Thread 105899825

mpv.io/installation/
ffmpeg.org/download.html
github.com/yt-dlp/yt-dlp

>>105899825 (OP)
why would anyone choose the ffplaypill over the mpvpill?

>>105899825 (OP)
>mpv has no UI
Use SMPlayer

What's good mp4 settings for 4chan posting?

I can't goon.

Do you expect me to download each yt video and watch it through mpv? And why would I need ffmpeg for this? Except as a dependency for yt-dlp, of course.

>>105900268
mpv <url>

>>105900052
ffmpeg -i file.mp4 -c:v copy -an out.mp4

There is no other use for Mp4 other than the fact that you can now post the shitty mp4 clips you find on the Internet without having to re-encode them again.

>>105900531
>-c:v copy
this would never work

>>105900568
It would work, it keeps the video stream in tact but just strips out the audio because 4chan is gay and doesn't allow it on most boards.

>>105900296
Wow, didn't know that, thanks! Tho I still wonder why would I do this instead of using the browser (with uBlock of course)

>>105900708
lightweight, scriptable, video takes up the entire window, better scaling, better colours, better ui, better subs etc etc

>>105900268
No, if you really want to watch YT in mpv use FF2mpv for links and Freetube to browse YT. Freetube has a button to open the videos in mpv and also marks them as watched when you do so

When I try to play a Rumble video with mpv, it always picks the lowest quality (320x180) resolution. Is there a way to fix that?

I tried adding this to my mpv.conf but it didn't help at all:
ytdl-format=bestvideo[height<=?2160]+bestaudio/best


It seems that the issue is yt-dlp and it just doesn't pick the "bestvideo" resolution for Rumble and just defaults to lowest quality 180p.
Can someone please tell me what's wrong?

>>105899825 (OP)
no, troonware is troonware, stay pleb

Reminder that libplacebo is awful.

>>105900052
None. File size limit is too low and x264 shits itself as a result. I have a feeling gookmoot only added it so that pathetic zoomers from xitter and tiktok could upload mp4 slop they brought from there without any effort. VP9 webms have significantly better quality and compression at the same size - it's not even close.

>>105901300
Many people disagree with you.

>>105901300
>libplacebo is awful
I like how easy it is to apply video filters with it (esp. tonemapping) in FFmpeg.

>>105900791
freetube is still electron web browser so it’s not lightweight
>>105900708
use a lightweight youtube client instead of command line

>>105901431
Unused RAM is wasted RAM

>>105901468
all "unused" ram are used for file system cache and makes all programs you run and files opened more responsive

>>105899931
mpv does everything i want and everything i'll want but not know currently, also i customized it so much i don't want to switch

>>105901300
>libplacebo is awful.
libplacebo is an awful replacement for mpv. it's a fantastic replacement for lavfilters.

>>105900952
add an asterisk after bestvideo

>>105899825 (OP)
mpvnet > mpv (trash bare bones)

>>105899825 (OP)
use a better slop

>>105901513
my whole system is in cache right now and still got more than enough RAM to start anything

>>105901740
MPV WON
THANK YOU MPV

>>105901786
mpv lost

>>105901688
whoa, thanks! that fixed it!

https://code.videolan.org/videolan/libplacebo/-/merge_requests/719
remember to rewatch all anime you watched on gpu-next

>>105901665
Why?

>>105902078
AIIIIIEEEEEE KASPER SAMA GOT NANAHI BACKED INTO A CORNER!!!! NOW SHE MUST USE LIBPLACEBO TOO!!!!! NOOO DON'T LEAVE US GPUBROS :'(

>>105902078
ok now fix the hwdec freezing, and slow start up time on vulkan pls

>>105902380
Slow start up time is amd driver bug I think. Nothing mpv can do

>>105899825 (OP)
>mpv.io/installation/
>ffmpeg.org/download.html
>github.com/yt-dlp/yt-dlp
sudo pacman -Sy mpv ffmpeg yt-dlp

>>105902494
> Nothing mpv can do
Default to OpenGL. And gpu-next doesn’t work well with it so also default to gpu

>>105902494
nvidia too
https://github.com/mpv-player/mpv/issues/13019

How can I use ffmpeg without it raping my cpu? Can't find a way to slow or throttle it.

>>105902925
That's a good thing

>>105902925
you don't

>>105902078
this got btfo'd already by jetcord, the behavior still doesn't replicate vo=gpu and had quality problems during testing

vo_gpu is undefeated

>>105904591
this? This was a fix for overlays 65k vertex limit. I think you are thinking about something else.

>>105902494
>>105902744
i have the feeling that most complainers about this are retards not caching shader compiles.

>>105902925
"kill" the ffmpeg process with SIGSTOP and SIGCONT in a loop with two sleeps.
or use "cpulimit" which does the same thing under the hood.
or use cgroups if you want to get fancy.
(assuming non-tard OS)

>>105904931
nvidia probes for opengl before initing vulkan applications because the vulkan stack is built on opengl stack. take your schizophrenia somewhere else

>>105902511
loading yt videos on mpv is slow as fuck

>>105905129
i have no startup time issues in both my workstation (AMD), and laptop (Nvidia).
been using gpu-next+vulkan for years.
i'm not a wintard if that's relevant.

>>105905307
inb4 you-know-who

>>105905307
yt-dlp is slow

>>105905381
Who?

>>105905351
do you shove things up your butt as well? no one cares, stop blogging

>>105902122
less niche syntax to remember, consistent parameter meanings, no extraneous lossy colorspace round trips, and most importantly everything actually runs on the GPU.

lads, is mpv-android good, or should i stick to vlc on my tablet?

>>105905924
On Android you should use whatever player has working hardware decoding so you don't hemorrhage battery which on many phones is only ExoPlayer.

>>105905955
its a tablet, samsung s9 if it helps

Finally a good thread but mpchc is still king

>>105902925
just restrict it to running on 1 thread

>>105905980
mpv is just mpc-hc2

>>105902078
basedsper dabbin

>>105902172
Nanahi loves big black dicks.

