Installation:
mpv.io
Wiki:
github.com
Manual:
mpv.io
User Scripts (including opengl shaders):
github.com
High quality video output profile (goes into mpv.conf):
profile=opengl-hq
mpv.io
Config example:
github.com
Input.conf:
github.com
Post your system specs and config if youre asking performance related questions.
Other urls found in this thread:
sprunge.us
github.com
pastebin.com
github.com
github.com
github.com
github.com
github.com
twitter.com
Tried 2nd, it didn't help, don't think 1st will help either.
can someone teach me how to use this:
To use the Code tag, book-end your body of code with: and
How do i make text inside white box?
I guess in the case of #2 that means your OpenGL compiler is smart enough to notice the redundant work and optimize it away. (Which is quite impressive given the complexity of the sampling expression)
...
>[ code] your text [ /code]
remove the spaces from the tag
why you do this?
I think he's a troll ironically counter-trolling the troll while in reality being trolled by the inverse troll
[ code] this is a test [ /code]
>sprunge.us
Which memes I'm missing?
whos the troll now?
>scale=ewa_lanczossoft
>cscale=haasnsoft
The ones were it doesn't blur your video, madvr.
That's just stupid and messy refactoring. :(
MadVR is bugged as hell since 0.92.x and has poor HDR support.
Its NGU was killed by FSRCNN-LQ and NGU-AA by RAVU.
Its AdaptiveSharpen and SuperRes algorithm is pure shit too.
>Its NGU was killed by FSRCNN-LQ and NGU-AA by RAVU.
even though is bait, it was a funny read
So mpv is just a fork of mpv.net because you thought the context menu was "bloat"? Well, I don't agree, but I'd fight to the death to defend your freedom to think so.
Just retry now before shitposting.
i have mpv installed on my hackintosh and on my macbook pro. i installed both with homebrew cask and i'm using the same .config/ directory and everything. as far as i can tell, the setups are identical.
on the hackintosh, when i open a video, the mpv icon shows up in the dock and the video window shows up. if i open a *second* video, the window changes to the second video. this is all pretty mundane — it's exactly how you'd expect a video player to behave (and how VLC and other players work in os x).
the problem is that on my rMBP, when i open a second file, a second mpv icon shows up in the dock. i can quit the first mpv instance, but the second video never plays. the second (and now only remaining) mpv icon in the dock becomes unresponsive, and i can't seem to kill the instance. i don't see it in the activity monitor or in htop/top. and, i suspect because there's that lingering mpv instance in the dock, another mpv instance can't be started. i end up having to force reboot.
it's the weirdest fucking thing, especially since mpv works fine on the hackintosh, where i would expect my experience to be buggier. i think i've generally ruled out the configuration and most everything else, but that leaves me with no idea what could be wrong.
does anyone have any ideas?
With 1st and 2nd, no difference.
pastebin.com
HW accel on or off lads?
My ISP is hosting some sort of video streaming site, And for some reason mpv can play the videos, however it get the 720p version of the video, while the browser player can get different resolutions+CC.
When I run mpv with --msg-level=all=trace it fails to play the file while playing it with --msg-level=all=v It plays, Why is that?
And I get this from it
[ytdl_hook] multi-arc video detected, building EDL
Any idea if I can get the CC or different resolutions as well?
Also I get this [ffmpeg] http: HTTP error 416 Requested Range Not Satisfiable
Also I get this erro
off
Did we ever figure out if angle or dxinterop is better now?
>we
hello plebbit
hello neckbeard
For me your v1 is 15k ms, v2 is 13.1k ms and v3 is 12.7k ms. You could probably also re-order your convolutions so you iterate over ‘x’ in the inner loop for your samples as well
Can you paste the fragment shader version as a comparison?
dxinterop is better for me.
Also you're still duplicating all of your weights. You should do the convolutions for each group of pixels simultaneously and re-use the vec4 for each dot(). If you search for a string like “0.03783298283815384, 0.03939036279916763, 0.004978751298040152, -0.010442780330777168” it will show up 4 times.
Although I think if you're lucky, the code gen might be smart enough to compress those down to one.
>For me your v1 is 15k ms, v2 is 13.1k ms and v3 is 12.7k ms.
With vulkan btw
with opengl your v1 is 18kms, v2 is 14kms and v3 is 13.7k ms
That's slower than your v2 and v3 for me
Interesting, I'll test it out with my i7 and GTX 1060.
For me fragment shader - ~1.5
v1 - ~5.5
v3 - ~32
When I pause a 1080p video for a couple of seconds and then unpause it, I get around 140 framedrops in a row and then it plays back flawlessly again.
