Playback related things that madVR does better than mpv:
>Proper RGB output where red means red (R:255 G:0 B:0), and green means green (R:0 G:255 B:0). No bullshit. This is currently not possible on mpv. >Custom modeline support that allows for perfect frame pacing. No resampling required (23.976Hz, 24.000Hz, 29.970Hz, 59.940Hz...). >Support for proper color management and 3DLUTs. Respects your intelligence and display calibration without raising your black levels. >Proper projector support (lens memory number, cropping masked borders, anamorphic lens support). >3D playback. >madshi's custom SmoothMotion algorithm that is smoother than mpv's blurry temporal interpolation (tscale=oversample is NOT "smoothmotion") >IVTC and deinterlacing
Rendering related tasks that madVR does better:
>*EXTENSIVE* HDR tonemapping options (mpv only has 2-3 tunable options) >Better debanding. >Can remove compression artifacts while keeping details intact. No CNNs required. >Can remove ringing artifacts that are part of the encode. >Support for optimally configured Jinc 3 (mpv is not optimally configured). >Optimally configured KrigBilateral support without pixel shifting. >Built-in 2D SSIM downscaler that is configured with Jinc (mpv shader does not support polar downscaling). >Super-xBR support, a cheap alternative to Jinc 3 for high quality chroma scaling. >Provides NGU upscaling CNN for high quality video output. Comes in various flavors: NGU Standard (better than FSRCNNX), NGU Sharp (not available on mpv), NGU AA (better than RAVU), NGU Soft (not available on mpv). >High quality sharpening filters, such as LumaSharpen, AdaptiveSharpen, and SuperRes. Comes with debloating artifact removal. >Support error diffusion dithering algorithms that are tuned by madshi himself for quality and speed. Better and sharper than mpv's gaussian blur blue noise smear pattern. >Better antiringing. >Better gamut mapping.
**UPDATE**
Better debanding has been confirmed.
mpv is a failure.
By ID do you mean the numbers in the video URL or the format ID? If you mean format ID, you can do yt-dlp -f number url. If you want to combine different audio/video formats you could do yt-dlp -f number1+number2 url.
Seems retarded to test the limits of how much you should deband on a test pattern. Where are the details that can potentially be destroyed? Might as well just smooth it completely, that's "better" in this case.
Funnily enough, changing the deband-threshold to 33 (literally ONE point) is enough to remove the banding from this test image. Is there any reason for this? Is everything from 33-64 just the same thing?
dynaudnorm is shit, so thankfully it doesnt let you do this, mpv users are so hearing impaired that they've never heard good audio normalization before
dynaudnorm is shit, so thankfully it doesnt let you do this, mpv users are so hearing impaired that they've never heard good audio normalization before
I don't think MPC has built-in debanding, the madvr one looks better but is way way more GPU intensive.
Is this a bot? There's no image on the right, it's a top/bottom image comparison.
Yeah that's what I meant. Oops.
https://github.com/mpv-player/mpv/issues/12358
I swear Traneptora posts the most braindead, borderline trolling comments.
I see you've modified the osc and provided screenshots. Did you know you can modify the osc and put it in scripts/ to use it?
He's the only mpv dev who's also an ffmpeg dev other than haasn. That's gotta count for something. In general this whole "wontfix because you can fix it yourself" attitude pervades the entire dev team, but this time it's more warranted since it's a dumb subjective cosmetic change.
3 weeks ago
Anonymous
> only mpv dev who's also an ffmpeg dev other than haasn
anyone can be an ffmpeg dev, just submit patches. And philipl regularly does work in ffmpeg related to nvdec.
>Shin Sekai Yori >Sokudo >Low effort speed mini encodes meant for 2Mbps and data cap third-worlders
Don't ruin a good show by watching a shit encode, especially a show with grain.
If you uploaded an encode with banding comparable to what's on the right to a hypothetical tracker I owned I would ban you. The defaults barely do anything. Eyes. Look.
https://github.com/mpv-player/mpv/issues/12358
I swear Traneptora posts the most braindead, borderline trolling comments.
I see you've modified the osc and provided screenshots. Did you know you can modify the osc and put it in scripts/ to use it?
Yeah I dunno, but to be honest, I really like this layout more than the default one. Seekbar fill could be a little thicker I guess, and maybe the chapter indicators could be more clear (maybe invert the colors on the chapter indicators and fill the seekbar completely?). But moving the alang/slang buttons to the top is a good idea IMO.
It should also make the osc look better when its truncated in windowed mode. But to be honest, there should be an option to make the OSC not scale with window size to go along with this.
One problem this modified seekbar would have though is that the timestamp pop-up as you move the mouse around would cover the filename/file title. Right now it just pops up on the seekbar itself.
>One problem this modified seekbar would have though is that the timestamp pop-up as you move the mouse around would cover the filename/file title
I'm pretty sure this is a problem with thumbfast.lua
https://github.com/mpv-player/mpv/issues/12358
I swear Traneptora posts the most braindead, borderline trolling comments.
I see you've modified the osc and provided screenshots. Did you know you can modify the osc and put it in scripts/ to use it?
looks interesting, but since he knows how to edit he should just do it, provide a modified version for people to test. He didn't even screenshot things properly (his screenshots are missing the fullscreen and audio buttons)
OK, placate my advanced schizophrenia for a second, I think this github user is the person who wrote the kokomins cargocult blog.
>Wow, your osc is nice and clean, could you provide a guide on how to edit it from original like that? I want to use latest osc from mpv source to make it look like yours! >To be frank it was kinda jury rigged from an old version, took days of testing to get the elements placed right. Some other 3rd party mpv builds have unique osc and I just copied code of elements I liked. Unfortunately how to edit the lua code isn’t well documented so even I can’t figure out how to properly make one from scratch.
Pic related, a file from the blog. Looks identical to this guy's OSC.
OK from digging further into the blog, it seems like kokomins blog actually provides an mpv build or something with a custom osc.lua. (I aint clicking that shit nigga)
It looks like this guy might've just downloaded it from this blog, and modified it to work with thumbfast.lua because the kokomins guy says:
>Gotcha, is it possible to implement this in your OSC? https://github.com/po5/thumbfast It would be the perfect combo in my opinion >Unfortunately this is out of my coding abilities, not do I currently have time to dedicate to hobbies anymore. What I suggest is that you could compare their osc.lua to the vanilla file and figure out how they implemented it, and if it only requires small modifications try to implement in mine.
My best guess is that he's just passing the kokomins osc.lua as his own creation. This explains his previous issue as well.