>>105902172
xhe can't use libplacebo because lappy is too old

Is there a way to fetch the highest quality format of an age-restricted youtube VR180 video?
I managed to grab the non-VR format with the default player_client but only android_vr can get me the VR formats, and youtube's new shitty PO token system prevents me from grabbing it using the android_vr client.

profile=high-quality

hwdec=auto
dither-depth=10

alang=jpn,eng
slang=eng

video-sync=display-resample
interpolation=yes
blend-subtitles=yes

osd-bar=no
osd-on-seek=msg

demuxer-mkv-subtitle-preroll=yes
autocreate-playlist=filter


This is all you need.

>>105906420
It is unclear why you've decided to minimize your needs, and it is doubtful that you are an advanced user.

>>105906561
There's nothing else you need. If you have more, then your config is probably a mess with bad options. Having more options does not make you an advanced user.

>>105906420
>dither-depth=10
Sir! Delete it! Sir!

How do we fix the watch_later system so it's not shit?

>>105906420
>interpolation=no

>>105908729
interpolation doesn't do what you think it does

with vsync aware display modes, having frame-blending for uneven pulldown and/or frame repeats is high quality.

>>105909020
What do you mean? There is no frame blending without interpolation=yes

>>105909047
yes, which is why the interpolation option stays on, otherwise you can't smooth out small inconsistencies between repeat frames due to uneven vsync ratios (yes, that even includes watching 24/1.001 content on a 120.000Hz display, you won't get even 5:5 pulldown)

>>105908578
for one, it should be called watch-later to be consistent with the rest of the directory naming of mpv (e.g. script-opts)

>>105907844
>not having a 10-bit display in 2025
sirs...redeem the shitty 6bpc+2frc TN panel sirs

>>105909554
>(yes, that even includes watching 24/1.001 content on a 120.000Hz display, you won't get even 5:5 pulldown)
i shouldve clarified that this is only the case with the default audio mode. display-resample should hand the uneven vsync ratio in this case, but technically this is still applicable to something like display-vdrop, where frame-blending will make the presence of uneven frame repeats less noticeable without having to resample the video speed

>set a hotkey for cycle-values screen
>the window does not change screens

There are only like 3-4 valid ways to handle refresh rate related stuff in mpv, if you disagree you're a massive retard and don't reply to me.

>method #1: VRR (G-Sync/FreeSync) capable display + --video-sync=audio

>method #2: --video-sync=display-resample --interpolation=yes

>method #3: --video-sync=display-vdrop --interpolation=yes

>method #4: absurdly high refresh-rate screen + --video-sync=audio
We're talking 240/360/480Hz displays here. at this point, the frame-timings are so low that dropped v-syncs are not humanly perceivable (as low as <3ms!!). Literally do not bother with anything except the defaults and fixed v-sync blocked presentation.

If you use anything else, you're NGMI and you should reconsider your config.

>>105909979
>--retardation=(You) --refutation=no

>>105909969
This is all downstream from the fact that 24fps was a mistake that people refuse to let go for no good reason whatsoever, instead of just using 30fps. Even fucking 20fps would be better since it at least lines up evenly with a 60hz refresh rate. 24fps doesn't line up evenly with virtually any display in existence, not even PAL with their 50hz retardation. There's no excuse.

>>105910030
It looks bad. Just don't use it.

>>105910046
>retard who sees the word "interpolation" and thinks it means SVP because he doesn't know what it is

>>105910062
No I know what is. I think temporal interpolation looks bad. Just set your display to an integer multiple.

>>105910095
The default interpolation kernel is oversample, which is just anti-aliased nearest neighbor. If you set your display to an integer multiple then there will be no interpolated frames in the first place. The point of having interpolation enabled is for every other scenario. If you're paranoid because you don't understand what oversample is, then put it in an auto-profile.

>>105910129
>The default interpolation kernel is oversample
I know and I think it looks bad. Why are you having a melty.

>If you set your display to an integer multiple then there will be no interpolated frames in the first place
false

>>105910152
>I know and I think it looks bad. Why are you having a melty.
There is no melty. Why are you schizophrenic?
>false
If you knew what nearest neighbor was then you would know that your statement is profoundly retarded. Oversample at integer values is literally identical to nearest neighbor (no blurring/blending). That's the point of the filter...

>>105910169
I was wrong about the integer multiple thing but I still think oversample looks bad on the 24->60 case.

>>105910032
At least PAL is actually 50Hz and not some frame dropping abomination.
Also the whole reason film is stuck at 24 fps is the idiots that bitch and moan about how it "feels wrong" because motion isn't a blurry mess, not because of a lack of willingness from the studios.

>>105910208
Such "people" are objectively wrong and I doubt they will notice a difference between 24 and 30 at all. I'm not even asking for them to raise the standard to 60 (because I'm not that optimistic), I just want them to pick a framerate that actually syncs up with the displays that the content is displayed on instead of leaving us trapped in interpolation hell.

>>105910203
Frame-blending is preferable to frame discontinuities caused by judder (think of it as a form of temporal aliasing) because it allows you to preserve smooth panning motions. Oversample still preserves all of the frames and there is no additional blur added to the original frames because it behaves like nearest neighbor, it simply anti-aliases in-between discontinuities by adding a blended frame. Required reading: https://github.com/mpv-player/mpv/wiki/Interpolation#smoothmotion

>>105910267
I prefer judder over the awkwardness of the interpolation panning. It looks too smooth and wrong.

>>105910284
>I prefer judder
>It looks too smooth
10/10 troll, you had me for a good 30 minutes, so good job.

>>105910208
48hz hobbit looked like shit. Just accept it. 24 is perfect for film and you don't need more.

>>105910267
>thinks aliasing means anti-aliasing
opinion instantly discarded go back to /v/

>>105910336
Movies should be 30 to appease retards, everything else should be 60 at minimum and sports should be 120. It's insane that we have screens pushing upwards of 320hz but everyone shoots at a value that's less than ~7-8% of that.

>>105910361
??? oversample is literally nearest neighbor but with blurring around edges to preserve sample shape. It's literally anti-aliasing no matter how much you want to piss and moan about semantics. It's not my fault that you hear the world "anti-aliasing" and just think about video games because that's the only thing you associate the word with. You won't see oversample-esque filtering on anything but pixel art games anyways.

