Installation:
https://mpv.io/installation/
Wiki:
https://github.com/mpv-player/mpv/wiki
Manual:
https://mpv.io/manual/master/
User Scripts (including opengl shaders):
https://github.com/mpv-player/mpv/wiki/User-Scripts
High quality video output profile (goes into mpv.conf):profile=opengl-hq
https://mpv.io/manual/master/#configuration-files
Config example:
https://github.com/Argon-/mpv-config/blob/master/mpv.conf
Input.conf:
https://github.com/mpv-player/mpv/blob/master/etc/input.conf
Post your system specs and config if youre asking performance related questions.
>>62476730
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?
>>62477068
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)
>>62477110
>[ code] your text [ /code]
remove the spaces from the tag
>>62477137
why you do this?
>>62477147
I think he's a troll ironically counter-trolling the troll while in reality being trolled by the inverse troll
>>62477137
[ code] this is a test [ /code]
>http://sprunge.us/RSed
Which memes I'm missing?
>>62477137
>>62477155whos the troll now?
https://github.com/mpv-player/mpv/pull/4880
SOON™
>>62477166
>scale=ewa_lanczossoft
>cscale=haasnsoft
The ones were it doesn't blur your video, madvr.
>>62477201
That's just stupid and messy refactoring. :(
>>62477292
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.
>>62477333
>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.
>>62477345
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?
>>62477133
With 1st and 2nd, no difference.
https://pastebin.com/cd06fMfc
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
>>62477398
off
Did we ever figure out if angle or dxinterop is better now?
>>62477529
>we
hello plebbit
>>62477542
hello neckbeard
>>62477396
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?
>>62477529
>>62468001
dxinterop is better for me.
>>62477580
https://github.com/igv/FSRCNN-TensorFlow/releases/download/0.2/FSRCNN_16-4-4_x2.glsl
>>62477396
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.
>>62477580
>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
>>62477595
That's slower than your v2 and v3 for me
>>62477591
Interesting, I'll test it out with my i7 and GTX 1060.
>>62477633
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?
>>62477201
>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
>>62477781
>>62477767
free and open source software
>>62477816
allows you to override style parameters of ASS scripts
>>62477702
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.
>>62477702
What hardware and drivers? Also slightly faster for me is switching to 16x16 instead of 32x8, but not by much
>>62477529
dxinterop is extremely gimped in `master`; it sets the frame latency to 1 which prevents pipelining across frames.
will be fixed in the PR
>>62477933
nvidia 660 384.69
>>62477767
nvidia? Check your power plan in nvidia settings. Should be Adaptive or better.
>>62478106
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?
>>62478216
Yes my son
>>62478007
Maybe it's a Kepler vs Maxwell difference?
>>62478235
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.
>>62478335
Yes.
>>62478106
>>62478214
Seems to work flawlessly now with Prefer Maximum Performance.
Will these power settings persist or do I have to add something to xorg.conf?
>>62477602
>Also you're still duplicating all of your weights.
They are not duplicated, there is an index at the end of each mat4
>>62478386
how?
how about we remove everything except the seek bar and the time on the left.
>>62478429
they presist
>>62478450
Dunno. Learn lua.
https://github.com/mpv-player/mpv/blob/7897f79217af1e04e6e65bd72e938058e84c451a/player/lua/osc.lua
>>62478478
i changed it to a box. lol. this is getting good. he he he
>>62478304
MadVR will feature mpv integration with opengl-cb.
>>62478490
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!
>>62478541
well, i just did that.
ill stick to this slimbox for awhile.
>>62478563
The Apple Macbook Pro is the official machine for mpv.
Can I lend hardware to train shaders?
>>62478563
Putautofit-larger=90%x90%into mpv.conf.
>>62478450
https://github.com/torque/mpv-progressbar
https://github.com/haasn/gentoo-conf/blob/xor/home/nand/.mpv/config
>cscale=ewa_lanczos
Is this placebo? I tried some other cscales and don't notice a single difference.
>>62478944
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?
>>62478885
i dont like this. i want to edit the existing OSC. not overwrite the already existing one.
>>62479238
then edit the osc.lua and put it in your scripts folder with osc=no in your config.
>>62479182
Delete line 487.
