> windows builds:
- https://kawaii.imo.uto.moe/
- https://sourceforge.net/projects/mpv-player-windows/files/
- https://bitbucket.org/rorgoroth/mpv-for-windows/downloads
> user shaders:
- https://gist.github.com/igv
- https://gist.github.com/voltmtr
> low preset without interpolation:vo=opengl:deband:blend-subtitles
hwdec=auto-copy
> high preset without interpolation:vo=opengl-hq:blend-subtitles
hwdec=no
> low preset with interpolation:vo=opengl:interpolation:blend-subtitles
video-sync=display-resample
hwdec=auto-copy
> high preset with interpolation:vo=opengl-hq:interpolation:blend-subtitles
video-sync=display-resample
hwdec=no
> video samples to test interpolation:
- https://raw.githubusercontent.com/haasn/cms/bac473311eda59e8730a48e858c2f0d867023b40/panning2.webm
- https://raw.githubusercontent.com/haasn/cms/bac473311eda59e8730a48e858c2f0d867023b40/panning3.mp4
Watch with interpolation on and off and notice the judder without it.
> http://forum.doom9.org/showthread.php?p=1768307#post1768307
I finally had some success combining the theory behind the SSIM downscaler and SuperRes. Which gave rise to a new experimental algorithm: Structural Simmilarity (SSIM) SuperRes, or SSSR for short. It's still a bit of a work in progress, and I don't want to give too much away yet but suffice it to say it'ssharp.
https://github.com/zachsaw/MPDN_Extensions/commit/43bfe6c16c134c0effcf0c350d1e52f0aa711fc5
Did anyone here try the new SSSR algorithm?
Is it really that sharp?
Asked in a previous thread but no answer: how can I easily check if hardware decoding is used for a file?
I open files by double click so no terminal output.
Is youtube-dl not working for anyone else? Using the latest build on windows, and when I doyoutube-dl --list-formats <link>nothing happens. It just hangs there forever.
It only started happening about 5 days ago.
>>54664012
it's working for me
>>54663926i show-text "Hardware decoding: ${?hwdec-current:${hwdec-current}}${!hwdec-current:n/a}"
Only work with 0.17.1+ (implemented ~two weeks ago).
I think I could also port SSIMDownscaler, but it requires result of the first pass of the internal downscaler.
>>54663647
Wait, I'm confused. Are you porting this to be compatible with mpv, or?
>>54663585
>> video samples to test interpolation:
>- https://raw.githubusercontent.com/haasn/cms/bac473311eda59e8730a48e858c2f0d867023b40/panning2.webm
>- https://raw.githubusercontent.com/haasn/cms/bac473311eda59e8730a48e858c2f0d867023b40/panning3.mp4
I am not seeing anything.
>windows builds
And why would I want this over MPC?
>>54665464
damn you are blind son...
>>54665499
than I am fucking blind kid, because there is nothing to see on those samples.
>>54663926
>Asked in a previous thread but no answer: how can I easily check if hardware decoding is used for a file?
>I open files by double click so no terminal output.
https://github.com/Argon-/mpv-stats
It displays it in the video line when hwdec is being used.
>>54665507
https://red.cachefly.net/learn/panning-24fps-180.mp4
https://red.cachefly.net/learn/panning-60fps-180.mp4
No difference at all for you?
>>54665609
Now there is a difference.
2nd one is 60fps
>>54664627
Is this not possible with current user-shaders API?
>>54665629
Now try the first clip withvo=opengl-hq:interpolation
video-sync=display-resample
>>54666663
>vo=opengl-hq
I use vo=opengl
without the -hq
exactly what I am missing
>>54663585
where can I get a scritpt that test multiple different setttings and take a screenshoot to compare wich works better for my specs?
I'm on an old laptop and this is kind of confusing unless you know about video.
are there setting options that cancel others? any link with "do and don'ts" ?
This script idea is useful and probably easy
>>54666788
sorry the typos, is not to troll purposes
>>54666734
>exactly what I am missing
What platform are you on?
There's no guarantee it will work with Windows QUALITY drivers. There's also no guarantee it will work on AMD QUALITY drivers. There's also no guarantee it will work with QUALITY compositors.
>>54665609
>>54666663
Not the other guy, but I did couple of tests.
>vo=opengl
judders happen, at a slower rate
>vo=opengl:backend=dxinterop
judders happen, but bit smaller scale, also smootherr rendering/video for some reason (shit opengl?)
>vo=opengl:interpolation
judders happen but smaller. slow rendering
>vo=opengl:backend=dxinterop:interpolation
judders happen but smaller, smoother rendering
All using hwdec=no and display-resample.
Will there ever be a proper fix? This feels like hack more than proper fix, as the difference between 60/24 still persists. The problem still there with at a reduced effect.
>>54666932
One of these days I'm going to get around to making a 60fps comparison clip between interpolation on and off, so there's no ambiguity about whether it's working or not.