This is some deep lore. This was also the image used on the april fools mpv page right? I remember an anon asking why the osc there looks different. >Unfortunately how to edit the lua code isn’t well documented
What did he mean by this
>This was also the image used on the april fools mpv page right?
Lol I made that image. It's not an actual april fools page, it's just a quick inspect element meme. When you search "mpv" and "anime" I'm pretty sure that image from the kokomins blog is what pop up first.
>llyyr tranny
Now now, llyyr may be many things (weeb, pajeet) but I have not seen evidence that he's a tranny. And statistically I've never seen an indian person succumb to trannyism (probably because they know they'll never pass)
yes, it's also the guy that gets irrationally angry if you say ffmpeg instead of "FFmpeg", they're the same guy
https://desuarchive.org/g/thread/95849923#95858179
And the actual schizo who thinks everyone on LULZ is one person and calls everyone a schizo is now talking to himself to force his retarded narrative.
>maybe if I samefag I can convince people to believe in my delusion that one guy spelling it as ffmpeg and hurting my tranny friend is the same as two other guys who made me sperg out by spelling it as FFmpeg and making fun of me for falling for the Reddit spacing hoax, and that everyone on LULZ that makes me lose my mind is a schizo, irrationally angry and one person
Seek help.
Yeah no surprise the newfag fell for that hoax is a schizo because only a schizo newfag would think reddit invented paragraphs and typing like an idiot is how oldfags typed
[...]
https://desuarchive.org/g/thread/95849923#95858179
And the actual schizo who thinks everyone on LULZ is one person and calls everyone a schizo is now talking to himself to force his retarded narrative.
Give me the qrd of that retarded meme "reddit spacing"
3 weeks ago
Anonymous
A redditor called Tioga060 uses the phone when using LULZ, gets exhausted from scrolling and wants to scroll less.
Because he knows how newfag infested LULZ is and how easily manipulated they are because of their stupidity and paranoia, he comes up with a hoax and spends two years forcing it until successfully brainwashing them that Reddit invented paragraphs and they can became an oldfag by Reddit typing i.e. putting the text in one line or making it appear as a block.
Mods and feds promote the hoax to make more money by buying less storage and printing less.
Phoneposters or anyone on a tiny screen promote the hoax to scroll less.
But those mostly Reddit typing or spamming "reddit spacing" when seeing paragraphs are paranoid post-2010 newfags.
3 weeks ago
Anonymous
You probably giggled to yourself while typing that out.
3 weeks ago
Anonymous
It's a mere coincidence that people started typing like idiots without being told to fuck off ever since "reddit spacing" has been mentioned 63000 times since 2016, "reddit spacing' being mentioned 8 times and 46 times between 2012-2014 and 2014-2016 respectively, normie overtaking normalfag in 2014 without anyone wondering why and gamergate happening in 2014.
>>if it's safe then it's not destroying anything >wrong, as a matter of fact, it's everything but safe
Then it's not safe retard. Note the "if". The claim that "because something it's safe it doesn't mean it's the best approach" is retarded and wrong (and badly spelled).
True. The picture looks identical when you just watch instead of zooming in on still frames. Kind of crazy how mpv uses less resources to do it too. No wonder the madvr envy uses a 4080.
also the more I look at it, it feels like there's wasted space on the bottom. it feels visually unbalanced that there's net thicker padding at top than the bottom. having the osc seekbar fill the entire height removes this illusion
also the more I look at it, it feels like there's wasted space on the bottom. it feels visually unbalanced that there's net thicker padding at top than the bottom. having the osc seekbar fill the entire height removes this illusion
Maybe code from some guy that thought PR meant "public relations" shouldn't be accepted...
One problem this modified seekbar would have though is that the timestamp pop-up as you move the mouse around would cover the filename/file title. Right now it just pops up on the seekbar itself.
I'm still not against the idea of extending out the seekbar and moving the buttons up though
I guess I don't use multimedia keys but have you tried testing the input?
Run mpv --input-test --force-window --idle and then press the key that you want to bind. What's the output tell you?
Or try --input-keylist for a known list. But this might not pick up all keys that are available to you like the first test.
>It should. There's a fuck-off-huge chart listing every pixel format
I didn't explain myself correctly, I meant swapchain format, on d3d11 and a 10-bit monitor the swapchain is automatically detected as R10G10B10A2 but on vulkan the log doesn't say anything about it.
nvm I am retarded, I found it:
[ 0.481][v][vo/gpu-next/libplacebo] Picked surface configuration 4: VK_FORMAT_A2B10G10R10_UNORM_PACK32 + VK_COLOR_SPACE_SRGB_NONLINEAR_KHR
It looks like it's using 10-bit too, pretty cool haasn.
I wonder what would happen if I set my GPU to 12-bit output, would it use "VK_FORMAT_R16G16B16A16_SFLOAT" or what?
Doesn't work, at least on Linux. mpv requires an external plugin to support multimedia keys which seems a little ridiculous for... a multimedia player.
https://github.com/hoyon/mpv-mpris
I guess I don't use multimedia keys but have you tried testing the input?
Run mpv --input-test --force-window --idle and then press the key that you want to bind. What's the output tell you?
Or try --input-keylist for a known list. But this might not pick up all keys that are available to you like the first test.
Can't do that on Android. And why do I have to use the terminal for such a basic fundamental feature that every video player on god's green earth has had since day 1?
>android
Yea... that would have been good info to have, wouldn't it? You're on your own with that one. Or you can ask the maintainer of the android version.
And using the terminal isn't the only way. Did you read the full post?
Nobody does that lmao. Who actually drags folders into their video player? Everyone goes into a directory, opens one file, and then goes to next and previous files from within the media player. It has always been like this. MPV is the exception.
There's only one thing to do. Become active with the mpv devel and prove you know your shit. Then after you've proven yourself get privilege to merge the good shit. ezpz
sounds like an excuse for the current devs to be lazy
>sounds like an excuse for the current devs to be lazy
It very well could be. But it's not like you're paying them. You have to be the change you want to see. Or just keep bringing it up in IRC with irrefutable evidence that they are retards for not merging certain PRs. But the evidence has to be solid.
adding swift to the codebase is worse than the sum total of what llyyr has done. I still have no idea what he was smoking or why the libmpv mac backend spans like 5 different files. https://github.com/mpv-player/mpv-examples/blob/master/libmpv/cocoa-rendergl/cocoa-rendergl.m is pretty much identical to the mac libmpv backend and is only ~300 loc. It's missing mouse and keyboard reporting, but that wouldn't take much more to add.