>>62479297
Thanks. I don't like looking at broken things.
>>62479238
>i dont like this.
isn't this what you wanted?
>>62479659
actually yes. but not like this.
i prefer to edit the existing osc. i'll deal with this tommorrow. i need to sleep.
>>62479238
just use layout=slimbox
>>62479834
yes. thats what im currently using. i'll stay on this until i find a way to edit the osc.
>>62476602
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
>>62480257
> expecting a modern media player to have basic functionality in 2017
top fucking kek anon
>>62480257
go back to VLC
>>62480257
>get mpv (preferably by compiling it yourself)
>get gnome-mpv
There you go.
why should i use this instead of vlc?
>>62480257
Are people not double clicking files anymore?
Why go through the extra effort of draggin the file into a window instead?
>>62480523
Image quality and cpu usage
What is placebo config?
>>62476602
am i retarded or the "Best quality" config files look more like a placebo more then anything ?
>>62480257
found the retard
>>62480523
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
>>62480581
sudo dnf builddep mpv
>>62480552profile=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.
>>62480572
I hope you never get into marketing.
>>62480523
Because it doesn't work, and spending months in a 24/7 support thread on 4chan trying to make it work will mean you "know computers."
>>62476602
What can mpv do that VLC cant?
>>62480644
Is FSRCNN better than Ravu? And where do I get a hold of it?
This thread is making me realize mpv is a meme.
>>62480745
Shit up the board with a 24/7 help thread.
>>62480749
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
https://github.com/igv/FSRCNN-TensorFlow/releases/
>>62480817
>needing a help thread for a media player
>>62480833
Oh that's neat, thanks.
>>62480745
make people butthurt >>62480817
>>62480948
vlc can do it too.
>>62480948
>be the only software corporation that uses 4chan as its official tech support forum
>wonder why users are mad
>>62480987
>see thread that upsets you
>click on thread
>get more upset
>reply to the thread and every reply you get bumps it again to the first page causing you to get madder and expose the thread to more people
What causes this behavior?
mpv is finished
>>62481059
>make thread on a board it doesn't belong
>keep making thread
>get told it doesn't belong
>keep making thread
>get told it doesn't belong
>keep making thread
>get told it doesn't belong
>keep making thread
What causes this behavior?
>>62481059
>it's ok because this totally fits here
>all this attentionwhoring is ok
>>62481093
>it doesn't belong
According to who? You?
>>62481059
>literally admit the purpose is advertising
>>62481084
I do wish mpv had better screenshot functionality. Having to disable subtitles to take a windowed screenshot just to get a shot of how the video is actually rendered is annoying.
>>62481131
>enabling subtitles in the first place
pleb
I wish they had support for lossless webp screenshots.
>>62477382
anyone?
>>62481131
How do you add a sub file to a video?
>>62480539
Do you know how many people have requested, no more like demanded, a "File -> Open" menu item because (apparently) opening a video player and then using said menu is the way they open stuff?
There have been so many over the time... that was also one of the incentives for the "Drop files here" screen so these people have a way to open files from the player (they refuse to double click files, the want to start the player first)
>>62480724
Nobody here is trying to make you purchase something though.
>>62481093
>discussing software and it's development is not technology
okay sweetie
>>62481241
This is an advertising thread, and you keep your tech support here to keep it bumped.
>>62481274
no you keep it bumped
>>62481150
It's not just subtitles though, I just used that as an example. The OSD, OSC, etc also show up in window screenshots so you have to make sure those aren't in the way either. Whereas the normal screenshot commands ignore everything but (optionally) the subtitles and takes the shot at native res. Which is good and convenient but the actual screenshots produced are lower quality.
>>62481208
I always thought the drop files here screen was just an inside joke for mpv devs.
I really wonder how many people actually use it.
Why doesn't mpv have telemetry and data collection? If they wanna keep up with other media players they better add it :^)
Anyone who renew your mpv pass in the next hour will get access to test vulkan.