>>105910400
Yeah, it is anti aliasing. So why did you call it aliasing? /v/tard.

>>105910432
I didn't jeet-kun. Stop speedreading, it's not as if you have anything better to do with your time.

>>105899825 (OP)
Thats great you keep spamming mpv shit here all the time. Can you explain what it is cause otherwise its just seen as some homosexual software you like the icon of.

>>105911145
click the links

>https://github.com/mpv-player/mpv/commit/86e884bc4a84adcc0dab896b7fa8ac9678520d3f
Please revert, no more improvements should be made to the legacy renderer as it is no longer the default.

>>105911184
all she has to do is implement a quick hack to get dovi p5 to work and there will be no reason to use gpu-next

i will not be using gpu-next until it fixes its presentation issues. im getting tired of watching muh animes and having the frames stutter and inexplicably travel back in time at random. this is frankly embarrassing

post ffmpeg commands

>>105911464
ffmpeg -i in -lavfi showspectrumpic=s=hd1080 out.jpg

>>105911184
Some other small tweaks that vo_gpu needs to be more feature complete OOTB:

>target-prim currently makes no assumptions about the user display even though other target options do, make it assume bt.709 on SDR content to be consistent
>stat screen still reports "Fornat" as the fbo-format, it doesn't properly show output format

>>105910236
> I just want them to pick a framerate that actually syncs up with the displays that the content is displayed on instead of leaving us trapped in interpolation hell.
That wouldn't necessarily fix judder or whatever you're complaining about, especially on low transition time displays, it is a fundamentally very different viewing experience to film. The only real options for dealing with 24/30 fps content on digital displays are interpolation or BFI. Anyway the 24/60 problem is solved by just changing the display refresh rate (as MPC can do OOTB and there are MPV scripts for this).

mpv now detects my monitor as Display P3 capable when it defaulted to BT.709 before, HDR is off. My uneducated guess is the HDR ready signal from the monitor gets picked up by mpv and the logic behind this is to make the screen switch to HDR mode automatically. Can I just --traget-colorspace-hint=no or can this fuck tonemapping HDR to SDR in any way? I don't have HDR footage to test.

>>105909969
you missed

--video-sync=display-resample --interpolation=no
+ autospeedwin.lua (script to automatically change refresh rate to whatever matches video fps)
+ CRU (utility to add custom resolutions, or in this case, refresh rates, to a monitor)

its the best method for those whose panel doesn't have VRR (or, more commonly, does, but it doesn't work well)

>>105910032
>24fps doesn't line up evenly with virtually any display in existence
lol?
120/144/240/360hz all line up with 24p

Thanks, it's a nice change from VLC looks simple and ive already figured out righ-clicking audio/subtitle lists the tracks so theres no sub menus like vlc had. Always hated the traffic cone logo.

>>105912436
There's no point in disabling interpolation. There is no frame-blending when the display refresh rate is an integer, because the filter will just behave like nearest-neighbor. Like I said to the other anon who was trolling me, if you're paranoid, then make an auto-profile to cover all possible fallbacks.

>its the best method for those whose panel doesn't have VRR
VRR is not the end-all-be-all, it's simply one of the solutions a person could use for getting smooth panning motions. If your display is sufficiently high refresh rate, VRR is completely unnecessary with --video-sync=audio because dropped v-syncs will be imperceptible. If anything, it could be more harmful because VRR is empirically more jittery than fixed v-sync intervals, and some high refresh rate monitors can have funky gamma curve behavior at lower refresh rates. VRR is only useful in a few scenarios over display- modes.

>1. you have a high refresh rate monitor, but it's on the lower-end of the HRF monitor scale where dropped v-syncs are still noticeable during panning (120Hz, 144Hz, etc.)
>2. you have a high refresh rate monitor, but your potato can't handle --display- modes
>3. you really like VFR videos for some reason and it's the only content you consume
>4. power efficiency

Outside of that, --display- modes should always be preferred over VRR when possible. The order of the list makes it seem like VRR is at the top of the list, but that was my mistake.

>>105912532
>There's no point in disabling interpolation
in the scenario i described, there is. if interpolation is on, the video gets interpolated before autospeedwin has a chance to change the refresh rate, so it doesn't get changed at all.
but if it was the other way around, then yes, you wouldn't need to specify interpolation=no.

>If your display is sufficiently high refresh rate, VRR is completely unnecessary with --video-sync=audio because dropped v-syncs will be imperceptible
ime on a typical 144-160hz panel, its still noticeable.
haven't ever used 240 or 360, maybe its fine there.

>>105912571
>if interpolation is on, the video gets interpolated before autospeedwin
I don't use mpv scripts for modeline switching, so I honestly don't know.

>ime on a typical 144-160hz panel, its still noticeable.
I know, that's why I said:
>1. you have a high refresh rate monitor, but it's on the lower-end of the HRF monitor scale where dropped v-syncs are still noticeable during panning (120Hz, 144Hz, etc.)
But regardless, if you have a 120Hz/144Hz monitor you could easily setup a 120/1.001 modeline as an easy catch-all for 90% of content, in which case you should still use --display- modes over VRR for all of the aforementioned disadvantages of VRR.

>haven't ever used 240 or 360, maybe its fine there.
The duration of a single v-sync on a 240Hz monitor is only ~4.16ms, mean while on a 360Hz display it's ~2.77ms. You would have to have superhuman eyesight to notice something at that point, especially at sub 3ms. You're more likely to notice whatever presentation jitter is present with VRR.

>>105900579
I never understood that. it obviously isn't about security, because audio is allowed on some boards. I don't know why they won't just make it muted by default and allow everywhere.

>>105912641
I used to passively agree but since 2018 I don't want every board spammed with tiktoks which is what would happen.

>>105911448
>i will not be using gpu-next until it fixes its presentation issues. im getting tired of watching muh animes and having the frames stutter and inexplicably travel back in time at random. this is frankly embarrassing
Vulkan? Never seen that outside of Vulkan and it happens when the gpu is not keeping up with rendering.