There's only one thing to do. Become active with the mpv devel and prove you know your shit. Then after you've proven yourself get privilege to merge the good shit. ezpz
The people have spoken. Having the up/down wheel seek is stupid. It makes for sense for the default to adjust audio.
Why is sfan5 so desperatly trying to hold onto this? He can create his own custom keybind.
First thing I disable and his arguments for reverting are weak. It has been sitting a while and everyone seems to overwhelmingly prefer volume, with no one actually speaking up about the change yet.
Everyone and their mother thinks seeking with the mouse wheel is weird. Just let it go sfan5. Window users will cry a storm whenever something breaks and they're always on the latest git.
The people have spoken. Having the up/down wheel seek is stupid. It makes for sense for the default to adjust audio.
Why is sfan5 so desperatly trying to hold onto this? He can create his own custom keybind.
I love how the original PR has a single thumbs down and it's by sfan lmao
lol a simple pr to change the int64 to double (which is strictly better than the current scenario) gets pushback https://github.com/mpv-player/mpv/pull/12356
no it wouldn't. Things just don't suddenly break beacuse you cast a double to an int. I bet you're one of the retards who treats floating point as black magic.
First of all, I thought haasn was a full-time uosc user now, who the fuck cares what he thinks about anything OSC or UX related? I swear to god if he merges this thinner seekbar shit I'm going to stop using mpv.
>I swear to god if he merges this thinner seekbar shit I'm going to stop using mpv.
That dude didn't even know what a pull request was. It should just be closed honestly.
kasper reasonably assumed that someone making osc code changes would be capable of making a PR. Unfortunately this doesn't seem to be the case and I'm betting that guy made his half-broken PR using github's webshit interface.
also while we're talking about osc, here's an improvement freebie: when the osc is not shown, there's no need to subscribe to playback position event. It's just wasting cycles. You should unobserve the property when osc is hidden, and re-observe when osc is shown.
if i had an account to do that, I wouldn't be reporting issues on LULZ. Although reporting issues on LULZ seems to be more effective anyhow (it helps that I only report issues that I'm 99% confident in, have already found the root cause for, and know roughly how to fix it but can't be bothered to PR).
>know how to program >familiar with the mpv code base >doesn't have a github account
bullshit you just don't want to get embarrassed like sizumam and become a meme
beats me, I didn't look at it. I just assumed that when you start the redraw, there's no guarantee that the property notification for the playback position comes before the next redraw
I don't follow what the issue would be. The playback position event is only used to fire the redraws (technically it calls "tick" which is a legacy from before wm4 refactored it to be observe-based instead of tick-event based). So a redraw would happens whenever the mouse moves anyway.
Also I'll do the hard work of debugging for you and tell you that when you do this, you'll run into an issue where OSC mouse events seem to suddenly stop working when you go into fullscreen. The root cause is because apparently the OSC reconfigures the region for which it receives mouse events (I had no idea you could do that) on every call to tick(), which causes the event region to be too small if you fullscreen while OSC is hidden. To fix you need to observe osc-width property and call tick() on that too.
Tbh the entire OSC code is kind of shitty, feels like pajeet code instead of clean wm4 code. You observe properties, but don't actually save those observed properties anywhere, so you query those exact same properties in the render function anyway. And like I mentioned for some reason every single draw cycle you reconfigure the mouse area.
3 weeks ago
Anonymous
>Tbh the entire OSC code is kind of shitty
OSC refactor soon(tm).
Bottombar will be better than ever.
3 weeks ago
Anonymous
i have no idea if this is a meme or a comment from one of the actual devs.
3 weeks ago
Anonymous
Looking forward to it
u-uh I was just memeing h-haha
3 weeks ago
Anonymous
Looking forward to it
3 weeks ago
Anonymous
>OSC refactor soon(tm).
Finally.
3 weeks ago
Anonymous
3 weeks ago
Anonymous
covers too much of the video
3 weeks ago
Anonymous
its just a drawing, how can you even ascertain that? lgtm
3 weeks ago
Anonymous
Volume slider should never be separate from play button
3 weeks ago
Anonymous
>Volume slider should never be separate from play button
there's literally no media player where that's the case, unless your only frame of reference is webshit like youtube or twitch
3 weeks ago
Anonymous
the play button could just be removed entirely, what purpose does it serve? the majority of users don't click on it and it's useless as an indicator that the video is playing. why is it taking up space on the osc when you can play/pause with click and spacebar. someone should make a PR removing the play button and also the play symbol from the mpv logo for maximum minimalism. with enough eyes emojis it'll get merged
3 weeks ago
Anonymous
I get where you're coming from. But we need some symbolic indication of when the video is played or paused, right?
3 weeks ago
Anonymous
you mean like the moving video and changing audio?
3 weeks ago
Anonymous
Exactly 🙂 lol
3 weeks ago
Anonymous
I have my right click bound to something else. What now?
[...]
well what if you have a video consisting of identical frames and no audio. What now, smartypants?
this needs to be hashed out in a pr. dokuro~~
3 weeks ago
Anonymous
if this nocoder da3dsoul1 guy can make a pr without knowing what a pr is then surely our dokuro can make one
>making a PR >when I barely even watch anything anymore
mpv makes me feel nauseated ever since the VRR incident. I haven't really sat down and enjoyed anything ever since.
3 weeks ago
Anonymous
exactly. if you're not going to be watching anything then focus on making the remove play button pr for us
3 weeks ago
Anonymous
have you tried the refreshing cone of vlc?
3 weeks ago
Anonymous
you mean like the moving video and changing audio?
well what if you have a video consisting of identical frames and no audio. What now, smartypants?
3 weeks ago
Anonymous
Isn't that a still image?
3 weeks ago
Anonymous
I have my right click bound to something else. What now?
3 weeks ago
Anonymous
top bars are disgusting
3 weeks ago
Anonymous
although seems the mouse reconfigure just sets some state over in C land, so it's not really more expensive than a property get, and we only do it at 30fps anyway. You can get the bulk of the benefit without rewriting the entire thing by just unobserving the playback pos event once hidden, which basically avoids unnecessary calls to tick().
https://github.com/mpv-player/mpv/blob/master/meson.build#L1290
because the check is only if the vulkan header exists, not if the target system is linux
https://github.com/mpv-player/mpv/blob/master/meson.build#L1290
because the check is only if the vulkan header exists, not if the target system is linux
Because the code compiles on windows even though it's impossible to use on that platform.
Talking about windows plus linux, has anyone tried running mpv under WSL2?