>>62481335
thank you plex-sama i will be sure to advertise your services on /g/ in the future
>>62481335
Does mpv still have that jew donate button on the windows builds?
gtx 750ti, watching 720p,1080p anime and 1080p live action, is this good?profile=opengl-hq
opengl-backend=dxinterop
opengl-shaders="~~/SSimSuperRes.glsl"
scale=ewa_lanczossharp
video-sync=display-resample
interpolation
tscale=oversample
>>62481456
it got taken off a while ago
halp /G/ I installed mpv and now my internet has 3 toolbars
>>62481498
>tscale=oversample
the biggest meme in the history of mpv
>>62481588
>he fell for the using any tscale besides oversample meme
o lord
>>62481588
What? why? It gets rid of the judder on 24Hz and 25Hz videos without adding any blur like all the other tscales. Sure it's useless if you already have a monitor with an integer multiple frequency but it's perfect for 60Hz monitors
>>62481588
It depends on your monitor but it really is the best overall.
>>62481619
unrelated but someone said interpolation doesn't automatically get turned off if your monitor refresh rate is an integer multiple of the video
is this true??
>>62481703
That's because no computer is perfect and neither is mpv or any other video player. There'll always be dirt in the gears that'll fuck something up leading to judder. Could be something as simple as needing to force your GPU into using higher performance clockspeeds. Could be something like a rogue software in the background causing DPC spikes in your system. Could be something else too.
>>62481498
What display and resolution do you have?
You could add an icc profile.
>>62481498
>interpolation
the biggest meme in the history of mpv
>>62481208
>>62481321
Might be helpful to Windows users. I used to use the "drop files here" screen to copy and drag stream urls to mpv. I switched over to Linux about a month ago so now I just use the terminal instead.
>>62481745
Completely unrelated topic but how do consoels manage to keep everything 100% smooth? Not talking about performance drops. Imagine a game that runs at cosntant 30 without drops. It stays smooth without any visible judder all the time. Is the console trying to vsync the game at all costs including huge ammounts of input lag? I mean, not even mpv knows for sure the true refresh rate of my monitor. It keeps fluctuating between 59.994 and 59.995. I've seen retroarch take an even longer time to measure the refresh rate
>>62481788
I made a custom hotkey that opens mpv with whatever is in my clipboard.mpv "$(xsel -ob)"
I don't like to keep terminals open.
Should I overclock my monitor to 72hz for anime or it's just a meme too?
>>62481838
I used to do that but it's a hassle whenver I want to watch 30 or 60fps videos so I just use oversample. The difference is nigligable, there's just a little more blur on pannings because of the lower refresh rate.
>>62481809
As for knowing the true refresh rate of your monitor that depends on the EDID. It varies from panel to panel. Some show 59.94Hz, some show 60Hz. And about the consoles I'm not quite sure what you mean.
>>62481809
They don't.
>>62481838
If you can I'd recommend it. No placebo config option will ever come close to it.
The difference is night and day to me.
Use this to test if it's actually working.
>>62481917
It's purple with oversample
>>62481941
It needs to be purple without interpolation.
>>62481917
There's a recommended pixel clock range?
>>62481941
For me it's flickering purple and red on 60Hz (with oversample) and constant purple pillar on 72Hz.
You can try and see the difference between these two.
>>62482007
I have a 60hz monitor and this is purple but flickering quite a bit. I can see a difference from the 72hz bar, but is my monitor locked at 59 hz or something?
>>62481917
Do you have a 24 and 25 fps webm for testing?
>>62481995
165mhz
>>62482070
Could be. Read your manual.
>>62482121
I only saved these two from an earlier thread. Didn't make them. There was also one for 50Hz but I wasn't interested in saving it.
>>62481084
WOW. After 5 years madVR users finally can make screenshots!
>>62482434
I think you didn't use madVR ar all.
It was always possible but now there are some meme options.
>mpv gets Vulkan and d3d11 renderer
>madVR can now make screenshots
Cute.
>>62482434
>added extended screenshot functionality
>>62482493
>It was always possible
Yeah, with print screen keyboard button. And then you have to crop it. LMAO
>>62482509
Bullshit! Vulkan is still slower than opengl and there is no available build with d3d11 renderer because it's probably a shit.
>>62482532
Pfff stupid shitposter!
>>62482557
>Vulkan is still slower than opengl
This makes no sense if it's true
>>62482532
He added extended screenshot features, most or them don't exist on mpv. Please!
>>62482603
>>62482557
>>62482578
Why'd you have to post these one after another? :O
>>62482591
Vulkan renderer needs some performance improvements but opengl renderer is already very optimized.