>>105912411
Show stats. I cannot guess based on this description, don't know what platform even. Display P3 is valid for SDR if you compositor prefers that.

>>105908729
Why?
I have it on and it works perfectly fine.

>>105909969
anyone knows what happens with --video-sync=display-resample + apple pro motion display?

>>105911448
>inexplicably travel back in time at random
had this happening with --vd-queue-enable=yes, idk if it also happens on vo=gpu when this option is used

>>105913552
it drops frames

>>105913360
I had just switched from the prebuilt binary of my distro to master and after posting I noticed the default value of target-colorspace-hint is now auto. It's Arch with Plasma/Wayland. The color change is pronounced here but no screenshot makes it justice.

>>105914928
Compositor is telling mpv this is the preferred output colorspace now. If that's not correct for your display, check options or report to KDE. You can use `--target-colorspace-hint-mode=source` to use source colorspace and send to compositor this data as is.

profile=high-quality
Rate my config.

Why should I mpv instyead of MPC?
Does it play DVDs? Can it play Blu rays?

>>105901300
>Reminder that libplacebo is awful.
qrd?

>>105913345
not vulkan, gpu usage is not high. if anything, vulkan might not be affected. the issue is related to the d3d11 renderer
>>105913603
idk but it doesn't happen with vo=gpu.

not gonna bug test, just want to be comfy and use what works

>>105909912
No but seriously how do you do this?

>>105915674
Outdated, new endgame is:
vo=gpu
gpu-api=opengl

>>105909912
>>105916311
The screen option doesn't really work like that but I guess it wouldn't be impossible to change it. It would require a different implementation on every platform (and of course wouldn't work in wayland but whatever). In the meantime, if you're creative with the --geometry option it should have the same behavior.

>>105916466
Have you gotten geometry options to change on the fly? I thought you could only use window-scale if the player's already running.

>>105916524
Runtime geometry changes should be working on all the platforms that support it. If not, that's a bug.

>vo: move vo_gpu_next above vo_gpu in probe order
why are we putting unnecessary maintenance burden on haasn? let the man spend time with his girlfriend

Is it possible to play SACD isos with mpv? I know ripped dsf files work but I couldn't get it to work with an iso file and sacd-ripper doesn't want to compile on my gentoo installation without some tardwrangling it seems.

>>105916659
It would require libcdio-paranoia to support SACD and maybe some mpv code changes, but I don't believe that library can read SACD.

>>105916659
>mpv
>SACD playback

>>105916534
Thanks, last time I tried it I was on an older build I think. Also just realized the fs-screen value gives me pretty close to what I want, although the window doesn't go back to the same screen when you exit fullscreen.

>>105916644
he had many years

>>105916685
This Anon again, used geometry and now it's fixed :^)

>>105917023
but has he put a baby in her yet?

>>105914928
>Format: rgba16
why can't we have this windowsbros?

>>105916316
How so?

>>105915694
only if you're on linux.

mpc + madvr still has the best framepacing and hdr mapping. only issue i have with mpc is that it doesn't handle PGS subtitles that well, while mpv can size and move them. there are ways to convert pgs to srt though, so easy to mitigate.

>>105918183
>there are ways to convert pgs to srt though
Yeah, using tesseract and it's completely useless without a gigantic corpus of autocorrects, which the scene is not willing to share. You get significantly more accurate results by using whisper.cpp but afaik nobody has made a mpv script for it and most people's rigs couldn't run it in real time anyway.

>>105918374
>whisper.cpp but afaik nobody has made a mpv script for it
https://github.com/GhostNaN/whisper-subs

where's the plasma faggot

>>105918374
>most people's rigs couldn't run it in real time anyway.
My 5 watt 5 year old iGPU consistently runs over 4x faster than realtime you'd need a very very bad rig to not be able to use it. The memory requirement isn't too bad either but you need a couple gigs free to not emulate the jank of old TV CCs. Handles long periods of silence much better than older automated methods too. The only downside is of course you can't get any SDH information but maybe in the future we can have a vision model running to supply that too :3

>>105918691
>My 5 watt 5 year old iGPU consistently runs over 4x faster than realtime
Is that with or without running mpv at the same time fucktard

>>105918183
>mpc
klite codec is better

>>105918836
>klite codec
what year is this

>>105918881
It's 2025? KLCP is the only MPC codec pack that still gets updates. You might be getting it confused with KCP. I really wish mediainfo-lite would get merged into the upstream. Any upstream.

>>105918934
Doesn't mpc come with lav filters? isn't that all you need?

>>105900296
this doesn't work for me. how do i fix?

>>105919216
terminal:
systemctl disable dbus-broker

>>105919054
No. Like I just said, I need mediainfo-lite. Also madVR.

#!/usr/bin/env sh

# encodes video with burned hardsubs to Sony PSP format with matching thumbnail
[ -f "$1" ] || { echo "usage: ${0##*/} [INPUT FILE]"; exit 1; }

file="$1"
output="${file##*/}"
output="${output%%.*}-h264"

# sony psp format
# H.264/MPEG-4 AVC Main Profile (CABAC) (AAC) and Baseline Profile (AAC)
# 480x272 resolution
# pad only applies to UHD content
# for 4:3 content, do not use 480:-1, set to VCD resolution -1:240 and no padding
ffmpeg -threads 0 -y -i "$file" \
-map 0:0 -c:v libx264 -crf 18 -profile:v main -preset slow \
-pix_fmt yuv420p -b-pyramid none -level 3 -refs 3 \
-filter_complex \
"[0:0]subtitles='$file':si=0[a];[a]scale=480:-1:force_original_aspect_ratio=1[a];[a]pad=480:272:(ow-iw)/2:(oh-ih)/2" \
-map 0:1 -c:a aac -ab 96k -ac 2 -ar 24000 \
"$output.mp4"

# generate matching thumbnail
ffmpeg -y -i "$file" -f image2 -ss 20 -vframes 1 -s 160x120 "$output.thm"

>>105918183
>mpc + madvr still has the best framepacing
use vo=gpu, the framepacing problems are currently a known issue in gpu-next, specifically when using video-sync=audio, which makes it seem like madvr is "better"
>hdr mapping.
not needed

>>105918183
>>105919274
dolby vision lol lol lol lol lol lol lol

>>105918183
>>105919710
Seems like there's a lot of whining about display-sync here... Why not just use madVR? madVR is currently the only video renderer with competent display synchronization.

