yt-dlp for YouTube is fixed. Update to the latest stable versions of yt-dlp, yt-dlp-ejs, and deno.

>>108007307
Bun works too right? Rust projects are aids to compile

>>108007323
I don't think so? It didn't last I checked. But that was a couple months ago.

>>108007325
Holy based.

File: 1641886880102.jpg (89.5 KB)

89.5 KB JPG

dementia autism schizophrenia

>>108007298
>>108007325
>>108007397
kill yourself

>>108007325
madVR BVLL on the bottom right

File: f277f3659d8f5d878af335e417e4a540.jpg (2 MB)

2 MB JPG

>>108007307
>Update to the latest stable versions of yt-dlp, yt-dlp-ejs, and deno
all these js dependencies
not a problem with my own youtube client

File: Zminok-2026-01-29.png (19.4 KB)

19.4 KB PNG

The reason ikatrans-chino (or however the burger is flipped) won't open source the software is because xe wants to make the code 'pretty' before showing it.

>>108007307
It broke? didn't noticed and I've been binge watching YT with mpv for a couple days now

>>108007652
The reason the faggot won't open source it is because he literally stole some faggot's script from github that is nothing but a bash script for yt-dlp+mpv. Then he added a bunch of gay animations to it and spammed every alt-chan on the internet.

I wrote something more usable in half a day.

>>108007600
You fuck black men.

ikafaggot ruined this general

This commit is still causing a 66% FPS drop on Arch

8407df979f6d8bb29e15b9412e769639bc3763c0 is the first bad commit
commit 8407df979f6d8bb29e15b9412e769639bc3763c0
Author: Kacper Michajłow
Date: Sun Apr 27 00:05:28 2025 +0200

 sub/osd_libass: don't rerender OSD if not changed
 
 Rendering OSD on every frame draw is quite heavy, even if some of the
 rendering parts are cached, it still takes significant amount of time,
 depending of payload can take 8-10 ms just to do the text shaping.
 
 Instead render only when we need it.

 sub/osd_libass.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

>>108007740
Sorry, 34% drop. I'm running on fumes here with 4 hours of sleep

File: a4b9e44a2c3817e941ad112b9a00b450.jpg (166.2 KB)

166.2 KB JPG

>>108007695
>>108007705
never happened

File: 1756073107108607.jpg (337.1 KB)

337.1 KB JPG

>>107704140
reporting in that this is now fixed for me as of

libva-nvidia-driver 0.0.15-1.1

:)

>>108007307
>yt-dlp, yt-dlp-ejs, and deno
I only needed the updated yt-dlp and node from nodejs.org (manually unzipped, no need for package managers), now yt-dlp lists all the formats.

File: Screenshot from 2026-01-29 18-27-39.png (5.3 KB)

5.3 KB PNG

finally yt-dlp has been updated

File: 1742375465934034.gif (69.7 KB)

69.7 KB GIF

RSS status for ikachan update?

>>108008442
OW

>>108008442
You know this general is meant for media players right? Go the fuck back to your trannycord circlejerk and stay there.

>>108007740
That's weird. I don't possibly see how this commit could effect performance. Actually I don't even understand how it supposedly does what the commit message says. All this does is add an assert.
https://github.com/mpv-player/mpv/commit/8407df979f6d8bb29e15b9412e769639bc3763c0

>>108008704
Nevermind, I'm dumb and I see it now but in theory this should save performance...

>>108007735
Fr bro this Vietnamese gook spams his shit on all of 4chan

>>108007298
kys IIyyr, you will never be a woman

>>108007803
I'm going to make it happen :3 *pounce*

File: 1748083680372790.webm (545.1 KB)

545.1 KB WEBM

>>107975246
>https://0x0.st/PN-j.sh
Thanks dude.
I finally manged to get it to work.
I was having some issues with resolving the hostname since my ISP is doing something.
Even when switching to different DNS I still couldn't get it to work.
SO I just spun warpproxy container and used proxy chains.
How to get better at these? and parsing and finding the source and decrypting it?
>uses hls non-standard extension to append params to urls.
What do you mean by this?
thanks BTW

File: 9bbf03d026645a9984b1a8248796984f.png (983.7 KB)

983.7 KB PNG

>>108008442
soon
going to work on it in a few days

Mpv is giving me HTTP/403 errors again. What gives?
I have:

ytdl-raw-options=js-runtimes=deno

in my config. yt-dlp works fine.