>>62482621
That's how a conversation works. ;)
>>62482660
I imagine a cute anime girl behind your posts, keep going ;)
>>62482591
>>62482628
Opengl does (limited) optimizing for you while with Vulkan you have to do it all. At least I think that is what this image is trying to say.
>>62482836
Yes vulkan is a disaster for opengl addicts and noobs.
>>62480164
What is wrong with Honma?
>>62482836
Given how autistic mpv devs already are, is this a problem?
Worst case scenarion we get Vulkan with similar performance as OpenGL. Best case scenarion we get Vulkan with some better performance.
>>62482974
But we need a d3d12 renderer. :D
>>62480164
>>62482904
What is Honma even supposed to be?
Are you in the correct thread?
>>62483156
>he doesn't know who Honma is
Back to doom9.
>>62483285
What the fuck is doom9?
I predict that vulkan will be in "master" on October 12th and d3d11 on October 22nd.
>>62483347
Multimedia for retards forum.
>>62481335
What?
>>62482434
Can now rather easily compare 720
>>62484185
to 1080
>>62484185
Wat?
>>62484185
>>62484201
>kuckchikirukia
>>62484216
>>62484231
No need to be upset. That one uses ordered chapters so your broken little player can handle it.
>>62484317
HAHAHAHAHAHAHHAHAHAHA
>>62484317
>implying that's the only problem with your rips
>>62484435
>needing a translator note for "I'm not good with tongue twisters"
>>62484317
Holy shit what kind of low life are you that you actually lurk threads of a media player you despise so much lol
>>62484651
He also hates qbittorrent and shits on it in every even barely related thread.
>>62484231
Oh I remember seeing that guy sperg out on PTP because one of his encodes got deleted.
>>62484714
I remember him from "Your Meme" anime. He made like 7 versions of it.
>>62484695
>qBittorrent gets endlessly shilled on /g/
>qBittorrent doesn't work
>backlash ensues
>mpv gets endlessly shilled on /g/
>mpv doesn't work
>backlash ensues
Notice a trend?
>>62484695
maybe two of my most favorite software :D
>>62484201
>>62484185
wtf, why does 720p look sharper and better than 1080p?
am i blind?
>>62484785
It's the downscaling kernel.
>>62484770
I notice a trend that you hate open source software with passion.
>>62484834
No, I hate freetards since you're complete fanbois who will overlook all flaws in your software and invent flaws in competing projects in an attempt to "prove" your superiority for using it. It's a religion, and religion irks me.
>>62484231
I always thought people were memeing but I really had to block him on various trackers because he's such an insufferable cunt
>>62484892
>>62484898
I doubt I know you exist.
How do I crop 1 edge on all sides?
>>62484924
Not sure what the fuck does "1 edge on all sides" mean, but this guy has a crop script:
https://github.com/occivink/mpv-scripts#croplua
>>62484770
>>62484892
Wait, what the fuck is wrong with qBittorrent?
It's literally the best, I've tried them all.
>>62484770
>qBittorrent gets endlessly shilled on /g/
no
>>62484990
different than what i want, thanks however.
>>62485037
It's ugly.
>>62484185
>>62484201
>720p upscaled to 1080p looks better than native 1080p
What sorcery is this?
>>62485107
>set rss downloaders
>set magnets to auto add without confirmation
>never look at it again
I just look in my torrent folder to see what's new and watch it.
script-opts=autospeed-nircmd=true,autospeed-speed=false,autospeed-nircmdc="c:\nir\nircmdc.exe",autospeed-monitor=1,autospeed-dwidth=1920,autospeed-dheight=1080,autospeed-bdepth=32,autospeed-rates="60;23",autospeed-exitrate=60,autospeed-minspeed=0.9,autospeed-maxspeed=1.1,autospeed-osd=true,autospeed-osdtime=10,autospeed-osdkey=y,autospeed-estfps=false,autospeed-spause=4
fs
fs-screen=1
profile=opengl-hq
video-sync=display-resample
Why does this config have a/v desync?
>>62476602
Is there a way to open two videos in sync?
I have a video which I would like to open twice at the exact same time and also where both videos can be paused with the same button.