What the hell is going on here?
>rename the built-in profile `opengl-hq` to `gpu-hq`
what does this option do?
sub-ass-force-style=Kerning=yes
a n i m e - h q
...
free and open source software
allows you to override style parameters of ASS scripts
With group size 80x20
//!COMPUTE 80 20 40 10
v1 - ~4.5
With bigger group sizes:
>error C5041: cannot locate suitable resource to bind variable "inp22". Possibly large array.
What hardware and drivers? Also slightly faster for me is switching to 16x16 instead of 32x8, but not by much
dxinterop is extremely gimped in `master`; it sets the frame latency to 1 which prevents pipelining across frames.
will be fixed in the PR
nvidia 660 384.69
nvidia? Check your power plan in nvidia settings. Should be Adaptive or better.
Current Mode: Adaptive
I'll try and mess with these settings.
I don't know if it's placebo but it doesn't seem to happen that badly when I enable interpolation.
madshi, are you ITT right now?
Yes my son
Maybe it's a Kepler vs Maxwell difference?
Whens the next BIG feature?
is it possible to edit the parts of the On screed display?
i want to remove the bar that containts the file name on top of the seekbar.
Yes.
Seems to work flawlessly now with Prefer Maximum Performance.
Will these power settings persist or do I have to add something to xorg.conf?
>Also you're still duplicating all of your weights.
They are not duplicated, there is an index at the end of each mat4
how?
how about we remove everything except the seek bar and the time on the left.
they presist
i changed it to a box. lol. this is getting good. he he he
MadVR will feature mpv integration with opengl-cb.
You can do that without editing any lua file with a config option.
>mpv
>doesn't recognize the taskbar on windows
>every other media player does
into the trash!
well, i just did that.
ill stick to this slimbox for awhile.
The Apple Macbook Pro is the official machine for mpv.
Can I lend hardware to train shaders?
Put autofit-larger=90%x90% into mpv.conf.
github.com
>cscale=ewa_lanczos
Is this placebo? I tried some other cscales and don't notice a single difference.
If you dont notice anything then dont use it.
Is there a way to get frame timings only in the dedicated page in stats.lua?
i dont like this. i want to edit the existing OSC. not overwrite the already existing one.
then edit the osc.lua and put it in your scripts folder with osc=no in your config.
Delete line 487.
Thanks. I don't like looking at broken things.
>i dont like this.
isn't this what you wanted?
actually yes. but not like this.
i prefer to edit the existing osc. i'll deal with this tommorrow. i need to sleep.
just use layout=slimbox
yes. thats what im currently using. i'll stay on this until i find a way to edit the osc.
What config do I use to get Honma healthy quickest?
>sudo apt-get install mpv
>"drag video here to play"
>no playlist functionality to be found
wtf
> expecting a modern media player to have basic functionality in 2017
top fucking kek user
go back to VLC
>get mpv (preferably by compiling it yourself)
>get gnome-mpv
There you go.
why should i use this instead of vlc?
Are people not double clicking files anymore?
Why go through the extra effort of draggin the file into a window instead?
Image quality and cpu usage
What is placebo config?
am i retarded or the "Best quality" config files look more like a placebo more then anything ?
found the retard
You shouldn't if you have to ask this question
Im trying to build mpv from the github repo but the waf config says that it didnt find any opengl outputs. What should I install to solve this?
I use fedora with xmonad as wm
sudo dnf builddep mpv
profile=opengl-hq
scale=ewa_lanczossharp
cscale=ewa_lanczos
# whatever you prefer
opengl-shader=~~/shaders/SSimSuperRes.glsl
#opengl-fbo-format=rgba32f
#opengl-shader=~~/shaders/FSRCNN_16-4-4_x2.glsl
# enable if you need to
#interpolation
interpolation-threshold=0.2
tscale=oversample
video-sync=display-resample
temporal-dither=yes
deband-iterations=2
deband-range=12
vd-lavc-dr=yes
vd-lavc-threads=16
# optional
hwdec=no
opengl-pbo=yes
I don't know if you can go more placebo than that.
I hope you never get into marketing.
Because it doesn't work, and spending months in a 24/7 support thread on Sup Forums trying to make it work will mean you "know computers."
What can mpv do that VLC cant?
Is FSRCNN better than Ravu? And where do I get a hold of it?
This thread is making me realize mpv is a meme.
Shit up the board with a 24/7 help thread.
Ravu is excellent in regards to how little computational power it uses, but it is by far not the best.
SSSR >= FSRCNN-LQ > RAVU > NNEDI3