>>108011770
Streamlink works for me with -4 flag.

YouTube seems to be blocking my IPv6 and adding force-ipv4= doesn't do anything. I know YouTube enforces source IP selection now but I have multiple IPv6 addresses. I think something must be broken somewhere.

>>108011770
I just keep yt-dlp and deno in the same folder and don't need anything in the config file

>>108011814
It works when I force IPv4 via a network namespace like:

sudo -E ip netns exec ipv4 sudo -E -u $USER -- env XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR WAYLAND_DISPLAY=$WAYLAND_DISPLAY  mpv -v ...

Is there any way to force the source address that ffmpeg uses?

>>108011831
I also tried with force-ipv6= which also doesn't work.

I wonder if I set a specific IPv6 route for the manifest.googlevideo.com and googlevideo server will that make it work?

It worked. I did:

ip -6 ro add  dev  src 

And no more issues. Fucking retards.

My default IPv6 route looks like this:

default metric 50 pref medium
        nexthop via fe80::xxxx:xxxx:xxxx:xxxx dev iface0 weight 1
        nexthop via fe80::xxxx:xxxx:xxxx:xxxx dev iface1 weight 1
        nexthop via fe80::xxxx:xxxx:xxxx:xxxx dev iface2 weight 1
        nexthop via fe80::xxxx:xxxx:xxxx:xxxx dev iface3 weight 1

I currently have 17 different IPv6 addresses:

$ ip -6 addr | grep global | grep -E 'inet6 2a' | wc -l
17

But Google expects the src IPv6 address to stay the same. Fucking kill all of your pajeets that implemented that right now. The src address could change at any moment. At least allow different src addresses in the same prefix.

>>108011922
Ideally Mpv would parse the /ip/2aXXXX/ and use that as the source address. Can it do that?

hmm 616 is back in the default format list

I just watch 397 and listen to 251 because I have limited data :skull:

>>108007735
>>108008823
Trannies always need to make everything about them

>>108010664
>What do you mean by this?
basically, the web player uses a non-standard HLS extension. so the HLS playlist URL which you can extract with some scripting (like the other site) can't be simply directly passed to yt-dlp (or any HLS client). you either need special support in the client(s), or more feasibly, do some scripting to replicate what the HLS non-standard extension does. the scripting solution should be feasible because the player code later uses a standard HLS playlist (i think. i wasn't willing to waste time on it to know for sure).
if you search for the extension name, you will see that a github user may have managed to make it work with streamlink, so maybe look for his code (if it was shared), or try to contact them.
https://github.com/streamlink/streamlink/discussions/6645

I have a NAS that I use for nightly batch downloads with yt-dlp, but it started failing at about 4%
Trying the same download with my normal pc works fine. Did youtube flag my server somehow?

i've never understood what the point of this thread was. what really is there to say about mpv? i've never even seen someone in here writing lua scripts. just chattering

File: 1769565225380290.gif (1.1 MB)

1.1 MB GIF

>>108008704
they really have variables named ass_packer, huh?

>>108013142
Leave then

the lgpl was a mistake
https://github.com/mpv-player/mpv/discussions/17356

>>108013711
Can we just drop X11 support from Mpv?

File: its joke.gif (850.2 KB)

850.2 KB GIF

>>108013172

>>108013741
Once wayland is ready so another 30 years.

Why does my yt-dlp do this at the end of a live stream?

[download]    7.06GiB at    1.82MiB/s (03:50:05) (frag 13829/13830)[download] Got error: The read operation timed out. Retrying (1/10)...
[download] Got error: The read operation timed out. Retrying (1/10)...
[download] Got error: The read operation timed out. Retrying (2/10)...
[download] Got error: The read operation timed out. Retrying (2/10)...
[download] Got error: The read operation timed out. Retrying (3/10)...
[download] Got error: The read operation timed out. Retrying (3/10)...
[download] Got error: HTTP Error 503: Service Unavailable. Retrying (4/10)...


ERROR: Did not get any data blocks
[download] fragment not found; Skipping fragment 13830 ...