>>62485161
Don't use autospeed, then it should work.
>>62485037
>>62485274
>confidence in his abilities to make it proprietary
What did she mean by this?
>>62485274
This is what peak autism looks like.
>>62485266
It's so convenient tho =\
Is there a way to adjust audio sync on the fly? Or perform functions similar to autospeed?
>>62485358
Peak autism AND lack of knowledge.
>>62484785
>>62485110
Does that mean 480p scaled to 1920x1080 would look even better?
Sorry why wouldn't you guys just use VLC? Install one thing bam done no?
>>62485745
>Install one thing bam done no?
So mpv.
>>62485745
Sounds like MPC-HC.
What am I looking at?
What is mpv?
What is good about it?
>>62485801
So many questions.
Only god knows.
>>62485801
>What am I looking at?
Honma.
>>62476602
why does no one on /a/ know about mpv?
are you even trying /mpv/?
>>62486264
/a/ is at the mercy of the cartel, and the cartel demands that they use MPC-HC to maintain their iron grip on anime piracy and consumption.
>>62486264
But I browse /a/ and use mpv anon.
>>62486264
/a/ is full of greenhorn weebs.
nothing of importance to discuss there.
they don't even know anything about the shows they are actively watching.
>>62486424
Have you ascended to /jp/?
>>62486264
Lots of retards on /a/ nowadays. The era of tech literacy is over. Normalfags outnumber /a/nons.
>>62486439
>ascended
The only thread I visit there is /djt/, but it has been months since I last did.
I still have some LNs to read before I get bored enough to check it again.
There's a hotkey to hide/show the title bar?
>>62486783
create a bind forcycle border
Is there a way to change the number of seconds the arrow keys skip through?
>>62481150
lavc actually supports lossless webp so enabling it for screenshots would probably be a 2-line change (+ documentation)
>>62481703
>unrelated but someone said interpolation doesn't automatically get turned off if your monitor refresh rate is an integer multiple of the video
Correct; it's also not a no-op (except for tscale=oversample). Disabling it for oversample would be a special case. And it wouldn't even help much. At best, you have to do one fewer sample per pixel. You still need to blit from a cache for the purposes of display sync, so...
>>62481809
Console devs know the exact hardware they're developing for. This allows them to do a few things you simply can't on PC:
1. Pre-compile all shaders so you just need to load the platform-specific bytecode instead of compiling them at runtime, preventing stutter when loading shaders
2. Pre-bake all resources and tune their size so that they don't exceed VRAM, then keep them in memory at all times - prevents any sort of microstutter when needing to swap resources in and out
3. They know exactly how many cores they have and how powerful they are, so they can do things like assigning specific jobs to specific cores at all times, and tuning the amount of ahead-of-time work that's done per CPU core based on how many jobs they know it can process per frame, etc.
4. They only need to test on exactly one implementation/platform/driver version/GPU; so they know exactly how to properly interface with the system instead of having to randomly stab in the dark with hacks like glfinish, glflush, 300 different swap extensions, etc. like you have with OpenGL.
Any time a console has any sort of benefit compared to a PC, it's simply because consoles have standardized hardware, while PC games need to run on everybody's mom's computers if they don't want it to turn into a PR nightmare.
>>62486995
Just to give you an idea of how much this is done, the PS4 Pro has more powerful hardware than the PS4 but pretends to be a base PS4 (clocks and all) in every game that doesn't have specific support for the more powerful hardware. Adding PS4 Pro support to a game is done by literally patching the game to tune it for the new hardware.
Imagine if PC games needed a patch to enable them to work on systems with a higher CPU/GPU frequency and core count.
>>62482509
but can mpv make screenshots with vulkan, huh tough guy??
>>62482591
It's no longer true as of latest commits
>>62484185
>>62484201
720 looks better than 1080p, is this the power of studio upscales?
>>62484898
The guy got himself banned from github as well. Fortunately these sorts of people find few friends in life, so they will probably end up dying due to loneliness
>>62486845RIGHT seek 10
LEFT seek -10
UP seek 60
DOWN seek -60
>>62486816
thanks fren
>>62482904
Some kind of neck or spine issue. But our boy's making a comeback!
>>62480539
I don't like rats.