It was a terrible hack. I wish just once DMG would actually finish and test something before merging it, because he seems to be on a serial rampage of merging broken shit and having to fix it multiple times.
yeah actually lets just delete all of the bcspline aliases and let users configure B&C params themselves, maybe they'll learn some cool math stuff along the way
This isn't a general constraint on the m-n class of bcsplines, it's just the subset of them that have quadradic (or better) convergence, per Keys. Popular exceptions are Hermite (0, 0) and Sharp Bicubic (0, 1).
3 weeks ago
Anonymous
>Sharp Bicubic
holy cr*p new meme just dropped
3 weeks ago
Anonymous
That's an old meme, it used to be the default in Premier or something.
3 weeks ago
Anonymous
>holy cr*p new meme just dropped
1/2
3 weeks ago
Anonymous
1/2
3 weeks ago
Anonymous
>Sharp Bicubic (0, 1)
you can spot something upscaled with this crap so easily, those weird blocky edges are so fucking ugly.
>israeli man in the 20th century dies poor
That seems like a rare occurrence. Well he was possibly a commie I guess so it's not like everyone is perfect.
Sharp Bicubic has been a meme for over a decade ever since madshi added AR to madVR. Lots of users were using Bicubic100 (B =0, C = 1) with AR especially for downscaling
Nah, there were people who turned it up to 100. Bicubic150 + AR and Bicubic100 + AR were very common for downscaling and people who wanted sharp memes for upscaling used Bicubic100 + AR. It's an old meme.
profile=gpu-hq
deband=no
>deband=no
I see you're a man of culture as well.
profile=gpu-hq
Rate my config.
Playback related things that madVR does better than mpv:
>Proper RGB output where red means red (R:255 G:0 B:0), and green means green (R:0 G:255 B:0). No bullshit. This is currently not possible on mpv.
>Custom modeline support that allows for perfect frame pacing. No resampling required (23.976Hz, 24.000Hz, 29.970Hz, 59.940Hz...).
>Support for proper color management and 3DLUTs. Respects your intelligence and display calibration without raising your black levels.
>Proper projector support (lens memory number, cropping masked borders, anamorphic lens support).
>3D playback.
>madshi's custom SmoothMotion algorithm that is smoother than mpv's blurry temporal interpolation (tscale=oversample is NOT "smoothmotion")
>IVTC and deinterlacing
Rendering related tasks that madVR does better:
>*EXTENSIVE* HDR tonemapping options (mpv only has 2-3 tunable options)
>Better debanding.
>Can remove compression artifacts while keeping details intact. No CNNs required.
>Can remove ringing artifacts that are part of the encode.
>Support for optimally configured Jinc 3 (mpv is not optimally configured).
>Optimally configured KrigBilateral support without pixel shifting.
>Built-in 2D SSIM downscaler that is configured with Jinc (mpv shader does not support polar downscaling).
>Super-xBR support, a cheap alternative to Jinc 3 for high quality chroma scaling.
>Provides NGU upscaling CNN for high quality video output. Comes in various flavors: NGU Standard (better than FSRCNNX), NGU Sharp (not available on mpv), NGU AA (better than RAVU), NGU Soft (not available on mpv).
>High quality sharpening filters, such as LumaSharpen, AdaptiveSharpen, and SuperRes. Comes with debloating artifact removal.
>Support error diffusion dithering algorithms that are tuned by madshi himself for quality and speed. Better and sharper than mpv's gaussian blur blue noise smear pattern.
>Better antiringing.
>Better gamut mapping.
**UPDATE**
Better debanding has been confirmed.
mpv is a failure.
Does madVR support anything besides mpc-hc?
MadVR Envy, get your shitbox for 10K$
>MadVR Envy, get your shitbox for 10K$
Honestly how would a mpv shitbox look like in comparison? For home theater use obviously.
MPC-BE, Kodi, PotPlayer... there are probably others but I think those are the big ones.
Does MadVR support any video playing software (like MPC-HC, VLC, MPV, Etc.) besides MPC-HC?
see
the confidence with which I can assure that you have 0 sex it's unimaginable
Can I not download a video by ID number in yt-dlp?
You can if it's a youtube video.
what's the syntax? -f {number]? what if I wanna combine audio of one number to another?
By ID do you mean the numbers in the video URL or the format ID? If you mean format ID, you can do yt-dlp -f number url. If you want to combine different audio/video formats you could do yt-dlp -f number1+number2 url.
format ID, alright thanks.
https://slow.pics/c/BO91B6Yb
New default deband sucks
Time to git revert mia?
That anon is posting fud, see:
https://slow.pics/c/ZAhHJues
Other anon is not posting the new default
>Other anon is not posting the new default
Well, according to his post he is.
No, he is clearly not.
Seems retarded to test the limits of how much you should deband on a test pattern. Where are the details that can potentially be destroyed? Might as well just smooth it completely, that's "better" in this case.
haasn had it right the first time, fuck lachs0r.
I don't think you understand the meaning of FUD
>I don't think you understand the meaning of FUD
FUD:
fear, uncertainty, doubt
deband=yes
deband-iterations=4
deband-threshold=64
deband-range=8
>deband-iterations=4
2
Funnily enough, changing the deband-threshold to 33 (literally ONE point) is enough to remove the banding from this test image. Is there any reason for this? Is everything from 33-64 just the same thing?
Why do madvr users constantly post in these threads?
last time i checked this was also an ffmpeg and yt-dlp thread, and mpc-hc+madvr supports both of those programs
How do make keybinds with various ffmpeg filters like dynaudnorm with mpc-hc+madvr?
dynaudnorm is shit, so thankfully it doesnt let you do this, mpv users are so hearing impaired that they've never heard good audio normalization before
Like this
m af toggle "lavfi=[loudnorm=I=-16:TP=-3:LRA=4]"
Where do you put that?
What does mpc-hc use?
input.conf file
Where is that for mpc-hc?
Sir, this is an mpv ffmpeg yt-dlp thread
I was only asking because of this anon
Is he mistaken?
LAVFilters are ffmpeg
and MPC-HC works perfectly fine with yt-dlp
But how can I make keybinds with various audio and video filters? Or can I with mpc-hc?
I don't think MPC has built-in debanding, the madvr one looks better but is way way more GPU intensive.
Yeah that's what I meant. Oops.
He's the only mpv dev who's also an ffmpeg dev other than haasn. That's gotta count for something. In general this whole "wontfix because you can fix it yourself" attitude pervades the entire dev team, but this time it's more warranted since it's a dumb subjective cosmetic change.