ERROR: Did not get any data blocks
[download] fragment not found; Skipping fragment 13831 ...
[download] Got error: HTTP Error 500: Internal Server Error. Retrying (4/10)...
[youtube] XXXX: Downloading webpage

Any way to tell it to merge the damn files anyway?

For reference I'm using a command line like this:

yt-dlp \
--wait-for-video 15 \
  --live-from-start -c --remote-components ejs:npm --cookies cookies.txt --js-runtimes node \
  --no-post-overwrites \
  --no-mtime \
  --write-description \
  --write-thumbnail \
  --convert-thumbnails jpg \
  --write-info-json --format "bestvideo+bestaudio/best" --recode-video mkv

I'm pretty sure everything is there and the server errors are just YouTube being dumb. I have two .f140.mp4.ytdl and f299.mp4.ytdl files now. I could merge them manually but shouldn't have to.

profile=high-quality

Rate my config.

>>108014194
Not high quality enough. You need to set --scale=ewa_lanczos4sharpest

If I use ctrl-z and then fg to suspend an encode with ffmpeg is there any risk of it introducing errors?

>>108015670
A more reliable way is probably to put the encode process in its own CGRoup and then use the CGroup Freezer to freeze/thaw it.
https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html#core-interface-files

This is a commission sample. Why are you reposting this artist's work without their permission?

Anyways, you guys/gals/others tried the latest wayland VO's?

>>108015869
Maybe he had permission?

https://github.com/kasper93/mpv360 this is nice for VR on pc.

>>108016648
Pretty neat. If I buy a Steam Frame (hoping to when it comes out if the price is right) then I'll try this.

>>108017256
isn't this just for viewing VR videos on desktop? I was looking for a good VR video player that ran on linux but couldn't find one.

>>108015670
Unless a resource is external (network), or some other similar complication, then no.
For short STOPs, there shouldn't be a problem in any case.
In fact, you can "kill -STOP ;sleep ;kill -CONT ; sleep " in a loop, and you would get working resource limitation, and this happens to be exactly what the tool "cpulimit" does.
And no, cgroups are not strictly necessary. They are good if you need them. But for simple cases, they can be overkill.

File: cinematography.jpg (397 KB)

397 KB JPG

>>108007298
Is there an extension to reskin the appearance of the OSC? No offense but I'm tired of the same look after seven years. Just want to swap it for a different look.

>>108018109
https://github.com/mpv-player/mpv/wiki/User-Scripts
search osc

If I have crappy DVD rips should I re-encode them to save space with av1 or something? Or leave them as is?

>>108018109
https://github.com/stax76/awesome-mpv?tab=readme-ov-file#on-screen-controller

Holy FUCK the latest build is fast. I have NEVER seen yt-dlp download at nearly 70MiB/s before without even hitching.

It did it again:

ERROR: Did not get any data blocks
[download] fragment not found; Skipping fragment 15495 ...


ERROR: Did not get any data blocks
[download] fragment not found; Skipping fragment 15495 ...


ERROR: Did not get any data blocks
[download] fragment not found; Skipping fragment 15496 ...
[download] Got error: HTTP Error 404: Not Found. Retrying fragment 15496 (1/20)...
Sleeping 5.00 seconds ...
[youtube] XXXX: Downloading webpage

I added these flags but it's still merging the audio and video for me:

--hls-split-discontinuity \
--no-abort-on-error \
--fragment-retries 20 \
--retry-sleep fragment:5 \
--fixup warn

Why is it still exiting with failure at the end of a live stream despite --no-abort-on-error?

Is watching pornhub with mpv working for anyone else?

>>108019880
i can't help you because you're not describing your exact problem properly.
but out of curiosity, what format is the livestream? Is it DASH?

>>108020059
YouTube live stream so I think so. When the stream ends YouTube seems to have some segments that can't be downloaded even though everything else is all there.

Current fix is to just merge the audio and video parts manually with Ffmpeg but ideally I'd have some more robust so I won't have to babysit it.

>>108020495
if it's DASH, there may be no indication of when the stream ends from the client side based on the playlist alone.
the playlist may define segments roughly as:
URL_PREFIX/.
where segment_id can be defined for example as time_stamp/segment_duration.
this allows for not requiring making regular playlist update requests, but you don't get immediate end stream indicator like HLS with "#EXT-X-ENDLIST".