Here's how you enable it...
rendering > general settings > present a frame for every vsync

Yes! It's that simple. Just remember to set a reasonable backbuffer queue depending on your hardware, and you're good to go.

Here's how you're better than the average mpvtard if you switch to madVR.

>madVR's vsync does not rely on jittery API timings, and is aware of your current vblank frequency to give you the best presentation possible
"B-But muh display-resamp--"
Not so fast, mpvtard.
>madVR's vsync does not speed up or slow down your video files, ruining the creator's intended vision
>madVR's vsync does not resample the audio, meaning that you will avoid random audio distortions and preserve bit-exact passthrough
>madVR's vsync does not allow the A/V clock deviation to exceed humanly noticeable thresholds
>madVR's vsync will calculate the rate of repeat frames, and will log if there's a presentation glitch.

Why use an inferior piece of software? Every film enthusiast deserves excellence because every pixel counts and every frame counts. And excellence is only achievable through madVR.

>>105919710
tried the vo=gpu, still not as smooth as madvr.
bladerunner 2049, first few shots of the solar farms are my benchmark.

>>105919817
Meeting all those requirements at the same time sounds like playing the video normally with vsync enabled on your operating system. how the fuck could there possibly be any difference if you're not changing speed or resampling?

>>105919849
>playing the video normally with vsync enabled on your operating system
This is not what "present a frame for every vsync" means. What you're talking about is dwm.exe enforcing VSync on applications and using DXGI API to handle its own redraws/repaints, which is completely different from what madVR does by default.

>>105919922
that doesn't answer a single part of my question. also i'm pretty sure what you're describing is strictly worse than adaptive sync which just werks in mpv if you use d3d11-exclusive-fs=yes

>>105919951
>adaptive sync which just werks in mpv
Unless something has changed in the 3 years since I last looked into this, this only works on Nvidia cards. For Intel and AMD you need a schizo config to duplicate frames to get into the adaptive sync range and even then NTSC framerates were actually presenting as integer framerates.

>>105919951
>that doesn't answer a single part of my question
I thought mpvtards were supposed to be super smart hackers? How could you be so dumb?

>also i'm pretty sure what you're describing is strictly worse than adaptive sync
Does he know?

madshi on adaptive sync...

>Games create a virtual world in which the player moves around, and for best playing experience, we want to achieve a very high frame rate and lowest possible latency, without any tearing. As a result with FreeSync/G-SYNC the game simply renders as fast as it can and then throws each rendered frame to the display immediately. This results in very smooth motion, low latency and a very good playability.
>Video rendering has completely different requirements. Video was recorded at a very specific frame interval, e.g. 23.976 frames per second. When doing video playback, unlike games, we don't actually render a virtual 3D world. Instead we just send the recorded video frames to the display. Because we cannot actually re-render the video frames in a different 3D world view position, it doesn't make sense to send frames to the display as fast as we can render. The movie would play like fast forward, if we did that! For perfect motion smoothness, we want the display to show each video frame for *EXACTLY* the right amount of time, which is usually 1000 / 24.000 * 1.001 = 41.70833 milliseconds.
>FreeSync/G-SYNC would help with video rendering only if they had an API which allowed madVR to specify which video frame should be displayed for how long. But this is not what FreeSync/G-SYNC were made for, so such an API probably doesn't exist (I'm not 100% sure about that, though). Video renderers do not want a rendered frame to be displayed immediately. Instead they want the frames to be displayed at a specific point in time in the future, which is the opposite of what FreeSync/G-SYNC were made for.

>>105920026
madshi is retarded you can use high-precision timers to avoid swapping the flip model until you want to and that's exactly what mpv does. again, dolby vision lol eat shit it's not 2010.

I'd use madvr on loonix if I could. mpv has a billion knobs and none of them just werk.

Tell me mpv fags, what's the closest I can get to madvr frame presentation. No audio speedup nonsense. Display-vdrop?

>>105920054
If you're using timers then you're syncing to system clock? That seems even worse than using the audio clock.

>>105920026
>>105920054
if you need more cope...

Nevcairiel (ffmpeg contributor, maintainer of lavfilters, d3d11 maintainer for **libplacebo**) on the topic of adaptive sync:

>No, video content is quite distinctly different to content actually created/rendered on-the-fly (like a game).

>Videos are all pre-rendered and pre-timed. To properly present them on such a "variable sync" system, you would for one need a super precise clock, since you have no Vsync that you can use for this purpose, and hope that the GPU and display actually present them without a varying delay (which a game could easily and automatically compensate for).

>Additionally, you cannot have any "drop" protection. madVR prepares and presents several frames in advance so that a small hiccup in the system wouldn't be noticed. If you cannot present in advance (because the GPU would display them immediately without a fixed sync) you lose this protection, and you may notice this.

>All in all, you lose quite a bunch of things that every video renderer leverages today to get smooth and stable playback.
Noone is saying that you couldn't build a video renderer specifically for this, we're only saying that no existing video renderer would be equipped to do this without a major design shift (and that its really not worth it when you can play videos at 100-144 Hz instead, plus this feature only being available in "gaming" screens)

Say what you want about godshi, call him a "retarded hack" all you want. But what about the person who contributed to the DXGI swapchain code for your precious libplacebo? Is he retarded too? Why are you using a retarded video renderer on your retarded player then?

Your hubris is amusing.

>putting quotes around something that wasn't said
LLM

KEK mpv trannies seething
You will never be a girl.

>call godshi retarded
>NOOOO i didnt say that! bot bot bot bot!! i have no retort please save me kasper!!
mpvtards will never not be hilarious. I'm gonna go enjoy my smooth frametimings with no presentation glitches now.

mpc + madvr can't do this

>>105920592
not sure i want it to...