> only mpv dev who's also an ffmpeg dev other than haasn
anyone can be an ffmpeg dev, just submit patches. And philipl regularly does work in ffmpeg related to nvdec.
I mean he has repo access you dirtclod
>dirtclod
what a weird insult
>Shin Sekai Yori
>Sokudo
>Low effort speed mini encodes meant for 2Mbps and data cap third-worlders
Don't ruin a good show by watching a shit encode, especially a show with grain.
>shinsekai yori
>good show
Rtard
it was homosexual garbage, typical A1 pictures slop
deband used to be good...
There's barely any difference between these pics
You are blind
Blind how? I'm not saying there's no difference, I'm saying the difference is very small.
All the banding is still there, it's just slightly masked but you can still see it.
what
its literally night and day
oof
Post the original video.
That literally shows it barely doing anything. Fix your monitor's gamma.
Fix your eyes, bilinearchad.
If you uploaded an encode with banding comparable to what's on the right to a hypothetical tracker I owned I would ban you. The defaults barely do anything. Eyes. Look.
Is this a bot? There's no image on the right, it's a top/bottom image comparison.
>2023
>ffmpeg still can't play SACD ISOs
https://trac.ffmpeg.org/
no.
not you again
https://github.com/mpv-player/mpv/issues/12358
I swear Traneptora posts the most braindead, borderline trolling comments.
I see you've modified the osc and provided screenshots. Did you know you can modify the osc and put it in scripts/ to use it?
trannyptora
He is brain dead, I thought that's common knowledge?
He regularly contributes to jpegxl after all
Yeah I dunno, but to be honest, I really like this layout more than the default one. Seekbar fill could be a little thicker I guess, and maybe the chapter indicators could be more clear (maybe invert the colors on the chapter indicators and fill the seekbar completely?). But moving the alang/slang buttons to the top is a good idea IMO.
It should also make the osc look better when its truncated in windowed mode. But to be honest, there should be an option to make the OSC not scale with window size to go along with this.
One problem this modified seekbar would have though is that the timestamp pop-up as you move the mouse around would cover the filename/file title. Right now it just pops up on the seekbar itself.
>One problem this modified seekbar would have though is that the timestamp pop-up as you move the mouse around would cover the filename/file title
I'm pretty sure this is a problem with thumbfast.lua
>both haasn and kasper approve
uh oh, new drama incoming?
looks interesting, but since he knows how to edit he should just do it, provide a modified version for people to test. He didn't even screenshot things properly (his screenshots are missing the fullscreen and audio buttons)
Why did he change his name to traneptora anyway. I preferred the_bombzen
needs a review from avih first.
Holy fucking shit guys.
OK, placate my advanced schizophrenia for a second, I think this github user is the person who wrote the kokomins cargocult blog.
>Wow, your osc is nice and clean, could you provide a guide on how to edit it from original like that? I want to use latest osc from mpv source to make it look like yours!
>To be frank it was kinda jury rigged from an old version, took days of testing to get the elements placed right. Some other 3rd party mpv builds have unique osc and I just copied code of elements I liked. Unfortunately how to edit the lua code isn’t well documented so even I can’t figure out how to properly make one from scratch.
Pic related, a file from the blog. Looks identical to this guy's OSC.
I want to believe
OK from digging further into the blog, it seems like kokomins blog actually provides an mpv build or something with a custom osc.lua. (I aint clicking that shit nigga)
It looks like this guy might've just downloaded it from this blog, and modified it to work with thumbfast.lua because the kokomins guy says:
>Gotcha, is it possible to implement this in your OSC? https://github.com/po5/thumbfast It would be the perfect combo in my opinion
>Unfortunately this is out of my coding abilities, not do I currently have time to dedicate to hobbies anymore. What I suggest is that you could compare their osc.lua to the vanilla file and figure out how they implemented it, and if it only requires small modifications try to implement in mine.
My best guess is that he's just passing the kokomins osc.lua as his own creation. This explains his previous issue as well.
https://github.com/mpv-player/mpv/issues/12349
This is some deep lore. This was also the image used on the april fools mpv page right? I remember an anon asking why the osc there looks different.
>Unfortunately how to edit the lua code isn’t well documented
What did he mean by this
>This was also the image used on the april fools mpv page right?
Lol I made that image. It's not an actual april fools page, it's just a quick inspect element meme. When you search "mpv" and "anime" I'm pretty sure that image from the kokomins blog is what pop up first.
that homosexual always posts the most useless retarded comments just like that llyyr tranny. he's a literal retard
he trooned out
Traneptora? More like trannyptora.
>llyyr tranny
Now now, llyyr may be many things (weeb, pajeet) but I have not seen evidence that he's a tranny. And statistically I've never seen an indian person succumb to trannyism (probably because they know they'll never pass)
>he trooned out
But his name is still Leo
always hated llyyr, indians should be banned from github
>replying to bait
it's not bait, it's the truth
look at the screencap below, he's very proud of it
Is he Indian?
Screenshot the updated post
Is that the cunt thats trying to force the "reddit spacing hoax" meme?
yes, it's also the guy that gets irrationally angry if you say ffmpeg instead of "FFmpeg", they're the same guy
https://desuarchive.org/g/thread/95849923#95858179
And the actual schizo who thinks everyone on LULZ is one person and calls everyone a schizo is now talking to himself to force his retarded narrative.
ffmpeg
>maybe if I samefag I can convince people to believe in my delusion that one guy spelling it as ffmpeg and hurting my tranny friend is the same as two other guys who made me sperg out by spelling it as FFmpeg and making fun of me for falling for the Reddit spacing hoax, and that everyone on LULZ that makes me lose my mind is a schizo, irrationally angry and one person
Seek help.
ffmpeg
Yeah no surprise the newfag fell for that hoax is a schizo because only a schizo newfag would think reddit invented paragraphs and typing like an idiot is how oldfags typed
Give me the qrd of that retarded meme "reddit spacing"
A redditor called Tioga060 uses the phone when using LULZ, gets exhausted from scrolling and wants to scroll less.
Because he knows how newfag infested LULZ is and how easily manipulated they are because of their stupidity and paranoia, he comes up with a hoax and spends two years forcing it until successfully brainwashing them that Reddit invented paragraphs and they can became an oldfag by Reddit typing i.e. putting the text in one line or making it appear as a block.
Mods and feds promote the hoax to make more money by buying less storage and printing less.
Phoneposters or anyone on a tiny screen promote the hoax to scroll less.
But those mostly Reddit typing or spamming "reddit spacing" when seeing paragraphs are paranoid post-2010 newfags.