>>108020648
Can I make yt-dlp still merge what I've got though?
If I've been recording for 3-4 hours then it needs to assume I want to merge that even if YouTube's servers keep shitting itself at the end if the stream ends abruptly.
Supposedly this is what --no-abort-on-error does but it doesn't work. It still aborts on error and doesn't merge what I have.

What are the shaders to use for live action content?

zewia = sizumam = linearschizo

Uh, based department? https://kotaku.com/jeffrey-epstein-five-nights-at-freddys-porn-via-4chan-links-emails-show-2000664667

>>108022317
>linearschizo
wot?

>>108024193
there was a guy who liked linear downscaling so much he wanted to use linear upscaling too which is nominally more trivially correct in like the same way as naive greyscale text rendering without hinting or rgb is more correct but it looks like shit and rings badly on real content and when you're interpolating it's all made up anyway so you should just use what looks better not what makes a mathematician happy. i believe he was the same as the catrom guy.

@llyyr Can you explain this meme to me? https://www.reddit.com/r/IndiaMemes/comments/1qs3f94/truth/

>>108021999
ArtCNN works great on live action too

>>108024544
My unpopular opinion is that it is precisely the schizos loonies and lolcows that keep this general alive not esoteric video player autism, it's fast food entertainment but nevertheless it's fun

>>108023713
He got banned from Xbox Live. Lmao.

>>108017875
CGroups were literally made to enforce resource limits. You might as well use them if you can.

You can set niceness, I/O priority, limit number of CPUs and their placement (cpusets), limit memory usage, limit CPU usage, and freeze/thaw tasks all from a simple and elegant pseudo filesystem abstraction.

>new release
>just more youtube client churn, again
>mobile site blocking background play
2026 is definitely the year of widevine on all videos.

>>108027938
Based, ikatube lost.
BBC won.

>>108027938
They would have done this long ago if they didn't need to support legacy clients.

are there any based and redpilled guides for getting started with this software?

playing from ikatube doesn't get subtitle

>>108028751

man yt-dlp

>>108029152
playing from ikatube doesn't get testosterone

File: 9403a0f91bb374e168eda312dc5eef4d.jpg (327.9 KB)

327.9 KB JPG

>>108029152
just fixed it in the latest release

>>108030652
Enjoy your vacation

>>108007735
>>108008534
This is the BBC general

I'm trying to get 60fps 1080p playback running smooth on my old laptop (Arch). This works (integrated GPU):

gpu-api=vulkan
gpu-context=waylandvk
vulkan-device="Intel(R) HD Graphics 4000 (IVB GT2)"
vo=gpu-next

I just installed NVK drivers, and if I swap out the integraded with the dedicated as the vulkan-device, performance is terrible. With proprietary drivers I got screen tearing.

File: 1747692395576444.webm (762.7 KB)

762.7 KB WEBM

>Pornhub
Is there a way to make it work?

>>108032252
>tearing with whyland
lol. use X11.

>>108033992
airpods and JBL btfo

>>108034099
Dropped frames doing x11vk like so with the dedicated GPU

gpu-api=vulkan
gpu-context=x11vk
vulkan-device="Quadro K1000M (NVK GK107)"
vo=gpu-next

Looks identical to using waylandvk

isn't this general supposed to be about a video player? Why are sharties upset again?

File: 1751161231234006.jpg (796.3 KB)

796.3 KB JPG

>>108033946
>help me coooom pls PLSSS
>provides ZERO useful information

>>108033946
stop watching porn

File: notes.png (10.8 KB)

10.8 KB PNG

>>108029768
why did you stop writing patch notes? it looks like youre updating ikatube with nothing since november.

>>108034432
i meant you should use the proprietary driver. if you get tearing in whyland, then try x11 (here, i was half-joking, because you shouldn't get tearing in either. but there is an irony here in whyland tards believing/claiming that x11 has an inescapable tearing problem).
and make sure you actually have hardware vulkan support with your gpu using the proprietary driver. otherwise, you could be falling back to llvm software vulkan rendering, which is extremely slow.
if you don't get real hardware vulkan support with the proprietary driver, use opengl.

>>108035691
She's busy sucking on Tyrone's big Black uncut juicy penis.