>>105918183
>mpc
I hope you mean MPC-BE

>>105920592
Is that supposed to be something I want?

The ikacord is real.

>>105920749
mpc-be is shitware lmao

if you want to have a good laugh, read this

https://github.com/Aleksoid1978/VideoRenderer/pull/212

>>105920755
Literal trannyware

>>105920772
send invite

>>105920777
>having your patch rejected because the maintainer is a complete retard that doesn't understand any of the underlying concepts
that's rough. poor guy.

>>105920772
is this real

>>105920634
>>105920755
this isn't stackoverflow

>>105920772
>ika!
>what alice
>I have an idea!
>.... what's your idea alice
>LETS MAKE FUN THREADS ABOUT YOUTUBE
>alice theres a board for that
>NO NO WE WILL BE TEAMS AND WE CAN SHILL BOTNET ADD IKA AD THE BEGINN SO ITS LIKE THIS DISCORD CHANNEL
>I like that idea
>shut up Lilith... don't encourage her
>AUTOMOD: IT IS TIME FOR YOU 6PM HRT SHOTS

>>105920634
>>105920755
yes if you watch youtube

>>105920592
your software is really nice, i just wish there was a way to log onto my youtube account so it can view my playlists

>>105920772
Welp, it's official
Ikadev/nanahi is a tranny

https://github.com/mpv-player/mpv/issues/15513#issuecomment-3076421797

/g/ wins again

https://desuarchive.org/g/thread/99635062/#99640235

>>105921396
And they called it a meme. shows how little mpvfags know.

>>105921396
i remember this based shizo

>>105921449
It has nothing to do with VRR. The schizo was not even right.

>>105920777
>MPCVR
Am I supposed to care? This is also the default renderer on MPC-HC..
At least MPC-BE can handle loading a YouTube playlist without locking up for an hour.

Can anyone help a boomer?
>have 2100ish youtube links in my bookmarks from 2011 - 2025 because I've refused to log into (((youtube))) ever for any reason
>"wait this is stupid"
>yt-dlp.exe & ffmpeg.exe
>set to download all my 2100 links
>watch it do about 20, it works, leave to work
>come back, 1000 finished, missing meta data/title so only 200 are usable
Now I can't view anything on jewtube without getting a "Sign in to prove you're not a bot". Will this ever go away? I have a static IP from AT&T fiber....

>>105921461
>It has nothing to do with VRR.
it's related to frame presentation with audio-mode. just because someone finds a root problem and claim it only solves x doesn't mean that it isn't applicable to yz scenario. even julestard says:
>I can reproduce on my 6700XT with default display-sync but not display-resample.
more than likely the anon didn't bother investigating further and just accepted that it fit his usecase

either way, every mpv maintainer has to apologize to /g/ as usual

>>105921468
>Am I supposed to care?
yes, because those retards maintain mpc-be. clsid2 and the pr author are the only people there who know what they're talking about.

>>105921507
The github issue is about presentation glitches that occur under extremely heavy load likely due to the vulkan swapchain not selecting the "best" image when you're in a scenario where a bunch of frames have to be dropped. Normal VRR usage will never run into that scenario.

>>105921468
>>105921521
It's fine to not know something but stubbornly insisting that you're correct when you have absolutely no idea what you're talking about is the bad part.

>>105921550
except this isn't the whole story because even in fixed refresh rate presentation, vsync jitter on vulkan is considerably lower with swapchain-depth=1 or swapchain-depth=2. this has been corroborated multiple times on /g/ as well. there's clearly a different issue here, and high load simply makes it more visible. its not a coincidence that lowering swapchain-depth supposedly fixes so many issues

>>105921580
>vsync jitter on vulkan is considerably lower with swapchain-depth=1 or swapchain-depth=2
first I've ever heard of this

>>105921567
>stubbornly insisting that you're correct when you have absolutely no idea what you're talking about is the bad part.
wow, they're just like /mpv/! im switching to mpc-be now

>>105921604
yeah but we're not the ones with push rights

>>105921588
And I can't reproduce. It's either the same (0 jitter) or swapchain-depth=1 is worse.

https://github.com/mpv-player/mpv/issues/14811#issuecomment-2335842154
>Note that I initially had the option swapchain-depth=1 option set, as this is needed for Nvidia gpus and recent drivers to get a no-stuttery output. I did a second run without the option, behavior is still the same... frame drops*
stutterpv is real, open your eyes

>>105921654
another victim of stutterpv

https://github.com/mpv-player/mpv/issues/6172#issuecomment-963224575
>The issue is still unchanged with --vo=gpu-next and latest libplacebo git-master, which unfortuantely makes --vo=gpu-next unusable on Nvidia Xorg (life without --video-sync=display-resample is possible, but meaningless).
https://github.com/mpv-player/mpv/issues/6172#issuecomment-1260294283
>Turns out this happens only with swapchain-depth= >2. CPU load with a value of 1 and 2 is normal, and it seems also playback looks fine with it (both composited Xorg windowed and uncomposited fullscreen).

>>105921676
>nvidia on linux of all things
might as well just chop your balls off at that point

>I can write the script so you don't need to make a feature request.
What did our dear friend mean by this

Is there an Mpv crop plugin that works with shaders?

>>105922738
doesn't --video-crop work?

>>105923241
I'm using the crop.lua script but shaders don't do anything. You can tell because they don't show up in the OSD if you go to the frame timings bit.

In mpv I use “sub-seek -1 exact” all the time. There is a short delay when I do that. Is there any tweak to make it faster? I’m willing to sacrifice some ram for smoother seeking.

>>105923555
No

>>105921521
>clsid2
The same guy who thinks its fine to blockingly load URLs with yt-dlp? Also blocking whatever source you're DnDing the link from? Maybe he can handle VR but the UX from MPC-BE is much better.

>>105919399
What?

>>105924240
it's pretty self explanatory if you can read

>>105924288
No, you just posted a code about creating mp4 files for sony PSP without any intro.
fucking zoomer.
Why would anyone uses psp to watch video even?
I admit I did watch porn on it back in HS.