>judders happen, at a slower rate
This sounds like you don't even know what's supposed to be going on, though.
>>54666932
Just use this:
vo=opengl-hq:backend=dxinterop
hwdec=auto-copy
And use the rorgoroth's latest builds.
>>54666978
I'll see if I can make a webm of what I'm seeing, I doubt it can be verified on anyone else.
Wish I had a HQ camera to record it.
Oh shit! Latest rorgoroth build fixed all my problems with my oldish AMD card except I still can't use dxinterop.
SuperRes and AdaptiveSharpen work great together.
Yeah mpv won the battle.
/mpvfags/
Anyone having problems with extreme load times? Whenever i try to load something from an online source it takes 3+ minutes to load the video. I have 50mbps internet and file size doesnt matter. I've got the newest ytdl and mpv.
>>54669131
Check your cache settings or try playing around with them
>>54669131
YouTube's counter-attack.
## Video ##
vo=opengl:interpolation:blend-subtitles
video-sync=display-resample
hwdec=auto
## Audio ##
audio-channels=2.0
softvol-max=600
## Tracks ##
alang=jpn,jap,Japanese,en,eng,English,dk,den,Danish,Dansk
slang=en,eng,English,dk,den,Danish,Dansk
## Subs ##
sub-codepage = utf8:cp1250
sub-scale-with-window = no
sub-text-font-size = 60
sub-text-color = 1/1/0
## Screenshots ##
screenshot-format=png
screenshot-png-filter=0
screenshot-png-compression=4
screenshot-template=D:\Pictures\mpv\mpv_ss_%tY.%tm.%td_%tH.%tM.%tS
## Other ##
msg-color=yes
geometry=50%:50%
autofit-larger=90%x90%
ytdl-format=best
cache-default = 8192
stop-screensaver = "yes"
[extension.webm]
loop-file=inf
>>54669143
I still get it with 8mb cache seems weird. It just hangs like this.
>>54669204
No matter the service
>>54669232
Try other windows build?
Damn why kodi has no deband or ewa_lanczos scalers.
>>54666734
>exactly what I am missing
--vo=opengl:scale=spline36:cscale=spline36:dscale=mitchell:dither-depth=auto:correct-downscaling:sigmoid-upscaling:deband:es=no
it's a shortcut for this. You also need a fast computer to take advantage of this through.
I use this madVR config:
Chroma: super-xbr100 + AR
Image: Jinc3 + AR
Double Luma: 1.5x or greater - super-xbr100
Double Chroma: 1.5x or greater - super-xbr100
Downscaling: SSIM 1D 100% + AR + LL
Upscaling refinement: SuperRes (1)
Artifact removal - Debanding: Medium/High
Artifact removal - Deringing: Off
Image enhancements: Off
Dithering: Ordered
How to do this with mpv?
>>54671609
almost any iGPU can handle it
>>54671624
there is no chroma doubler in mpv, and still no SSIM Downscaler.
try thisvo=opengl-hq:scale=ewa_lanczos:user-shaders="~~/CrossBilateralSR.glsl,~~/SuperRes1.glsl":cscale=bilinear
you will need to download those shaders
https://github.com/mpv-player/mpv/wiki/User-Scripts#pixel-shaders
mpv is shit
MPC-BE is where it's at
>>54672440
retarded joke
>>54672498
still true tho
I don't see any reason to use mpv, when mpc-be does so much more
but people have different uses I guess
>>54672527
>when mpc-be does so much more
haha
>>54669131
>>54669232
Have that same problem
You aren't the only one
>>54671842
It's just awesome.
I saw that dxinterop has better vsync, but what does it mean exactly?
>>54672440
The mpc-be development is almost dead and was focused on windows only.
>>54671624
>super-xBR
>SuperRes
>SSIM
Please kill yourself
>>54672796
Why? It's fucking good with 480p and 720p videos.
>>54672796
said some retard, who doesn't even know what are those words mean.
>>54672838
>superXBR
made for upscaling pixel art
>SuperRes
family of filters all characterized by their distinctive warping of image features
>SSIM
for downscaling (useless)
>>54672836
>Why? It's fucking good with 480p and 720p videos.
>>54672769
>fixes some bad behavior since OpenGL support on Windows is so bad.
better vsync because directx is better optimized.
>>54672901
>literallyphotoshopoilpaintfilter.jpg
>>54672859
only 1st is true
>for downscaling (useless)
you forgot IMHO here
Why not creating a kind of installer that include all these glsl files and preconfigured presets like the ranpha's installer?
https://imouto.my/tutorials/watching-h264-videos-using-compute-unified-device-architecture-cuda/
>>54672960
>oil painting filters
wew
superres on right
>>54663585
does interpolation not work on windows or something? not seeing a difference with it turned on.
>>54664012
Try another website and check if there is a new version out
>>54673129
nice, but how did you get same pic dimensions, if SuperRes works only when image has been upscaled? What downscaler if it was downscaled?
