatted like skipping from episode 10 to episode 20.

I just started using loonix, so I'm not sure what to do here.

I've set up mpv and copy pasted my config file from my old install but it's not reading it, I assume it's some issue with root. It reads it if I do sudo mpv but I don't wanna run root every time I wanna watch something. How to fix?

onakaitai@onakaitai:~$ mpv
[file] Cannot open file '/etc/mpv/input.conf': Permission denied
Failed to open /etc/mpv/input.conf.
[input] Can't open input config file /etc/mpv/input.conf.

Thanks in advance

Attached: seagull.png (1920x1258, 1.98M)

Is linux dead now that M$ is everywhere?

--no-window-dragging
For resizing I don't know.

open your file browser
ctrl+h to show hidden folders
open the .config folder
make a folder called mpv if it doesn't exist
put mpv.conf and input.conf in there

wtf it's in /etc? I don't have a mpv folder in /etc.

[protocol.ydl] isn't working, what am I doing wrong?

Is Baka-MP a good frontend?

When I installed it on xubuntu it created one in /etc/mpv/

I used this repo
sudo add-apt-repository ppa:mc3man/mpv-tests

Attached: Capture.png (1353x278, 51K)

the user specific one is ~/.config/mpv/mpv.conf

First time using mpv, what settings are essential/recommended?

are all of you autistic?

only a mild case of aspergers

profile=gpu-hq

where do you think you are?

yes

>N af toggle lavfi=[dynaudnorm=f=200:g=5:r=0.1]
Nice

Attached: Slender.Man.2018.720p.BluRay.x264-DRONES_00_37_22.jpg (1804x755, 355K)

Is there a way to create thumbnails / multiple screenshot images using mpv?

Extra points if there is a batch process that can do it for a whole folder, for example.

Attached: thumbs20071204094442.bmp.jpg (600x462, 95K)

github.com/occivink/mpv-gallery-view

Are any of the ffmpeg filters good at removing background noise like low frequency humming/static typical on low quality audio recordings?

>4k 60 with HDR downloaded Youtube video

My Vega FE GPU is literally at 99% usage trying to apply all the post-processing using Vulkan and Wayland. Someone please tell me I am doing something wrong and my GPU is supposed to be able to handle this.

Attached: Screenshot from 2018-10-20 18-56-08.png (1920x1080, 2.22M)

I get high CPU usage on W10 with vulkan. Not sure if my GPU usage is normal. It's a laptop with an i7-7700HQ and GTX 1050

>vulkan
CPU: ~60-70%
GPU: ~60%

Attached: vulkan.jpg (3840x2160, 2.47M)

D3D seems more sane. Maybe I had the vulkan settings wrong?

>D3D11
CPU: ~3-4%
GPU: ~60%

Attached: d3d11.jpg (3840x2160, 2.58M)

That is not the same as mpc-hc

n af toggle "lavfi=[highpass=f=200,lowpass=f=3000]"
is the best i could find. it would be cool if you could add/decrease both values at runtime to fine tune it. i bet it is doable with a lua script.

Are you doing any sort of debanding/luma upscaling and etc? Because I can decode 8k without doing post-processing but as soon as you start doing any sort of post-processing on it, shit goes bad real quick.

>tfw linux doesnt have vp9.2 (hdr) hw acceleration on pascal

Attached: screen-watch?v=LXb3EKWsInQ&t=38s_2018-10-21-00:02:52.jpg (3840x2160, 1.89M)

good av1 when

d3d is using hwdec.

how do i get mpv to do the video output screenshot? pressing s/S gives pic related

Attached: screen-2018-10-20-22.09.34.png (1690x1761, 2.5M)

I used profile=gpu-hq
video-sync=display-resample for both

I get that. But why is the GPU usage the same on both then? Resource usage in total is way higher with vulkan.

I used ctrl + s

it works fine if i turn off everything vulcan related

forgot pic

Attached: screen-watch?v=LXb3EKWsInQ&t=38s_2018-10-21-00:02:23.jpg (3840x2160, 3.42M)

Oh yeah, I forgot ctrl + s doesn't work with vulkan. I had to use the prtscr key.

Attached: spaceghostannoyed.jpg (640x360, 37K)

just tried it too. seems vulkan doesn't do tonemapped screenshots for some reason.

ctrl+s does screenshot window, so if you screenshot a 4k video fullscreen on a 1080p monitor you will only get a 1080p screenshot. theres not really any benefit to using ctrl+s now that regular screenshots with s have filters and tonemapping.

>now that regular screenshots with s have filters and tonemapping.
well I used s and S with vulkan and you see the results:
without vulkan:
hdr shows active on both
i have a 4k screen

I use vcs on linux. Search video contact sheet. There is a bash script, and python scripts that should run on windows. You can make a keybind for it in mpv like x run /path/to/vcs ${path}

Attached: vcs.jpg (1813x1374, 598K)

is there any quality reduction from using nvdec?

around one of the lastest updates of the git version of mpv i started getting frame drops, yet i dont see this in smplayer, what could be the cause? did the mpv devs fuck up something? my config is very simple
volume-max=200
profile=gpu-hq
--script-opts=osc-layout=box
keep-open=yes
--loop-file

no
deband is slightly slower now.

anybody? I have ytdl written in my profile and not ydl.

as far as i know protocol.ytdl is only active when you explicitly open a url or file beginning with "ytdl://"

Hi, does mpv build in lua load file with symbolic link?
my setting right now have a link
mpv/scripts/abc.lua -> mpv/scripts/def/abc.lua
but the console shows error
Lua error: mpv/scripts/abc.lua:1: unexpected symbol near '!'
if I replace the link with real file, it will pass.

Fuck I've not been paying attention to mpv for a while and have just seen in the example for FSRCNNX using all FSRCNNX 16's at the same time, is that how it's supposed to be used or just a demonstration that it can be used like that?
I always assumed x2 meant doubling, x3 tripling and x4 quadrupling, was I wrong?

a demonstration that it can be used like that

I don't understand how to do custom screenshot filenames. I feel like there should be no way I could fuck it up, but I am somehow.

mpv.io/manual/master/#options-screenshot-template
>--screenshot-template=
>Allowed format specifiers:
>%[#][0X]n

I literally try --screenshot-template=%[#][0X]n and it proceeds to do nothing. I already changed the screenshot directory location and specified the output to be png, so it's not like I'm editing the wrong file or anything. I can even just put keyboard spam in place of the templates it provides, and they work. I don't understand what I'm doing wrong in using the templates, can someone please spoonfeed me on this.

my mpv.conf
path: %AppData%\mpv\mpv.conf

screenshot-template="%F_%p_%04n"
screenshot-directory=C:\Users\your_name\Pictures\