>>105923555
le --play-direction= but it may not work for all files i think

>clsid2 and the pr author are the only people there who know what they're talking about
lmk when they can come up with a player that doesn't oversaturate the output. Only SDR PNG of a DV file but demonstrates the problem.
This is default settings for all and regardless of MPCVR or madVR.

>>105909673
>>105909554
not the op but if my display is refreshing at 119.88Hz, that means I have perfect frame sync for ntsc content?
weirdly, in windows I have both 119.880hz and 120.000hz modes available, but in kde wayland it's only showing 199.88hz (pic related).
anyone know how i can get the 120hz refresh mode back under linux?

>>105925549
add it to your edid

>>105899825 (OP)
Choosing Best Video, and Best Audio compiles to .webm, and an audio codec that can't be played on Windows Movies & TV's vidya player. What am I doing wrong? What audio codec should I be using to make it play on windows and not have to use vlc? AV1 won't even play.

Are there any python scripts that can archive an entire youtube podcast channel, about 2hr long videos and saves thumbs and meta data like description? Anything with a delay? I don't care if I only have to download 10 videos a day or have a 10 min delay per video or what ever. Just need to avoid getting rate banned.

>>105926267
yt-dlp

>>105926240
Just install a codec pack. LAV filters is probably the most popular but I am not clear whether they will just work with Media Foundation based players like yours.
>not have to use vlc
Is the default Windows player really any better

>>105926240
>>105926367
>I am not clear whether they will just work with Media Foundation based players like yours
They do not
https://www.codecguide.com/media_foundation_codecs.htm
Just run through that link and install the store packages for your player.
https://apps.microsoft.com/detail/9mvzqvxjbq9v

/g/ how to fix mpv? What would you change?

>>105926751
speed change doesn't drop video frames

>>105926751
custom color options apply before first frame is displayed

>>105918183
The learning curve of madVR is massive. But I try.

>>105921396
Fuck I have 2 as that got me slightly lower jitter when I last checked. Is 1 better?
>>105921588
>>105921580
>or swapchain-depth=2. this has been corroborated multiple times on /g/ as well.
It's what I saw with side by side testing, but I can't claim my testing was scientific. Was a cargocult last year.

>>105899825 (OP)
Maybe I can ask here and I don't need to open a thread.
could you suggest me a video edition software? I need to upload some videos. I'm talking about proffesional stuff like: copy, paste, rotate, save as... and other video hacker tools.

>>105927186
good morning sir

>>105927186
ffmpeg!

>>105927212
not a pajeet, but it seems I can do it with vlc, I'm looking into it now

>>105925335
Last time I checked, fire and sparks are red and orange. Not salmon coloured. mpv is not reflecting reality correctly, this is a self-own my guy.

>>105925335
Is that vo=gpu or vo=gpu-next because I still don't remotely trust gpu-next for any HDR content, neither do I remotely trust madVR for DV

a short poem in celebration of gpu-next being the default *ahem*

roses are orange
violets are teal
i use libplacebo
and now so do you!

>>105930264
The white is teal meme is dead. Kasper fixed it months ago. We must find new material tbdesu

>>105926367
yt-dlp doesn't do bulk downloading by default with delay paramaters does it? Has anyone made a script that takes all of these things into account with options for json stuff, thumbs, saving as a certain string, etc? Something that can be edited, stopped, and restarted? It's about 800 2hr videos so I will be switching filepaths, hard drives, and stopping and restarting.

>>105930329
>we
too obvious, go back to #mpv-devel

>>105926751
>pause video
>press m to mute --volume
>unpause
>audio still plays after unpause

>>105926951
the learning "curve" for madVR is massive because it's purposefully obtuse and doesn't work OOTB. hell, even after tweaking it, it still doesn't work.

>>105930031
The HDR image clearly won't look like that SDR cap. Also its not fire and sparks..

>>105930051
>Is that vo=gpu or vo=gpu-next
Config file only has osc=no but MPV identifies it as gpu-next with d3d11.
>neither do I remotely trust madVR for DV
MadVR and MPCVR give the same colors here but the same configuration for both with MPC-BE matches the output from MPV.

It also doesn't appear in a portable MPC-HC from GitHub so it appears to be a configuration problem on my end. Though even the portable MPC-HC with MadVR doesn't work right as it will apparently switch to SDR on fullscreen..
Whatever, is there any reason to use MadVR anymore for HDR content? Apart from its smooth motion setting.

The fuck am I doing wrong?
ffmpeg -i /media/kyle/NOTHILL/VIDEO_TS/"concat:VTS_03_0.VOB|VTS_03_1.VOB|VTS_03_2.VOB|VTS_03_3.VOB|VTS_03_4.VOB|VTS_03_5.VOB|VTS_03_6.VOB" -to 00:03:00 -c:v libx264 -crf 16 -c:a aac -b:a 256K -sn -map_metadata -1 -f mp4 "Notting_Hill (1999).mp4"

>>105910267
>https://github.com/mpv-player/mpv/wiki/Interpolation#smoothmotion
>Convolution-based interpolation
>you see its all very technical, it does signal processing complx stuff and essentially does convolution
Lmao easily the worst description I've read. Though DSP is usually filled with midwits online spewing shit and misunderstanding the simplifications.

>>105931894
LLM reply

>>105931944
we need nanahi's opinion on this.

>still can't change yt format at runtime on any player except for MPV
Guess it wins by default

>>105932037
vlc can do it to if you play the full manifest url
it also has real adaptive stream selection unlike mpv

>>105932188
>it also has real adaptive stream selection unlike mpv
would you explain what is this and its benefit?

>>105930264
"we are too retarded"
"to use them `--image-lut`s"
"we are fully committed"
"to never edit these CLUTs"
you will never satisfy your eyes
or ever have monthly clots

>>105931829
this is invalid syntax (LLM!)!
cd to the directory or pass full paths to all files.

>>105933478
auto adapt stream bitrate to network condition

>>105931685
>>105926951
not sure what's difficult?
you install it, select it as video renderer, madvr systray icon gives you settings. go to rendering 'folder', smooth motion, toggle 'enable smooth motion' and you're set.