>>54673129
filter makes babby sad (distorts mouth features)
>>54673182
>nice, but how did you get same pic dimensions, if SuperRes works only when image has been upscaled?convert -filter Lanczos -distort resizeon the left
>>54672985
>only 1st is true
show me a superres algorithm that doesn't warp features
>you forgot IMHO here
Downscaling filters are useless because downscaling is a solved problem. A nice linear light cubic spline or windowed sinc (appropriately widened) is more than you'll ever need. Downscaling is just throwing information (high frequency content) away. There's basically nothing you can screw up.
Upscaling is a hard problem because to do it properly you would have to guess information that isn't present in the source.
Compare mpv's defaultvo=opengl-hqconfig against ANY of your meme downscalers if you want proof. There's no difference except when the latter ends up distorting the image instead of just downscaling perfectly.
>>54673211
babby go sad
>>54673262
>Compare mpv's default vo=opengl-hq config against ANY of your meme downscalers if you want proof.
Done, I won
>>54673359
>adding shittons of ringing is a good thing
If you like ringing that much, trydscale=sinc:dscale-radius=16
>>54673359
Imo, the left one looks better, the right one is overly sharp, and even has ringing on the "unera" text. I'd like to see an example of faces, I imaging the SSIM one will have hideous sharpened pores.
>>54673359
Oh and the ringing on the windmill square thingies, holy shit lol
/g/, I'm having some issues with my mpv. I use some sub preferences to set color, but they didn't work anymore. It was working fine but some update may have caused it to break. The video plays fine, but the sub is white instead of yellow.
Error:Error parsing option sub-text-color (option requires parameter)
/home/anon/.mpv/config:3: setting option sub-text-color='' failed.
Error parsing option sub-text-border-color (option requires parameter)
/home/anon/.mpv/config:4: setting option sub-text-border-color='' failed.
Error parsing option sub-text-shadow-color (option requires parameter)
/home/anon/.mpv/config:5: setting option sub-text-shadow-color='' failed.
My config:# Write your default config options here!
#keep-open=yes
sub-text-color=#e6e6e6
sub-text-border-color=#000000
sub-text-shadow-color=#000000
sub-text-border-size=3.5
sub-text-font-size=65
no-osc
Something I may be missing?
>>54673359
>>54673374
It's not even sharper or anything, nor does it have more information. It's just literally the same with shittons of added ringing.
You could get the exact same result out of taking the left image and passing it through a strong ringing filter
Unbelievable, simply ridiculous
>>54673412
Try setting the color codes in "quotes".
>>54673403
>>54673414
easy fix
>>54673374
Hey cscale=sinc:dscale-radius=16 is a good way to sharpen!
>>54673434
How?
>>54673434
>>54673467
slightly less ringing is still ringing
>>54673424
Thank You, anon! It worked perfectly. I feel dumb for not thinking that before.
>>54673477
You meant slightly more than mitchell? At least without blur.
>>54673514
the sharpness is literally the exact same >>54673477
What's this ringing meme? I don't see any rings.
>>54673541
literally not even close
but whatever, I don't care about SSIM Downscaler
>>54673359
wow so sharp
wow such ring
quality
wew
>>54673552
>I don't know what a word means, so it must be a meme
>>54673552
>>54673580
Ringing? I don't see no rings. What is this ringing you speak of?
>>54673580
>>54673623
nice photoshop skills
>>54673623
Damn, that's sharp, please share your madvr settings, please do this in a 400 line tutorial explaining what tab and button to click.
>>54673636
Not photoshop. See >>54673444
>>54673636
>>54673693
I also added an extra sharpening filter or 10, because madVR memesters love sharpening filters, right?
>>54673693
>>54673706
looks crappy
>>54673569
>literally not even close
All the pixels are the exact fucking same, the bottom two just have boosted contrast and extra ringing
Look at the shape of the letters. The pixels match 1:1. There is no extra “sharpness” at all.
>>54673721
you blind or something?
http://i.4cdn.org/g/1463865708460.png
2nd pic is the same only with less ringing
>>54673750
*with more ringing
>>54673750
Yeah, less than the other SSIM downscaled one, but it's /still/ exactly the same as a mitchell downscaled image (as in the exact same pixel data, not more "detail", just more artificial contrast (which results in ringing obviously, as seen in >>54673477)).
>>54673796
sure, no body said it has more details.
>>54673721
You don't understand at all, the contrast and ringing placed algorithmically to boost the perceived cognitive quality of the image.
>>54673129
how you did this
pretty nice
>>54673865
Open Photoshop(r), -> Filters -> Oil Paint
What the fuck do all these image comparisons mean?
Does it mean we can't have better than ewa_lanczos?
>>54673966
yes
>>54673966
and mitchell
>>54673988
>>54673975
Even in 2020? So why ewa_lanczos is not the default for opengl-hq?
>>54673848
>the contrast and ringing placed algorithmically to boost the perceived cognitive quality
Yep, that's the same kind of thinking that gives us loudness war and the stupidly oversaturated / torch mode displays you see in shops.