File: paradox_paradigm.jpg (118.3 KB)

118.3 KB JPG

>>108035975
It's a Kepler card, which the Arch wiki says is unsupported. When I did the proprietary card, I tried out a bunch of settings, and I don't remember for sure what happened with opengl, but I think it just gave me a segmentation fault. Regarding hardware support, I don't think this Quadro K1000M has good hardware support, because any hwdec settings I've tried alwaays fail. Anyway, I've been trying for Vulkan because the NVK and Nouveau sites say that they support all Kepler cards. Where all this leaves me: I dunno, I guess I could try the proprietary drivers again, but only if I could find some new setting to tweak. All the obvious settings yielded no real results. Either seg faults, or tearing.

>>108036109
Time to upgrade that GPU

File: a31d505c03292569073013418e4f25f8.jpg (154 KB)

154 KB JPG

>>108035691
the notes are only for new features not bug fixes

>>108036777
why do you fuck niggers?

>>108036109
make sure the you use the correct proprietary driver, and with a kernel version that supports it.
with old enough gpus, you may need a dkms package from the AUR, and possibly an older kernel (lts may work for now).
with nvidia, you're always either on borrowed time or stuck with subpar "reverse-engineered" drivers, when it comes of OOTB support.
both of these things may improve with official open drivers which can "theoretically" be independently maintained by volunteers (e.g. supporting newer kernel versions), and modern open drivers who can work well with a lot of functionality effectively shoved into firmware. but the latter doesn't help you considering your gpu's age. and i don't about open-drivers and your specific gpu since i don't buy nvidia.

>>108036777
not same anon but last time its troublesome when it stopped working, it turned out you added new working extractor. I wouldn't know until I found changes in docs.

>>108034765
>>108034846
It fails to get the stream

[ytdl_hook] ERROR: [PornHub] 69040eb7b700d: Executing JS on webpage failed with returncode 1
[ytdl_hook] 140492646471552:error:25066067:DSO support routines:DLFCN_LOAD:could not load the shared library:dso_dlfcn.c:185:filename(libproviders.so): libproviders.so: cannot open shared object file: No such file or directory
[ytdl_hook] 140492646471552:error:25070067:DSO support routines:DSO_load:could not load the shared library:dso_lib.c:244:
[ytdl_hook] 140492646471552:error:0E07506E:configuration file routines:MODULE_LOAD_DSO:error loading dso:conf_mod.c:285:module=providers, path=providers
[ytdl_hook] 140492646471552:error:0E076071:configuration file routines:MODULE_RUN:unknown module name:conf_mod.c:222:module=providers; please report this issue on  https://github.com/yt-dlp/yt-dlp/issues?q= , filling out the appropriate issue template. Confirm you are on the latest version using  yt-dlp -U
[ytdl_hook] youtube-dl failed: unexpected error occurred
:

>>108038014
>>108038014
>

Executing JS on webpage failed with returncode 1

That sounds like you didn't install Deno properly, faggot.

>>108038090
I have phnatomjs

>>108038090
I did that as well and it didn't work

>>108038103
>>108038112
Idk if they're using ejs scripts for that site but doesn't hurt to try with Deno and

--remote-components ejs:github

It works here, I can play anything outta there in mpv

>>108038154
eta is 60 years, hold tight while it's done

>>108038192
kek thanks for the laugh anon, No disagreements, too give a picture of the state of things on i2p, I think seeding to a good client has an average/median or whatever of >100KBps. Not particularly impressive at all but just a smidge better than 60 years lol

>>108038192
>Idk if they're using ejs scripts for that site but doesn't hurt to try with Deno and
I've installed deno, how do you use it?

>>108038014
wadafuk is libproviders.so?
no such a library exists.
this discussion points to builds from retards who don't know how to build openssl:
https://github.com/openssl/openssl/discussions/24433
so, what OS is this? and where did you get yt-dlp (and deps) from?

>>108038192
>eta is 60 years, hold tight while it's done
use 1 hop total. plausible deniability but decent speeds.
>t. last ran i2p when the alternative c++ impl didn't even exist lol

>>108038300
>so, what OS is this? and where did you get yt-dlp (and deps) from?
Endeavour, PIP

>>108038242
It's enabled by default

>>108038229
It shouldn't be that slow with the amount of routers in the network, I wonder if it's people running very low rate limits what's clogging everything up.