You probably giggled to yourself while typing that out.
It's a mere coincidence that people started typing like idiots without being told to fuck off ever since "reddit spacing" has been mentioned 63000 times since 2016, "reddit spacing' being mentioned 8 times and 46 times between 2012-2014 and 2014-2016 respectively, normie overtaking normalfag in 2014 without anyone wondering why and gamergate happening in 2014.
>>if it's safe then it's not destroying anything
>wrong, as a matter of fact, it's everything but safe
Then it's not safe retard. Note the "if". The claim that "because something it's safe it doesn't mean it's the best approach" is retarded and wrong (and badly spelled).
>phoneposting
>MPV and MadVR when both are configured correctly one still looks ever so slightly different from the other.
Yeah im not gonna be that autistic about a few pixels looking slightly different.
I could put that effort into actually watching cool shows.
True. The picture looks identical when you just watch instead of zooming in on still frames. Kind of crazy how mpv uses less resources to do it too. No wonder the madvr envy uses a 4080.
>Kind of crazy how mpv uses less resources to do it too
because madshi can't into C
More like madshi can't into GPU programming, vulkan/d3d12, or probably low-level optimization in general
No, madshi literally can't into the Windows API. He literally cannot make a window show up on the screen.
sizumam is getting really uppity lately, someone needs to peg him down
for me it's deband=never
RIP
you could mask out the portion behind the tooltip
also the more I look at it, it feels like there's wasted space on the bottom. it feels visually unbalanced that there's net thicker padding at top than the bottom. having the osc seekbar fill the entire height removes this illusion
Maybe code from some guy that thought PR meant "public relations" shouldn't be accepted...
I already said this was going to be an issue...
I'm still not against the idea of extending out the seekbar and moving the buttons up though
I'm just glad someone is finally brainstorming ideas. Bottombar has always felt lacking and like the seekbar was unnecessarily thicc
SHUT THE FUCK UP HOLY SHIT TALK ABOUT THE SOFTWARE
I'm just waiting for ewa_lanczos4sharpest to be mapped...
What's the reason mpv doesn't support multimedia keys like mpc-hc?
But I pause with my media key every single day...
I guess I don't use multimedia keys but have you tried testing the input?
Run mpv --input-test --force-window --idle and then press the key that you want to bind. What's the output tell you?
Or try --input-keylist for a known list. But this might not pick up all keys that are available to you like the first test.
It does. Kill yourself.
It's literally native.
It should. There's a fuck-off-huge chart listing every pixel format + codec + graphics api combo somehwere on the ffmpeg wiki but I can't find it.
>It should. There's a fuck-off-huge chart listing every pixel format
I didn't explain myself correctly, I meant swapchain format, on d3d11 and a 10-bit monitor the swapchain is automatically detected as R10G10B10A2 but on vulkan the log doesn't say anything about it.
nvm I am retarded, I found it:
[ 0.481][v][vo/gpu-next/libplacebo] Picked surface configuration 4: VK_FORMAT_A2B10G10R10_UNORM_PACK32 + VK_COLOR_SPACE_SRGB_NONLINEAR_KHR
It looks like it's using 10-bit too, pretty cool haasn.
I wonder what would happen if I set my GPU to 12-bit output, would it use "VK_FORMAT_R16G16B16A16_SFLOAT" or what?
NEXT playlist-next
PREV playlist-prev
PLAYPAUSE cycle pause
werks on my machine
meant for
Doesn't work, at least on Linux. mpv requires an external plugin to support multimedia keys which seems a little ridiculous for... a multimedia player.
https://github.com/hoyon/mpv-mpris
mpv doesn't detect multimedia keys.
Just use a socket dummy.
I use Windows and it just works... so not really my problem I guess.
>mpv doesn't detect multimedia keys.
Works fine on mac.
>1990+43
>Still no native option to play next file in directory
No I'm not going to install a "script"
But there is a native option to play the next file in a directory. Play the folder via terminal or drag the folder onto the mpv window.
Can't do that on Android. And why do I have to use the terminal for such a basic fundamental feature that every video player on god's green earth has had since day 1?
>android
Yea... that would have been good info to have, wouldn't it? You're on your own with that one. Or you can ask the maintainer of the android version.
And using the terminal isn't the only way. Did you read the full post?
Nobody does that lmao. Who actually drags folders into their video player? Everyone goes into a directory, opens one file, and then goes to next and previous files from within the media player. It has always been like this. MPV is the exception.
Can Vulkan output rgb10_a2 presentation surfaces like d3d11? Using W11.
>103 open PRs
Why are mpv devs like this?
it's because of llyyr spamming useless PRs https://github.com/mpv-player/mpv/pull/12337
still a better /dev/ than Akemi
sounds like an excuse for the current devs to be lazy
>sounds like an excuse for the current devs to be lazy
It very well could be. But it's not like you're paying them. You have to be the change you want to see. Or just keep bringing it up in IRC with irrefutable evidence that they are retards for not merging certain PRs. But the evidence has to be solid.
>still a better /dev/ than Akemi
Akemi wasn't a pajeet last time I checked
Hes German I think.. and a brain dead Apple user.
adding swift to the codebase is worse than the sum total of what llyyr has done. I still have no idea what he was smoking or why the libmpv mac backend spans like 5 different files. https://github.com/mpv-player/mpv-examples/blob/master/libmpv/cocoa-rendergl/cocoa-rendergl.m is pretty much identical to the mac libmpv backend and is only ~300 loc. It's missing mouse and keyboard reporting, but that wouldn't take much more to add.
There's only one thing to do. Become active with the mpv devel and prove you know your shit. Then after you've proven yourself get privilege to merge the good shit. ezpz
https://github.com/mpv-player/mpv/pull/12329
>be mpv dev
>create PR
>merge it because I can
>be another mpv dev
>dislike that PR
>make my own PR that reverts it
>have no balls to just merge it.
See
without wm4 to call them all retards these guys are truly helpless
The people have spoken. Having the up/down wheel seek is stupid. It makes for sense for the default to adjust audio.
Why is sfan5 so desperatly trying to hold onto this? He can create his own custom keybind.
Agreed, the volume via mouse wheel is the first thing I looked into changing when I first started using mpv.
quicktime seeks. Do you not trust the hig folks at apple?
see 2nd half of
First thing I disable and his arguments for reverting are weak. It has been sitting a while and everyone seems to overwhelmingly prefer volume, with no one actually speaking up about the change yet.