It completely destroys fidelity, distorts the source, ruins the quality and makes it seem like cheap trash
but idiots like it, I guess
>>54674010
>So why ewa_lanczos is not the default for opengl-hq?
It's too slow for most normie's shit PCs
>>54674238
>but idiots like it, I guess
Yes they buy $600 GPU just for that:
http://www.projection-homecinema.fr/forum/topic/1532-tuto-madvr-configuration-x64bitpasserelle-integration-htpc-avec-tablettesmartphone/?page=1
And some of them buy $600 CPU (Avisynth+ ^^') for even more:
http://www.projection-homecinema.fr/forum/topic/324-htpc-optimisation-avisynthffdshowlavmadvrmpc/
http://www.projection-homecinema.fr/forum/topic/1946-htpc-optimisation-avisynth-x64-madvr/
>>54674238
Hopefully VR will save us...
>>54674335
so exactly why are you guys hating on these kids?
>>54674271
so what settings should i use with eaw_lanczos if I own a 970, well if I can even use it
>>54674381
Not kids but blind grandparents!
Look at these wonderful ^^' pictures:
http://www.projection-homecinema.fr/forum/topic/324-htpc-optimisation-avisynthffdshowlavmadvrmpc/?do=findComment&comment=4448
>>54674416
Nothing, ewa_lanczos is already perfect!
>>54674428
it doesn't look bad? I don't get it
why are you hating on them
>>54674441
Seriously you like it? If so, I can't do anything for you... :(
>>54674491
They made a tutorial to have this kind of result.
>>54674440
except that it makes dark brighter sometimes (probobly because of noise?) and rings.
>>54674491
a fucking troll.. what did I expected from this place.
>>54674521
OK almost perfect...
>>54674491
You're a fucking faggot, dude.
>>54674549
Fortunately SuperRes can fix those.
>>54674541
Not a troll just real things.
>>54674562
No fags here, just perfectionists.
>>54671842
>there is no chroma doubler in mpv, and still no SSIM Downscaler.
>try this
>vo=opengl-hq:scale=ewa_lanczos:user-shaders="~~/CrossBilateralSR.glsl,~~/SuperRes1.glsl":cscale=bilinear
>you will need to download those shaders
>https://github.com/mpv-player/mpv/wiki/User-Scripts#pixel-shaders
results/comparation for this
I turned an old C2D machine into an HTPC.
What am I missing from my mpv.conf?vo=opengl
save-position-on-quit
fullscreen=yes
>>54674851
>-hq
>>54674491
Jesus fucking christ that's hideous
>>54674870
Not supported, it keeps defaulting back to a legacy opengl context
MadVR's nnedi3 for chroma upscaling is very good.
>>54675084
NNEDI3 is an overhyped, ridiculously slow piece of shit that's basically just like a less strong super-xBR and also chroma scaling is virtually irrelevant on real world content.
good subtitles settings/fonts you guys use?
also difference between
vo=opengl-hq and vo=opengl
>>54675337
Former is better quality at the cost of more GPU usage
Latter is low GPU usage at the cost of quality
>>54675337
>vo=opengl-hq and vo=opengl
manual
Anyone willing to do a comparation between mpv interpolation and svp?
>>54665609
I'll do it if you tell me what software you used to record?
>>54675504
MPV interpolation doesn't do what SVP does. They're completely different things that solve different problems
>>54675504
>I'll do it if you tell me what software you used to record?
I didn't record these, they are existing test clips.
I don't know of a good way to record mpv's interpolation output in realtime, sorry.
I would try approaching the problem by coaxing mpv into making a frame-by-frame perfect reproduction of the interpolated output.
Rather than relying on timing data, use mixing coefficients that assume a perfect 60 Hz display.
>>54675545
Or just dump the frames as raw images and write a simple C program to generate the interpolated output.
>>54675540
>They're completely different things that solve different problems
what does mpv solves?
>>54675707
https://github.com/mpv-player/mpv/wiki/Interpolation
Can someone give me a gpu raping line.I dont time to try 50 configs and read 1000 posts of aspbergs fighting
>>54676232
>Can someone give me a gpu raping line.apt-get install nvidia-drivers-196.75
>>54676232
lanczossharp with nnedi3 64 neurons
>>54676232
>Can someone give me a gpu raping line.
Best config.vo=opengl-hq:scale=ewa_lanczossharp:cscale=ewa_lanczossoft:user-shaders="~~/shaders/finesharp.glsl,~~/shaders/crossbilateralsr.glsl":dscale=mitchell:tscale=bcspline:interpolation:tscale-param1=0:tscale-param2=0.6283185:tscale-radius=2:interpolation-threshold=0.01:scale-radius=3:temporal-dither:dither-depth=8:deband:correct-downscaling:prescale-passes=2:prescale-downscaling-threshold=1.6:blend-subtitles=yes
video-sync=display-resample
hwdec=no
framedrop=vo
>>54676438
>finesharp.glsl
what's this?