>>105931829
not using MakeMKV is what you're doing wrong

What are these errors?
libEGL warning: failed to open /dev/dri/card1: Permission denied

libEGL warning: DRI2: could not open /dev/dri/card1 (Permission denied)
MESA: error: ZINK: vkEnumeratePhysicalDevices failed (VK_ERROR_INITIALIZATION_FAILED)
MESA: error: ZINK: failed to choose pdev
libEGL warning: egl: failed to create dri2 screen

>>105935487
>Permission denied

>>105935511
I need sudo to run mpv now?

>>105935524
if you're not using a full systrmd/udev/<other shit>, make sure you're a member of the "video" group.
otherwise, figure it out on your own.
you can start by "ls -l /dev/dri/card1" to see the permissions. it's not that hard.

>>105935524
you need to add your user to the group that owns card1

why do trannies shill mpv troonslop so hard? pic related is simple and just werks for everything.

>>105935959
Agreed. MPV should drop Windows support because better options exist for that platform.

I love mpv!!!!

>>105921654
should add big disclaimer on their manual --video-sync=audio stutters heavily!!!

Ikatube-anon, is there a way to setup a custom command to launch when pressing the play button instead of autodetecting MPV? I'm writing a custom TU that i would like to launch instead.

I use a lua script to drop my screen's framerate to match the video's along with --video-sync=audio but this method has issues with variable framerate content. I was wondering if it was possible to make a conditional profile for such videos where the script is ignored and video-sync=display resample is used instead but I have no idea if it's possible to detect fps variability in such a way that all variable framerate videos are covered.

>>105937835
display modes do not work with variable refresh-rate content, what you're asking for is untenable. if you already have a script that drops your monitor's refresh rate, you should be using display-resample at all times. read the manual

>>105899825 (OP)
I have taken the mpvpill.

>>105937556
manual already has a normative statement that recommends display-resample if it works for your setup. what more do you want?

>If you use this option, you usually want to set it to display-resample to enable a timing mode that tries to not skip or repeat frames when for example playing 24fps video on a 24Hz screen.

>>105899931
More minimalist and part of libav itself. I stuck with ffplay for years until they deprecated it.

>>105900101
mpv works on stripchat's CDN.
Combine this with --vo=tct and you can have infinite female buttholes directly in your tmux session.

>>105935569
Unless the hotplug deamon is misconfigured.
I wonder if he's using devtmpfs so the kernel just sets the owner to root or whatever.

if you watch anime, add this to your config for accurate lineart scaling

scale=ewa_lanczos4sharpest
scale-antiring=1
dscale=bilinear
correct-downscaling=no

thank me later

>>105939568
How is this better than mine?
vo=gpu-next
hwdec=vaapi
gpu-api=opengl
dscale=catmull_rom
linear-downscaling=no
video-sync=display-resample
interpolation=yes
tscale=oversample

>>105939858
I feel like turning on hardware decoding would make my laptop blow up.

>>105939883
>hardware decoding would make my laptop blow up
Are you baiting?

>>105939898
The battery is already swelling and pushing the case apart. I don't think it would take a lot. It always gets so hot when I use the GPU for anything. I don't know why.

>>105939923
What?
Dude, using HW decoding actually lower temps.

>>105939443
I have 30-24 fps variable .mkv files and frames are def being interpolated because I set tscale=bicubic and 24 fps panning shots look blurry as hell compared to when I just leave it out. What's going on?

>if you already have a script that drops your monitor's refresh rate, you should be using display-resample at all times. read the manual
Audio lags behind on 24 fps content if I don't use audio mode.

>>105939997
That hasn't been my experience.

>>105939858
>opengl
>gpu-next
sirs...

>>105940028
>Audio lags behind on 24 fps content if I don't use audio mode.
this is not possible unless your system is so slow that mpv's dynamic rate control can't apply the resampling filter fast enough

>>105940110
Then you have other problems friend.

>>105940219
I'm not your friend, pal.

>>105940185
Because vulcan is shit.

>>105937663
you can set the player_path in the config to the custom program you write, then it will run that program with the same commandline parameters used for mpv
as long as your program can parse these parameters then it will work

>>105900268
just drag the link into the mpv window bro its not hard

>>105939997
It can but if the video is low enough resolution it will actually take more power for all the overhead than you're saving (definitely anything below 720, even 720 itself on newer hardware). Also if the heat pipe has shitty contact with the GPU die he's fucked.

>>105899825 (OP)
honestly liking haruna a lot, feels like the best of mpv and mpc

>>105937835
Solved half of the problem, at least low framerate VFR videos are not a choppy mess now:
[variable-framerate]
profile-cond=math.abs(estimated_vf_fps - container_fps) > 1
profile-restore=copy
interpolation=yes
video-sync=display-resample
tscale=catmull_rom

I still don't know how to make mpv ignore the script when this profile is triggered, I guess it parses the scripts first then applies the auto profiles making that impossible

ikadev, does use_full_manifest always start with lowest quality? I have mpv built with subrandr, do you know why sub with named timedtext rendered with subrandr but the rest with libass even they are all webvtt? also sound hiccup when change sub, what is the reason? thanks

>>105941924
It's the same for lower res video as well.
>heat pipe has shitty contact with the GPU die
This is probably the cause.

>>105900976
what's troonware here? implied yt (agreed) ore general free software(and you are worng)

>>105944351
>ikadev, does use_full_manifest always start with lowest quality?
it should start with the highest quality by default, check --hls-bitrate
>I have mpv built with subrandr, do you know why sub with named timedtext rendered with subrandr but the rest with libass even they are all webvtt? also sound hiccup when change sub, what is the reason?
no idea about that, maybe the auto format detection is buggy

>>105945821
>it should start with the highest quality by default, check --hls-bitrate
okay, thats explain it. I actually have hls-bitrate=min in my conf. thanks. sorry for trouble.

why does using -vf setpts=2*PTS cause the output bitrate to be halved compared to not using the filter and how do I avoid this

>>105936250
>>105941115
>>105945821
why are you a pedo

>>105946553
Discord tranny coal >>105920772