>everyone seems to overwhelmingly prefer volume
The change isn't even released to repos yet, I doubt most people are building mpv from head
all windows users are (which is probably like 50% of mpv's userbase)
Everyone and their mother thinks seeking with the mouse wheel is weird. Just let it go sfan5. Window users will cry a storm whenever something breaks and they're always on the latest git.
can confirm, my mom uses VLC
>Having the up/down wheel seek is stupid
it's bad.. because... it just is okay!?!?!
it's bad because no one uses the scroll wheel to seek
The seek bar is visualized horizontally. Having a vertical motion (up/down wheel) control it is extremely unintuitive.
what now vlcisters
sfan5, my dude. Listen, the majority thinks that using the mouse wheel to seek creates more problems. And it's not even accurate for most people.
do you want drama of devs reverting each other's commits
yes
I love how the original PR has a single thumbs down and it's by sfan lmao
just add a bot to auto-close PRs after being ignored long enough, problem solved
lol a simple pr to change the int64 to double (which is strictly better than the current scenario) gets pushback https://github.com/mpv-player/mpv/pull/12356
see
because in its current state it'd break stuff due to a mismatch of types. Did you read or just skip over blocks of text?
>it'd break stuff due to a mismatch of types
Not a single person in that discussion said this.
no it wouldn't. Things just don't suddenly break beacuse you cast a double to an int. I bet you're one of the retards who treats floating point as black magic.
9/10 chance this simple PR causes an issue eventually because it's authored by dmg. Only half joking
Is there any reason why I shouldn't just use ravu-zoom and krig for everything? My frametimes are under 1ms.
https://github.com/mpv-player/mpv/issues/12342#issuecomment-1707491434
haasn is basically admitting that vo_gpu is superior
Can't we just set the default action for scroll up/down to handle playback direction.
that's even dumber good job
NOOOOOOO THAT WOULD BLOAT MY INPUT.CONF!!!!!
Everyone listen up... I solved it...
AXIS_UP ignore
AXIS_DOWN ignore
>AXIS_
Why not use Wheel_ ? Are you afraid of rubber?
>Deprecated aliases for ``WHEEL_*``.
>implying that everyone uses a physical scroll wheel
how do you have a virtual scroll wheel wtf
found the apple user
First of all, I thought haasn was a full-time uosc user now, who the fuck cares what he thinks about anything OSC or UX related? I swear to god if he merges this thinner seekbar shit I'm going to stop using mpv.
Doesn't he use plplay ?
>I swear to god if he merges this thinner seekbar shit I'm going to stop using mpv.
That dude didn't even know what a pull request was. It should just be closed honestly.
Some of the ideas were good though. Moving buttons to the top to let the progess bar expand across the whole window is a good idea.
kasper told him to make a PR, the dude clearly wanted to discuss the changes first.
kasper reasonably assumed that someone making osc code changes would be capable of making a PR. Unfortunately this doesn't seem to be the case and I'm betting that guy made his half-broken PR using github's webshit interface.
What happened to thumbfast?
po5 had a meltdown
typical discord user moment, dont use scripts from unstable people
Any good alternative?
no
thumbnails are bloat anyways, and if youre a windows user it literally wears down your drive
put it in ramdisk, problem solved
>ramdisk
what is this, 2009?
also, just dont use scripts from mentally unwell individuals, thats way easier
WHERES THE GUI
The what?
also while we're talking about osc, here's an improvement freebie: when the osc is not shown, there's no need to subscribe to playback position event. It's just wasting cycles. You should unobserve the property when osc is hidden, and re-observe when osc is shown.
open an issue so I don't forget when I'm free in a week or so
if i had an account to do that, I wouldn't be reporting issues on LULZ. Although reporting issues on LULZ seems to be more effective anyhow (it helps that I only report issues that I'm 99% confident in, have already found the root cause for, and know roughly how to fix it but can't be bothered to PR).
>know how to program
>familiar with the mpv code base
>doesn't have a github account
bullshit you just don't want to get embarrassed like sizumam and become a meme
bingo
Wouldn't that create a race condition when you show the osc again?
how so? osc events are coalesced anyway, and the property values that are observed are not actually saved anywhere, it's only used to fire the redraw
beats me, I didn't look at it. I just assumed that when you start the redraw, there's no guarantee that the property notification for the playback position comes before the next redraw
I don't follow what the issue would be. The playback position event is only used to fire the redraws (technically it calls "tick" which is a legacy from before wm4 refactored it to be observe-based instead of tick-event based). So a redraw would happens whenever the mouse moves anyway.
Ignore my dumbass. I already said I didn't look at the code.
Also I'll do the hard work of debugging for you and tell you that when you do this, you'll run into an issue where OSC mouse events seem to suddenly stop working when you go into fullscreen. The root cause is because apparently the OSC reconfigures the region for which it receives mouse events (I had no idea you could do that) on every call to tick(), which causes the event region to be too small if you fullscreen while OSC is hidden. To fix you need to observe osc-width property and call tick() on that too.
Tbh the entire OSC code is kind of shitty, feels like pajeet code instead of clean wm4 code. You observe properties, but don't actually save those observed properties anywhere, so you query those exact same properties in the render function anyway. And like I mentioned for some reason every single draw cycle you reconfigure the mouse area.
>Tbh the entire OSC code is kind of shitty
OSC refactor soon(tm).
Bottombar will be better than ever.
i have no idea if this is a meme or a comment from one of the actual devs.
u-uh I was just memeing h-haha
Looking forward to it
>OSC refactor soon(tm).
Finally.
covers too much of the video
its just a drawing, how can you even ascertain that? lgtm
Volume slider should never be separate from play button
>Volume slider should never be separate from play button
there's literally no media player where that's the case, unless your only frame of reference is webshit like youtube or twitch
the play button could just be removed entirely, what purpose does it serve? the majority of users don't click on it and it's useless as an indicator that the video is playing. why is it taking up space on the osc when you can play/pause with click and spacebar. someone should make a PR removing the play button and also the play symbol from the mpv logo for maximum minimalism. with enough eyes emojis it'll get merged
I get where you're coming from. But we need some symbolic indication of when the video is played or paused, right?
you mean like the moving video and changing audio?
Exactly 🙂 lol
this needs to be hashed out in a pr. dokuro~~
>making a PR
>when I barely even watch anything anymore
mpv makes me feel nauseated ever since the VRR incident. I haven't really sat down and enjoyed anything ever since.
exactly. if you're not going to be watching anything then focus on making the remove play button pr for us
have you tried the refreshing cone of vlc?
well what if you have a video consisting of identical frames and no audio. What now, smartypants?
Isn't that a still image?