Does tscale=oversample do anything at all for video smoothness or is it essentially equal to having interpolation off?
>>54676601
yes, it does
I prefer it over mitchell
>>54674851
vo=opengl:scale=spline36:tscale=triangle:interpolation
Try that
Hey guys, I recently got a t420 to use as my htpc. It has a i5-2520M, and 6 gigs of ram.
I copied over my main laptops mpv config, and had performance issues.
The performance part of my config isvo=opengl-hq:scale=ewa_lanczossharp:cscale=ewa_lanczossoft:tscale=mitchell:temporal-dither:blend-subtitles:pbo:target-prim=bt.709:target-trc=bt.1886:gamma-auto:icc-profile-auto:3dlut-size=256x256x256
I couldn't even keep in opengl-hq.
Can my processor really not handle that, or is something else bottlenecking it. (HDD maybe?)
Is the difference between sandybridge and haswell that big?
Running ubuntu 16.04 lts.
Any config reccomendations?
Removing the entire performance section and going off defaults gets me smooth performance.
I don't understand the whole icc-profile thing, can someone explain?
What different does it have over not having it set?
How do I downmix 5.1 surround to stereo?
>MPV interpolation
When will /g/ jump on the buttery smooth Vapoursynth train?
>>54677593
>I couldn't even keep in opengl-hq.
>Any config reccomendations?
opemgl-hq should run just fine, on my T420 I use this on a 1600x1200 monitoropengl-hq:scaler-resizes-only:dither-depth=8:temporal-dither:icc-profile-auto:3dlut-size=128x128x128:blend-subtitles=video
Sometime it has problems with 1080@30p, I need to disable debanding on those clips
I can do 4K@24p with thisopengl:scaler-resizes-only:dither-depth=8:temporal-dither:target-prim=bt.709:target-trc=bt.1886:blend-subtitles=video
Oh, both of this is with hwdec=no, try it that way, there's a performance hit with hwdec
All of this is on Win 7 though, but performance should be better on *nix with this oldish Intel GPU's
>Is the difference between sandybridge and haswell that big?
There's quite a difference in the GPU department, next time get the Thinkpad with a dGPU instead of relying on Intel's bullshit
When will H.265 actually be superior to H.264?
>>54677816
My other laptop is intel gpu also, although the gpus improved a lot more than the cpus, maybe that's it.
>>54677829
When decoders and encoders are better.
>>54675016
That's because you probably don't have the opengl 3.x driver.
You can still input this on pretty much any shitop that has functional opengl 2.x driver.
>>54671609
>>54677816
Thanks for the config, seems to be running well for me.
>>54663585
Is it worth to download/build this player if I'm just going to use default settings? I'm no expert, but I can notice the bad quality of things like VLC and usually install MPC-HC in Windows with other stuff in default settings.
>>54678540
just usevo=opengl-hqit will already have much better quality than with those 2.
What do I need to put in the config file to bind the scroll wheel to control volume?
I set up MPV on Win10 and it works fine except whenever I try to add any shader to my video settings it fails. It just dumps the full fragment shader source to the command line (in red text) and ends with "fragment shader compile log (status=0):" followed by a bunch of error C1013 lines. I've tried it with a couple of different shaders and they all get the same result. For reference the video setting line I'm trying at the moment with the adaptive-sharpen shader is -
>vo=opengl-hq:post-shaders="~~/shaders/adaptive-sharpen.glsl"
Anyone have any ideas what's wrong and/or how to get it working with the glsl shaders?
>>54664012
fyi you can use -F, eg. youtube-dl -F url
>>54672527
It literally doesn't. Please remember that madvr is not part of mpc.
>>54664012
ensure you put the link in quotes
>>54679067
Pls respond, this seems like an easy solution, but I just can't find the command anywhere.
Also, is there any way to get Popcorn Time to recognize gnome-mpv, or to stream Popcorn Time to gnome-mpv some other way?
>>54679399MOUSE_BTN3 add volume 2
MOUSE_BTN4 add volume -2
saved it as input.conf
>>54679244
what are those errors?
>>54679608
When I run a video the output looks like this:
>AO: [wasapi] 48000Hz stereo 2ch float
>VO: [opengl-hq] 640x480 yuv420p
>[vo/opengl-hq] fragment shader source:
Then the whole source code for the shader is shown with line numbers and after that it reads:
>[vo/opengl-hq] fragment shader compile log (status=0):
>[vo/opengl-hq] 0(109) : error C1013: function "hook" is already defined at 0(56)
>[vo/opengl-hq]
>[vo/opengl-hq] shader link log (status=0): Fragment info
>[vo/opengl-hq] -------------
>[vo/opengl-hq] 0(109) : error C1013: function "hook" is already defined at 0(56)
>[vo/opengl-hq]
>[vo/opengl-hq] after rendering: OpenGL error INVALID_OPERATION.