>>62485037
It doesn't connect very well. It also seem to have a bug at the moment where it will fail to recheck, making it lousy for cross-seeding.
It's fine for pleb-tier downloading from well seeded swarms on public trackers, but that's about it.
>>62487127
And madVR.
>>62487750
I've been using it for a bit and have 4TB upload and around active 400 torrents and never had a problem with rechecking or cross-seeding.
>>62487770
Wait, so you're telling me the shitty one was madVR? holy kek
I guess that means the sharp one was mpv with RAVU?
>>62487840
It's native resolution. You're not even a very good cultist.
>>62487826
>4TB
Are you on dial-up, or have you only been torrenting for a month?
>>62487855
Wait then why did you say “the power of madVR”?
>>62486264
>>62486306
>>62486337
>>62486424
>>62486454
Pretty much every fansub website in existence has a usually pretty huge/long description of how to setup MPC-HC.
A fucking GUI video player.
When the users are so stupid that they can't configure a VLC-level program themselves, then mpv is really not for them.
>>62487990
To be fair, MPC-HC is shit out of the box, so you do at least need someone to tell you what else to install.
>>62476602
fuck is mpv and what makes it so great? the only thing i see when i click on the first link is download links.
>>62487960
Did you not see the 720 isn't 720?
>>62488020
>what makes it so great?
It's free as in freedom.
>>62488019
KCP, CCCP, and K-Lite take care of that. Alas, mpv needs a 24/7 help thread for ~10k worldwide users.
>>62488064
>KCP
Abandonware that's hopeless outdated.
>CCCP, and K-Lite
Both shit.
>>62487855
But it's not
>>62488064
last time i used k-lite codec pack it was like 2004 and i still downloaded stuff off kazaa lmao
>>62488099
Then what is?
>>62488023
But I was talking about the 1080p screenshot, not the 720p screenshot
>>62488131
It's an upscale, with the power of studio upscales... and madVR, apparently
>all of this autism
The other guy was just shitposting that madVR did the magic upscaling. It's not that complicated.
>>62488020
It's not for you, fuck off
>>62488153
Which it did, since it's the reason the 720 screenshot is in 1080.
Hi /madvr/! In your opinion, what are best settings for madvr on my configuration?
I've an Intel i5-4590; an MSI R9 270X 2GB; 8GB RAM.
Thanks!
>>62488181
Try NNEDI3 32 neurons all around.
>>62488196
I tried it but now my GPU is at 100% and my fann is spinning like crazy WTF?
>>62488246
That's because it uses your GPU.
If you bought a video card with a shitty fan, just use Jinc.
>>62488360
human eye can't see past spline36
>>62488360
mpv I just tried mpv and the fan spinning stopped
I'll stick to it I guess. Thanks anyway!
>>62488196
Enjoy your center shift
>>62488391
Enjoy your bilinear.
>>62488452
>he doesn't use scale=anime for the best quality
Kodi implemented smoothmotion/interpolation!!!!
https://github.com/xbmc/xbmc/commits/master
>>62480523
Personally, i started using it because it could actually play some 6 fps videos I downloaded while VLC kept shitting the bed whenever I tried. But that's just me.
>>62489081
>6 fps videos
wut
>>62489097
framerate
>>62489227
I know that. I just didn't know 6 fps video even existed.
>>62489245
You can make it any frame rate you want. I slowed down a 30fps video to 1/5th speed and Youtube handles it just fine.
>>62489306
>>62488196
Here's nnedi3 64
I'm unsure why anyone would ever want to use this
>>62489245
The videos (pic related) are very roughly animated, like barely at all, so it didn't really affect the way it looked much. Oddly enough, not all of them seemed to be at such a low framerate with others at like 30.
>>62489350
SSSR
>>62489359
What's up with the random stretching?
>>62489350
>>62489359
2nd one looks much better
>>62489401
>I like everything sharpened
That's what sharpeners are for.
>>62489385
I lost the frame because I had multiple mpvs open (oops)
That offset is caused by nnedi3 for whatever reason, it's not present without shaders or SSSR
>>62489424
Considering nnedi3 is supposed to be "very good" at enlarging images 2x, it does a pretty piss-poor job compared to SSSR and even RAVU
>>62489442
Not the offset. Entire regions shift.