I have my right click bound to something else. What now?
top bars are disgusting
although seems the mouse reconfigure just sets some state over in C land, so it's not really more expensive than a property get, and we only do it at 30fps anyway. You can get the bulk of the benefit without rewriting the entire thing by just unobserving the playback pos event once hidden, which basically avoids unnecessary calls to tick().
What is the "displayvk" gpu-context?
>VK_KHR_display. This backend is roughly the Vukan equivalent of DRM/EGL, allowing for direct rendering via Vulkan without a display manager.
Weird, why is it listed on the windows build?
Because the code compiles on windows even though it's impossible to use on that platform.
https://github.com/mpv-player/mpv/blob/master/meson.build#L1290
because the check is only if the vulkan header exists, not if the target system is linux
windows is just that good
Talking about windows plus linux, has anyone tried running mpv under WSL2?
I don't think you can use graphical applications with WSL2 yet.
?feature=shared&t=225
>there's a Microsoft Azure server engineer RIGHT NOW using mpv to watch anime during his shifts
I believe
i don't see the mpv demo anywhere? what's the timestamp
Yes. Microsoft literally did a demo where they ran mpv as one of their examples. This was a few years ago.
huh
sauce? sounds funny
https://github.com/mpv-player/mpv/commit/a20ae041
its what you use when you get fed up and uninstall the crapware known as waymeme
running mpv over drm with vulkan
test
can yt-dlp download pornhub now? It worked for me few months ago but idk if that's still possible
yis
>Busty Ebony Step sis Sarah Banks Gags On Step Bro's Fat White Cock
https://github.com/mpv-player/mpv/pull/12362/files
why remove the auto option
It was a terrible hack. I wish just once DMG would actually finish and test something before merging it, because he seems to be on a serial rampage of merging broken shit and having to fix it multiple times.
He needs 204 more commits to surpass pigoz and have his name listed right next to haa-chan.
I'll read Demonbane is his honor when that happens.
@haasn, the BC parameters for hermite are 0:0, pls fix:
https://github.com/haasn/libplacebo/blob/master/src/filters.c#L483
oh yeah I'm sure all of the hermite users were really complaining about this one...
Then why have the filter at all?
Better delete it than having an incorrectly defined filter.
yeah actually lets just delete all of the bcspline aliases and let users configure B&C params themselves, maybe they'll learn some cool math stuff along the way
You just literally need to follow B+2C=1
uh yeah? just dont put the formula in the manual, let them figure it out themselves using desmos
This isn't a general constraint on the m-n class of bcsplines, it's just the subset of them that have quadradic (or better) convergence, per Keys. Popular exceptions are Hermite (0, 0) and Sharp Bicubic (0, 1).
>Sharp Bicubic
holy cr*p new meme just dropped
That's an old meme, it used to be the default in Premier or something.
>holy cr*p new meme just dropped
1/2
1/2
>Sharp Bicubic (0, 1)
you can spot something upscaled with this crap so easily, those weird blocky edges are so fucking ugly.
send a PR loser
sorry no we rely only on our dokuromam to represent us on the githubs. he is like our lawyer he speaks on our behalf. take it up with him.
Delete all BCSplines and all ewa_lanczos variants.
luckily for you the ewa_lanczos variants aren't even available options yet
oh also delete lanczos, all you need is sinc for this
scale=sinc
scale-window=sinc
scale-radius=3
fixed your media player bloat, thank me later
You can't a window for sinc on mpv anon.
catmul catmul catmul!
Liz Love!!
screenshot-png-compression=0
screenshot-high-bit-depth=yes
I use compression 4 and then run ect -9 before upload
>screenshot-high-bit-depth=yes
screenshot-high-bit-depth=no
rgb8 is enough
spline36 is for chads. Guaranteed to be bug free since it was wm4's scaler of choice, no need to mess about with parameters, it just werks.
Okay now explain how Derche derived Spline36, Anon.
it came to him in a dream
lanczos16 plus AR is endgame
Lanczos = israeli
Catmull = American
Derche = German
Robidoux = French
choose your path wisely
Isn't Robidoux canadian?
Isn't he from Quebec though? He mostly communicates in french.
>Quebec
Still a leaf
I mean I hate israelites but those guys were good at math
Well everyone listed there was good at math so its a bit of a moot point. Catmull probably has the most accomplishments.
Lanczos almost certainly has more things named after him therefore he has more accomplishments.
Yeah well, like most mathematicians from that era Lanczos probably died poor, and Catmull made big bucks on LucasFilms, Disney and Pixar.
>israeli man in the 20th century dies poor
That seems like a rare occurrence. Well he was possibly a commie I guess so it's not like everyone is perfect.
you mean numbers schizo
stop bullying sizumam
I'll consider it when shitposting decreases
stop bullying llyyr
Sharp Bicubic has been a meme for over a decade ever since madshi added AR to madVR. Lots of users were using Bicubic100 (B =0, C = 1) with AR especially for downscaling
I think you're confusing it with Bicubic75+AR, which is what MadVR used to use for chroma by default I'm pretty sure.
Nah, there were people who turned it up to 100. Bicubic150 + AR and Bicubic100 + AR were very common for downscaling and people who wanted sharp memes for upscaling used Bicubic100 + AR. It's an old meme.
https://code.videolan.org/videolan/libplacebo/-/commit/c4900a41ca8ccb64f913a02b50d0d67e842a642d
Pretty based, now we can use dither-size-fruit=8
uh, ok
retard here. what benefit is there in using a larger dither matrix size?
if this nocoder da3dsoul1 guy can make a pr without knowing what a pr is then surely our dokuro can make one
What the fuck is this ugly shit?
https://github.com/mpv-player/mpv/pull/12307
Looks related to this and should be a bug going by PR and commit descriptions.
wdym? looks good
It's useless/redundant because no cropping is happening
There's cropping 1920x1088 to 1920x1080
Somehow missed that and mediainfo reports 1920x1080. Still think it's ugly as fuck and it sits in the center of stats and is the longest stat.
>Somehow missed that
So it should be bigger?
It shouldn't be on that line.
Should it be in the middle of the screen so it's more obvious?
Sure. 100% dudemanguy would merge that without looking
There's no indication those are values for cropping
There's now two 'native' resolutions reported
Shouldn't y be non-zero if it's cropped?
Why was this merged?
the final solution to the dudemanguy question is to make all of his PRs undergo avih review
Why is kasper so obsessed with cropping anyway? He has been fucking with it for what feels like months
How are pajeets so good at tech support https://github.com/mpv-player/mpv/issues/12364
just use system volume what the fuck