>AV: 00:00:00 / 01:31:19 (0%) A-V: -0.033 DS: 2.000/0 Cache: 10s+256MB
>[vo/opengl-hq] after rendering: OpenGL error INVALID_OPERATION.
>AV: 00:00:00 / 01:31:19 (0%) A-V: -0.047 DS: 2.000/0 Cache: 9s+256MB
>[vo/opengl-hq] after rendering: OpenGL error INVALID_OPERATION.
>>54679826
>vo=opengl-hq:user-shaders="~~/shaders/adaptive-sharpen.glsl"
with latest build from
>https://bitbucket.org/rorgoroth/mpv-for-windows/downloads
>>54679842
Just tried using the mpv-latest-Win64.zip file and I got the same results. Actually it was a little different as with this version the video appeared as a blue screen (audio seemed to be working) but the errors came out the same.
>>54679895
>user-shaders
>>54679915
Well shit. I could've sworn I'd tried that before and it didn't work but damn if I was wrong. That fixed it. Thanks.
>>54677018
Useless because interpolation needs video-sync=display-* option. :(
Is there a fix yet for interpolation being real shit since the last update?
Damn linear scaling is shit.
Why finesharp and adaptive are better than just changing cscale-radius, scale-blur and scale-radius to increase sharpen?
I am on windows and I still don't know why I should use an OpenGL based software? Why shouldn't I use a D3D based video player like MPDN?
When a deringing option in mpv?
>>54676438
>correct-downscaling
why
>prescaling-passes=2
why?
>dither-depth=8
why??
>>54677735
>Trumpposters don't understand the distinction between VapourSynth (the scripting framework) and MVTools (the interpolation filter).
When will this meme die?
>>54681436
There'sscale-antiring
>>54681436
>unironically using destructive video 'enhancements'
Just stay on plebVR, pleb.
>>54681469
Not the same thing, deringing remove ringing from image source.
>>54681473
Banding and ringing are real.
>>54681500
If the ringing is in the source to begin with you're fucked. Only way to “remove” ringing is by heavily distorting it then.
>>54681514
Just like deband...
>>54681500
>Banding and ringing are real.
And they can't be fixed without objectively damaging the quality of the video. The only video quality enhancements you should use are non-destructive ones like superior upscaling algorithms.
Also what this guy >>54681514 said, if your files have ringing in the first place you downloaded some absolutely shit-tier encodes.
>https://gist.github.com/igv
Why FineSharp.glsl or SuperRes1.glsl but adaptive-sharpen.glsl?
>>54681539
What does finesharp and superes do anyway?
>>54681522
On the contrary, actually. Banding is relatively easy to move with non-invasive filters, since recognizing gradients in otherwise featureless regions of the image is easy.
Recognizing and counteracting ringing, on the other hand, is hard and will mostly either introduce aliasing, or make everything look artificial.
>>54681530
Deringing should not be always used, only if there are ringing artifacts on source.
>>54681115
bump
>>54681585
It shouldn't be used even when there's ringing in the source. You're just deteriorating the image quality even further by distorting the ringing away.
How can I force mpv to always save position on quit? Some lua script?
What subtitle fonts do you bags of rancid human trash use? Getting sick of serif-sans here.
What is between spline36 and ewa_lanczossharp? Is it robidouxsharp?
My laptop doesn't handle ewa filters unfortunately.
>>54681660save-position-on-quit
>>54681718
robidouxsharp *is* cheaper to compute than ewa_lanczossharp, but since it only uses a radius of 2, much of EWA's benefit goes away. (Needs radius 3-4 to really make a difference)
My advice would be to stick with spline36 if your GPU can't handle ewa_lanczos.
>>54681743
How to get sharper result?
>>54681768
Simplest way is to download a higher-res source image. You can't invent details that aren't there.
Otherwise, you get a tradeoff between sharpness and aliasing when compared to an ideal (frequency-invariant) reconstruction filter.
>>54681718
>>54681768
ewa_lanczossharp isn't sharper than spline36
>>54681918
Then whats the point of using it? why does it have a higher performance hit?
>>54682004
no aliasing
Is there an option to show info about the file (video and audio format specifications) on the screen somewhere? I'd like to bind it to a keyboard shortcut.
>>54682520
https://github.com/Argon-/mpv-stats/
>>54663585
why not add this link
https://github.com/mpv-player/mpv/wiki/User-Scripts
for all this noobs?
I (with contributors) have made a series of scripts that play videos from various webpages (supported by youtube-dl) in mpv. It is especially useful for nut-cases (like myself) that have JS off by default and still want to be able to play videos.
https://github.com/agiz/youtube-mpv
Criticism is more than welcome.
>>54682629
How is this better thanmpv URL
>>54682678
In a sense that I can maintain my own youtube-dl. Another benefit is that with mpv URL youtube-dl module is loaded every single time thus slightly increasing the time needed to start the video. If you rarely use mpv URL then I guess that doesn't matter.