>>62489474
Just nnedi3's center shift
>>62489541
That's not a center shift, moron.
So are you guys actually watching videos or just comparing screenshots the whole day?
>>62489573
Bit of both
>>62489573
thats what multimonitors are for.
>>62489573
Would you like to see the 300 Sailor Moon screenshots I recently took with mpv?
>>62489599
>mpv
No thanks.
>>62489306
mpv --speed 0.2 :^)
>>62489464
>"very good"
nobody believes this except braindead madVR fanboys
>>62489686
I'm just going off the documentation, as I said in my post I have no idea why anyone would actually use it
Almost twice as demanding as SSSR too...
>>62489541
That's not the center shift, the entire image is warped - like a heat-distortion type effect. But as anon pointed out, he missed the frame. If the original animation had some sort of heat effect then it could be that. Even meme neural networks aren't going to cause that significant of a shift at only 8x4 window size. It must be a difference in the frame.
Also, the center shift of NNEDI3 is very tiny - only a half pixel. mpv also automatically corrects it if you use any scaler except bilinear. (e.g. ewa_lanczos)
>>62489726
You're correct, according to mpv I was a few ms off
00_04_57.630 vs 00_04_57.963
That being said the 'distortion' or whatever you want to call it is also present in other frames
I have a good comparison but it spoils the latest episode of Made in Abyss so I'll try and find something similar in last weeks ep
>>62489726
>>62489776
It's not as significant though, as you pointed out *
What is haasn's favorite anime?
Here's a good example
Check out the line going through his chin
>>62489866
Seems to be a 2-3px deviation
>>62489799
prolly katanagatari
>>62489866
uh what the fuck, something is seriously messed up with the coordinate calculation or something
That doesn't seem normal at all. try on https://0x0.st/7GU.ppm
>>62489910
NNEDI3
>>62489953
is this the power of neural networks?
>>62490014
Pretty interesting that you can literally see the offset in the thumbnail
>>62489953
Can you try the old NNEDI3 versions (before the 4pass optimization)? I suspect there might be some kind of regression
>>62490042
If you can point me to where I can find them I'd be more than happy to, I'm just unsure where I can find the old versions
Is there anyway to use the start option with the resume playback option? I have playlists that I use and I want to save the time for the video it's playing and resume correctly but skip the intro when it loads the next file
>>62490078
mpv-prescalers, just checkout some older version (`git log`)
>>62490132
nnedi3-nns64-win8x4.hook from https://github.com/bjin/mpv-prescalers/tree/cd9742657732dbd8f8f27dee233f41190a7d9be1
>>62490168
>>62490168
Here's a comparison to the other regs
Sorry it's not exactly the same position but you can tell it's noticeably better than the current NNEDI3
>>62476602
Where are the configs for mplayer?
guys, sssr is a post processing algorithm which first requires you to upscale the image with another algorithm. so comparing ewa-lanczos + sssr to just nnedi3 is not very fair. that's like comparing "ewa-lanczos + debanding" to just nnedi3.
if you want a fair comparison, compare "ewa-lanczos + sssr" to "nnedi3 + sssr". or can't mpv do "nnedi3 + sssr"? technically, nnedi3 + sssr should be possible.
and btw, you also need to not cherry pick the samples. nnedi3 has the very specific purpose to upscale highly aliased sources, and it's much better at that than ewa-lanczos (with or without sssr). but yes, if you upscale decent quality sources which aren't heavily aliased but rather soft, then nnedi3 can't show it strengths and ewa-lanczos + sssr looks sharper/better. nnedi3 also shows much less bloating than ewa-lanczos + sssr, but once again, you won't notice that if you only test with very blurry non-aliased sources.
>>62490677
I don't believe NNEDi3 + SSSR works
My screenshots were all ewa_lanczos + SSSR/NNEDi3
I didn't cherry pick my results, I saw the comment by the other anon, opened the first file I saw and took some screenshots, regardless the above posts point to there being something wrong with the current iteration of NNEDi3 (unless I've stuffed up the comparisons)
maybe mpv doesn't support nnedi3+sssr, but in madvr nnedi3+superres is possible, so there's no technical reason why it wouldn't work. nnedi3 has a pixel shift which would need to be fixed for sssr to work, but that's not really hard to solve.
sssr produces nicely sharp results, but it isn't nowhere near the anti-alias capability of either nnedi3 or ravu. also it produces more bloating than either ravu or nnedi3.
if you look back through the last couple of mpv threads, iirc there were some image comparisons where ravu ran all over ewa-lanczos+sssr, same should apply to nnedi3. if you can't find them, i can look for some image where lanczos+sssr is much worse than ravu or nnedi3.