SVT-AV1-HDR defaults are already optimized out of the box. At that CRF level I wouldn't pass anything else but --fast-decode 1 and --tile-columns 1 to make it potato friendly.

>>108038311
I have to try that, weird that it never crossed my mind.

>>108038320
https://desuarchive.org/g/thread/107952779/#108007051

>>108038328
Why would anyone use yt-dlp-git?

>>108038345
Why not?

>>108038363
Bleeding Edge Instability
Compilation and Build Failures
Manual Maintenance
Dependency Issues
Lack of Quality Assurance

>>108038432
>Bleeding Edge Instability
yt-dlp is unstable by nature
>Compilation and Build Failures
Can't even remember the exact year I had this issue, pretty sure clearing the AUR helper cache fixed it.
>Manual Maintenance
Not a problem on Arch based distros, you basically run paru/yay.
>Dependency Issues
does not apply here
>Lack of Quality Assurance
never had an issue stemming from a bad commit or one that installing a release build ever fixed

>>108038345
why use pip? yt-dlp is in the repos.

>>108038600
yt-dlp recommend using the pip version.

>>108038607
well. considering you have an error about a non-existent library. it's clearly not working out that way. use a distro package, and see if that error goes away. if it doesn't, try to know where it's actually coming from.
or if you really must, use a full python packaging/env solution like uv, although i never used it myself.

>>108038631
I used the git version same issue.

>>108038686
run with

tracexec log --no-show-env --no-diff-fd -- yt-dlp --verbose 

let's see where your error is actually coming from.

>>108038345
>git
Then cffl wouldn't work

File: c9b450f0c8b6d6c82e2bacb732f52907.png (362.7 KB)

362.7 KB PNG

>>108037486
whenever i upload a new version it's likely to fix any existing critical issues

>>108038432
>Lack of Quality Assurance
I have some really bad news for you.

>>108038103
Uninstall it. PhantomJS is severely unmaintained and a security risk at this point.

>>108038830
Cffi works if you build it with support for it. I have no idea what the yt-dlp-git AUR maintainers are doing though.

>>108013711
wm4 didn't need x11 support in the proprietary media service shit he was working for

>>108040476
I'm pretty sure there are still a few extractors that need phantomjs specifically and won't work with deno.

>>108042803
I think they're going to be migrated at some point as soon as somebody does the work though. PhantomJS is a liability at this point

File: G__YNo-acAAgXbn.jpg (86.2 KB)

86.2 KB JPG

Dont, mind me, just testing.

>>108007298
kys IIyyr, nobody's ever liked you

>>108038738
Same error man
>>108040476
I did that and installed deno, didn't solve it
>>108040497
You're right but cffi works with pip, that's why I've kept it.
DESU, my pip tree is shit.
I should have migrated to pipx by now.

>>108044582
>Same error man
post the full output including tracexec info

as of tonight/now, svt-av1-hdr doesn't work with gst-plugins in arch linux i guess. sad:(, i just wanted to watch toradora with my dad with av1 and opus in peace

>>108047247
>gst
lol

>>108046219
Why you need me to install tracexec?

>>108048778
so we know which exact process(es) the error(s) come from.
if you can figure that out without tracexec, have at it?

>>108049033
Ytdl, it's not running the JS framework

>>108049055
you're clearly clueless. install tracexec and post complete output.

>>108047247
> av1 and opus
yuck

File: 1753519042349361.webm (551.4 KB)

551.4 KB WEBM

When I do -ss after -i it takes too long.

>>108052175
How to get nice looking teeth?
Mine is yellowed.

>>108047247
>he installed it systemwide
lmao

>>108052175
no shit, it's reading the whole file instead

>>108052210
ffmpeg -hide_banner -i ugly_chompers.ivf -vf libpacebo=saturation=0.12 -pix_fmt yuv420p10le -c:v libsvtav1 -crf 26 -preset 3 -svtav1-params enable-variance-boost=1:variance-boost-strength=3:variance-octile=5:enable-tf=0:sharpness=1:fast-decode=1:tile-columns=1:tile-rows=0 -c:a libopus -b:a 192k pearly_whites.mp4

>>108053940

[AVFilterGraph @ 0x564422feb100] No such filter: 'libpacebo'
Error opening output file pearly_whites.mp4