>>54682546
Neat.
I'm having trouble using it though. I placed it as told in the ~/.config/mpv/scripts directory, but pressing "i" doesn't trigger it. Am I missing something? Do I need to enable scripts with a config option in mpv first?
>>54682774
>Do I need to enable scripts with a config option in mpv first?
no
>>54682806
>>54682774
It's because I used the config line which disables all the default keybindings.
Now I'm trying to figure out how to use the
>script-binding "<name>"
keybinding option.
I tried
>i script-binding stats.lua
but it's not working. The documentation on the mpv website isn't really helping and it's kind of vague with no examples. Does anyone know how to exactly do this?
>>54682932
>>54682546
>>54682774
Figured it out.
k script-binding stats/display_stats
>>54681438
Correct downscaling is better. Read the manual. Pre scaling passes are a remnant when I was trying out pre scalers. Specifying dither depth is required because it's off default.
>>54683153
>Correct downscaling is better.
It's enabled by default so why is it in your config. Same goes for most things
>Read the manual.
Maybe you should read the description of opengl-hq
>Specifying dither depth is required because it's off default.
This is bullshit. It's on by default for opengl-hq, and by default it uses your monitor's auto-detected dither depth, which is not necessarily always 8.
Hard-coding it as 8 is just stupid, especially when you are going to recommend your bullshit VO line to others.
Damn new SSIM SuperRes is ugly...
Am I right in that executing script does not affect playback? If the script will take some time to read and arrange properties in some output string before showing them it will not create stutter?
I think I have weird behavior with mpv.
Is it normal I can use this config with an Intel HD3000:
vo=opengl-hq:deband=no:interpolation
video-sync=display-resample
hwdec=dxva2-copy
fullscreen=yes
ytdl-format=bestvideo[height<=?1080][vcodec!=vp9]+bestaudio/best
I thought my laptop was too limited for interpolation. However I can't go forward without desync and blurish playback.
>>54684741
sure
>>54684768
And why when I go forward (right key), I have messy playback and I need to quit and go fullscreen several time to retrieve a smooth playback?
>>54684741
Add check-pattern to your vo
vo=opengl-hq:deband=no:interpolation:check-pattern=1-1
You can see miss interpolated frames and dropped frames when you run a video from the command line, if they are not very high sure it is working
>>54684790
You can use cscale=bilinear if you want to go easier on your iGPU
>>54684794
This is what I get:Playing: https://www.youtube.com/watch?v=g708PmJAbuI
(+) Video --vid=1 (*) (h264)
(+) Audio --aid=1 --alang=und (*) 'DASH audio' (aac) (external)
[dxva2] Using fallback memcpy (slow)
AO: [wasapi] 44100Hz stereo 2ch float
Using hardware decoding (dxva2-copy).
VO: [opengl-hq] 1920x1080 nv12
[vo/opengl-hq] GLX_SGI_video_sync not available, disabling.
AV: 00:00:00 / 00:02:47 (0%) A-V: 0.012 DS: 1.000/2 Dropped: 12 Cache: 8s+7KB
Audio/Video desynchronisation detected! Possible reasons include too slow
hardware, temporary CPU spikes, broken drivers, and broken files. Audio
position will not match to the video (see A-V status field).
AV: 00:00:20 / 00:02:47 (12%) A-V: 0.000 Dropped: 56 Cache: 10s+17MB
[ffmpeg] tls: Failed to send close message
[ffmpeg] tls: Failed to send close message
Exiting... (Quit)
>>54684877
>GLX_SGI_video_sync not available, disabling
Yeah, I don't think interpolation is active, your drivers don't support needed OpenGL extension.
You can try with an ANGLE enabled build
by adding backend=angle to vo
https://sourceforge.net/projects/mpv-player-windows/files/64bit/
>>54684877
>Using fallback memcpy (slow)
you better use dxva2, not dxva2-copy, or disable it.
>>54684987
No GLX_SGI error with ANGLE but video needs 20s to get smooth playback.
>>54685074
Using hwdec=dvxa2 make it worse, choppy playback (fullscreen multiple times doesn't work) and I get vivid colors.
>>54685133
>but video needs 20s to get smooth playback.
If you are getting A/V desynchronisation notices in the command line, video-sync and interpolation are getting disabled
I think this is the least demanding config for interpolation for your case, you can try different hwdec options. Try on a local, non streaming videovo=opengl:interpolation:check-pattern=1-1:backend=angle
video-sync=display-vdrop
hwdec=dvxa2
If you are getting too much dropped/missed frames or A/V desync, your iGPU is simple too weak for it
>>54685269
interpolation:tscale=nearest may be less demanding
Is interpolation noticeable for live action content, or is it mostly useful for animuh?
So when I go to double click and open a file that I have assoc. with MPV, I get an error message saying that MPV stopped working, but if I right click the same file and use Open With -> MPV it will open the file fine. Any ideas as to what could be causing this?
>>54669131
i had the same issue once, it fixed itself after a few hours. i still don't know what it was.