This retard already posted some distorted nnedi3 images in few previous threads. He doesn't even fucking know how to take a screenshot or how to compare with the same frame. I tested one of his pictures he provided as source and the nnedi3 (latest version) result looked normal as expected, compared to his mess. Don't take his (((comparisons))) seriously. He doesn't even understand that after doubling with shaders his pic gets downscaled to match his monitor resolution. Also probably the same person who couldn't figure out how to usefor two days even though people spoonfed him.
>>62490861
This is literally the first time I've ever posted about NNEDi3
Keep attempting to derail though
>>62484201
>>62484185
what's with the hooves?
>>62490956
>>62490168
Yeah that looks much more like it
>>62491018
See >>62490270
It definitely looks much better
>>62490677
>or can't mpv do "nnedi3 + sssr"? technically, nnedi3 + sssr should be possible.
mpv can, the NNEDI3/SSSR shaders can't
>>62490861
uh anon I was able to reproduce his results. it's the NNEDI3 shader, has a bug since the 4-pass optimization by bjin
>>62490881
You even have the same file name styling, pretty pathetic of you to pretend to be another person.
>>62491033
>uh anon I was able to reproduce his results. it's the NNEDI3 shader, has a bug since the 4-pass optimization by bjin
Works fine for me.
>>62491049
>Getting his nickers in a twist over a shader
And you call me pathetic? About the file names, I don't know what you're expecting, if I leave them alone 4chan-x scrambles them
Here is (((nnedi3))) he posted in previous thread.
Here what it looks like with normal nnedi3 with normal mpv setup.
>>62491146
Pastebin your NNEDi3 code
>>62491167
Latest mpv-prescalers-master branch. nnedi3 from the gather folder.
>>62491177
I'm still getting the strange lines with that specific version
>>62491191
Post pictures.
>>62491049
>Works fine for me.
>>62491209
Post source, haasn.
>>62491202
I don't know what you're expecting, the offset is still there
>>62491226
Nice spoliers faggot.
>>62491220
That file was posted earlier, have you even read the thread?
>>62491231
Good point
>>62491243
You mean >>62489910 ? I missed it. Can you upload it here, i don't wanna click on fishy links.
>>62491266P1
9 9
1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1
1 1 1 1 0 1 1 1 1
1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1
Anyway nnedi3 for me looks normal even compared to RAVU or just regular scalers. Tested on multiple pictures.
>>62491209
Also it looks like this if I also add --scale=ewa_lanczos
That said, I think what's happening here is that the errors from NNEDI3 are getting compounded due to sigmoidization and the extreme scaling factor. It's worth pointing out that I also tried the older version and it looks exactly the same. Are you sure when you tested the older NNEDI3 that it was actually being applied?
GUYS, GUYS! WTF i just used nnedi shader on picture and look how WEIRD and DISTORTED it looks! :D HAHA SuperRes works OK! ;) BTW how do i tkae screenshot? I print screen and crop in paint right nwo. BTW what is stats.luma? Yes i didnt check but i can FEEL nnedi is applied!
@ EVERYONE
HOw to make words in a white box? Pls explain how to post text inside a white box?
>>62491136
Something is fucked in his setup. Looks like bilinear upscaling.
>>62491452
Please do not stretch my wife haasn, that's the job of my penis.
New >>62491603
nnnedi guy: try this test image instead
--vf=format=gray
>>62491627
my result
>>62491635
>the fucking edges
10/10 scaler
>>62491627
Looks the same as bjin's nnedi3 from December 2016. Better than madVR BTW.
>>62491673
I mean you have to remember the image basically looks like this
>>62491673
Now compare to ewa_lanczos, dumb fuck.
>>62491775
ewa btfo
>>62491775
>>62491753
haasnsoft :^)
>>62491803
at least unlike nnedi it doesn't leave a darker coloured band at the right edge