>>108053994
fixed:
ffmpeg -hide_banner -i ugly_chompers.ivf -vf libplacebo=saturation=0.12 -pix_fmt yuv420p10le -c:v libsvtav1 -crf 26 -preset 3 -svtav1-params enable-variance-boost=1:variance-boost-strength=3:variance-octile=5:enable-tf=0:sharpness=1:fast-decode=1:tile-columns=1:tile-rows=0 -c:a libopus -b:a 192k pearly_whites.mp4

>>108032252
use opengl

Rate my config
`interpolation
tscale=sphinx
tscale-radius=3.0
tscale-blur=0.55
tscale-clamp=0.15
tscale-antiring=0.6
tscale-taper=0.15`

>>108007298
so how do i use fsr and framegen in ffmpeg to upscale low res videos? i only read in the news that they added them in ffmpeg but haven't seen any example script how to actually use them.

This has been annoying me for several months but since we have /mpv/ threads again:
Does anyone else experience flickering on the bottom of the mpv window with border=no? It's either a solid white horizontal bar across the whole bottom of the window or flickering video frame. It's possible the bottom of the window is even constantly resizing instead of a true bar/flicker.

I reliably reproduce by doing the following:
Open mpv, play a video
Fullscreen, exit fullscreen
Open a different app/window maximized above the mpv window
Minimize the maximized window

If the above doesn't noticeably trigger it, minimizing and bringing the mpv window back should eventually.

Arch, KDE/Plasma 6.5
--no-config --border=no is enough to reproduce

I think I noticed this around when KDE/Plasma 6.5 released so I don't know if it's a KDE/Plasma bug or mpv/libplacebo since I build both mpv and libplacebo from git every few weeks.

>>108054014
>>108053940
>av1
also no difference

File: glyph missing.png (5.6 KB)

5.6 KB PNG

The drawtext filter has changed in ffmpeg!!

Before (September 2025) I could give

text='hello\nworld'

And it would print
>hello
>world

But now (February 2026) ffmpeg prints
>hello[]
>world

[] is that box replacement character for when a font is missing the glyph. It still linebreaks *after* the box. Who fucked this up? Where can I even read patch notes for the drawtext filter specifically? My code or font file hasn't changed, only the version of ffmpeg.

>>108058151
Why would you want that? FSR is designed for low-quality realtime upscaling of video games. If you're using Ffmpeg then you obviously have no time window you need to meet so can use slower but much more high quality shaders, etc, that don't need to operate in realtime.

Admittedly I don't know what people are using for that but FSR ain't it.

>>108061141
I also attempted using a text file:

textfile='path.txt'

The UTF8 .txt contained two lines:
>trying
>hard
ffmpeg (drawtext) prints
>trying[]
>hard

>>108059963
I am 99% sure this is a Plasma bug. Plasma lies to mpv about whether it has borders. You can tell because --border=no is supposed to imply --script-opts=osc-windowcontrols=yes but Plasma is the only desktop in the entire world where that doesn't actually happen.
>>108061141
https://github.com/search?q=repo%3AFFmpeg%2FFFmpeg+drawtext&type=commits&s=committer-date&o=desc

>>108062697
>https://github.com/search?q=repo%3A...
thanks, so i guess it means that they haven't discovered the new bug created or the bug is actually a feature

>>108062763
People think ffmpeg is all high quality code thanks to a few hand written assembly decoders but the truth is they merge a lot of very shitty code due to an equally shitty code review process. Especially in the nischer parts. Go read some merge requests it's fucking scary how incompetent many of them are.

>>108064387
More recently they get a bunch of people allegedly writing AI slop without disclosing it so that doesn't help. Even if they mean to good they can end up touching unrelated parts of the codebase or introducing crap overly verbose code that potentially breaks something.

I don't envy the Ffmpeg maintainers having to deal with that.

File: 7.gif (475.6 KB)

475.6 KB GIF

>>108064387
>>108064473
open-source is scary. at the start of the Russia-Ukraine war I remember that dependency that probably 1000s of projects used that people just happily merged without checking what was changed. it created a text file on your desktop in support of ukraine. harmless enough all things considered but it could have also erased your hard drive if ran after a certain date. open-source projects can have 100s of dependencies and it's so easy to just trust them without checking anything.