>>54685500
It's extremely noticeable for live action, especially pans.
In fact, it's more noticeable for live action than for anime, because anime is mostly low fps garbage either way
>>54685269
>>54685329
My iGPU is too weak for mpv. :(
That's sad because I can run madvr's smoothmotion with spline36AR chroma upscaling (cscale).
Where exactly do you enter the presets in the OP?
>>54685706
It could just be your opengl drivers being shitty(Window opengl drivers historically has been really shitty), or your playbacking a resolution that's too demanding for your igpu.
Is it possible to NOT blink while switching to fullscreen mode?
>>54684503
Yes but the displaying of some text on the OSD itself is actually quite a heavy operation that can lead to dropped frames if your computer is already at the limit with just playing the video.
>>54681741
out of the blue mpv is saving positions,i dont have this in my config.is there another switch im missing?
What fonts do you use?
pic related is Tex Gyre Adventor
>>54686351
Are you using Q instead of q to quit?
>>54686482
Nope.
Any reason not to use--osd-fractions? Like, is there a performance hit or something? No negatives mentioned in the manual, but I just wanted to be sure because some of these osd things apparently had a big impact.
>>54686656
No, it's purely aesthetic
>>54685691
Well anime very regularly has 24 fps pans,
Got some sample live action scenes which I should check out?
>>54686656
I guess the biggest negative is that you don't care about milliseconds in most cases, do you?
>>54686731
>>54686759
Okay, thanks. Wanted to know timestamps for webms and shit, but didn't see a good way to see them first.
>>54685880
Oh I thought angle and dxinterop fixed that.
Should I install a linux based os to get the best out of mpv?
>>54686795
>Should I install a linux based os to get the best out of mpv?
Yes, absolutely. mpv on windows is a hack at best, a joke at worst. Using Linux will help you get the best out of your computer either way.
>>54686759
http://www.spirton.com/uploads/InterFrame/20110618-Sample-Original.mkv
http://www.spirton.com/uploads/InterFrame/20130218-Sample-Original.mkv
>>54686795
angle translates every mpv opengl call to directx counterpart bypassing driver supplied opengl libs. it should have a negligible overhead, slim likelihood you would get a better performance under linux & native opengl drivers. It's the interpolation requiring display-sync made them too demanding for low end
>>54686957
>>Should I install a linux based os to get the best out of mpv?
>Yes, absolutely. mpv on windows is a hack at best, a joke at worst. Using Linux will help you get the best out of your computer either way.
kek
https://github.com/mpv-player/mpv/wiki/FAQ
>Nvidia should generally not tear. Sometimes, it tears in windowed mode, but not in fullscreen (solution unknown). There are additional problems on multimonitor systems.
kek
>Intel tears out of the box. Intel users on Linux are going to have a pretty bad time.
top kek
>AMD users on Linux are going to have a pretty bad time.
pic related
>>54686995
Does it mean madvr's smoothmotion doesn't use something like display-resample?
And that 's why it works without choppy playback.
>>54687049
Works on my machine
I've played around with all sorts of shaders and configs,and all i notice is they make motion blurry as hell.
I've been using vdpau for years and it looks way better then anything i've found in any of these threads.Am i missing something?
>>54687049
>Windows
>Whether tearing happens with OpenGL may depend on the backend. Try various with --vo=opengl:backend=NAME, where NAME is the name of the backend. dxinterop seems to perform best generally, but it will crash randomly with some older buggy Intel drivers.
>--vo=direct3d should never tear. (Maybe.)
>>54687071
>Does it mean madvr's smoothmotion doesn't use something like display-resample
Not that I know of, madvr docs recommended ReClock for that last I looked
>>54687049
>>54687101
Tearing and display-sync are two entirely different things. Just because it won't tear doesn't mean display-sync will work even remotely well.
>>54686086
Please respond. In madvr I would just disable exclusive fullscreen.
Definitely install Linux to use MPV.
Windows sucks. It has no graphics drivers or audio so it makes mpv look performing like shit when really Windows sucks.
>>54686086
Yes, use Linux.
>>54687145
>Install another OS in sake of anime
>>54687071
madVR does timing by polling the DirectX status on a virtual window at 1000 Hz, or something stupid like that.
(Part of the reason it has such a high CPU usage)
>>54687180
>Uses Windows
>Watches Anime
Mixing /a/ and /v/ does not give you /g/
MPV works best on Arch Linux. Remove Windows it sucks.
>>54687071
madVR's smoothmotion is the equivalent of enablinginterpolation:tscale=oversamplein mpv. It's just an algorithm
madVR's video synchronization method uses active polling. mpv's video synchronization method (with video-sync=display-*) relies on blocking swapbuffer calls (with working vsync).
ReClock is something separate altogether, but it's the equivalent to mpv'svideo-sync-max-video-chancefeature.
>>54687213
>Arch Linux
Does mpv has this?
>>54687497
Yes
