BtbN changed the topic of #ffmpeg to: Welcome to the FFmpeg USER support channel | Development channel: #ffmpeg-devel | Bug reports: https://ffmpeg.org/bugreports.html | Wiki: https://trac.ffmpeg.org/ | This channel is publically logged | FFmpeg 7.0 is released
System_Error has quit [Remote host closed the connection]
Juest has quit [Ping timeout: 260 seconds]
System_Error has joined #ffmpeg
Mister_Magister has quit [Quit: bye]
Juest has joined #ffmpeg
Mister_Magister has joined #ffmpeg
HarshK23 has quit [Quit: Connection closed for inactivity]
iive has quit [Quit: They came for me...]
noobaroo has joined #ffmpeg
<noobaroo> I found the command `mp4box -zmov` that uses ISOBMFF box compression but the videos are unrecognizable by ffmpeg afterward, is this unsupported?
flotwig_ has joined #ffmpeg
flotwig has quit [Ping timeout: 240 seconds]
lucas_ta has joined #ffmpeg
lucasta has quit [Ping timeout: 272 seconds]
Suchiman has quit [Quit: Connection closed for inactivity]
milkt has joined #ffmpeg
lemourin has quit [Quit: The Lounge - https://thelounge.chat]
milkt has quit [Remote host closed the connection]
milkt has joined #ffmpeg
lemourin has joined #ffmpeg
zenloading has joined #ffmpeg
minimal has quit [Quit: Leaving]
<Oksana> galad: .mp4. Recorded by Windows Snipping Tool. How do I check what the codec of that file is?
Juest has quit [Ping timeout: 268 seconds]
Juest has joined #ffmpeg
waleee has quit [Ping timeout: 240 seconds]
Vonter has joined #ffmpeg
<emanuele6> ffprobe?
lucas_ta has quit [Quit: Leaving]
MootPoot has joined #ffmpeg
Keshl has quit [Ping timeout: 245 seconds]
Keshl_ has joined #ffmpeg
Keshl_ is now known as Keshl
Ox7C5 has joined #ffmpeg
milkt has quit [Remote host closed the connection]
milkt has joined #ffmpeg
Juest has quit [Remote host closed the connection]
johnjaye has quit [Ping timeout: 260 seconds]
AlexandrKirilov has joined #ffmpeg
AlexandrKirilov has left #ffmpeg [#ffmpeg]
johnjaye has joined #ffmpeg
Juest has joined #ffmpeg
Muimi has joined #ffmpeg
Tinos has quit [Ping timeout: 250 seconds]
xx has joined #ffmpeg
nadir has joined #ffmpeg
nadir has quit [Client Quit]
<yrc> Hello :-)
jagannatharjun has joined #ffmpeg
kts has joined #ffmpeg
beaver has joined #ffmpeg
<yrc> Offspring: I have to admit I don’t fully understand what you write. In short, the crop expression is "crop=<croped width>:<croped height>:<left offset>:<top offset>". So if you start with a width of 1920 and want to remove 5px on the right and 2px on the left, then you’d get a width of 1920-7=1913, and a left offset of 2. Beware, though: the _final_ width and height should have W and H multiple of 8 (if
<yrc> not 16), and must at least be multiple of 2.
<yrc> Hi all. Regarding the issue I was reporting earlier (~12h30 ago), I may be jumping to conclusions too fast, actually, given my limited video & audio knowledge.
<yrc> What I experience is this: in the final MKV where all chapters are put one after the other, when played with mpv or vlc, I can see each chapter mark at exactly the right time location, and I can move along the timeline freely, keeping a perfect sync between video, audio, and subs.
<yrc> BUT the whole film (vid+aud+sub) ends _after_ what should be the end of the last chapter; the timeline is thus longer than it should be.
<yrc> Also, FYI, this is an MKV using ordered chapters (so that following Edition 1 = film, therer is Edition 2 = bonus), and when I watch the “extra chapter” of the film, half of it is the end of the ending credits, and half of it is the beginning of the next Edition :-/
<yrc> (still with perfect syncing). And when I do play the next Edition (bonus), it starts off, because the actual beginning of it is at the end of the film Edition.
<yrc> Basically, all footage looks correctly synced, _but_ spread more than it should, so I guess it plays imperceptibly slower than it should.
<galad> Oksana: make another file with windows snipping tool, and use something like untruc to fix the other file
hjckr has joined #ffmpeg
georgereynolds8 has joined #ffmpeg
The_Blode has joined #ffmpeg
ivanich has joined #ffmpeg
kts has quit [Ping timeout: 260 seconds]
HarshK23 has joined #ffmpeg
kts has joined #ffmpeg
Muimi has quit [Quit: Going offline, see ya! (www.adiirc.com)]
<Offspring> yrc: sorry if I'm not explaining it clearly, I'm still new to learning how to use ffmpeg for the encoding, but ran into this fun situation. Here's the full thing, I hope: the video is 1920x1080, but the left has a couple (2px) of black pixels, and the right has 5 pixels that I'm trying to crop out. I was able to crop out the 5 pixels with your help earlier with the 1920x1076:0:0 command. The problems I'm running into are two-fold: 1) how do I keep
<Offspring> the AR if I want to encode it in a 720p resolution? 2) how do I remove the 2px on the left, and the 5px on the right while answering the first question as well?
<furq> crop=w=1912:h=1080:x=2:y=0,scale=-2:720
<furq> or crop=w=iw-8:h=ih:x=2:y=0,scale=-2:720
Blacker47 has joined #ffmpeg
Suchiman has joined #ffmpeg
<Offspring> wouldn't that crop both sides the same amount, though?
michaelni has quit [Ping timeout: 255 seconds]
lavaball has joined #ffmpeg
Ox7C5 has quit [Quit: Lost terminal]
<vincejv> hello! I'm having some weird color shift with ffmpeg screenshot via ffmpeg -y -ss "$time_point" -i "$input_video2" -vframes 1 "$output_frame2 here's the comparison pic https://slow.pics/c/3amrNSXU the same video file, one is ffmpeg, one is taken from MPV
<vincejv> specifying the pixfmt didn't help
<vincejv> tried rgb24 yuv420
<vincejv> fmpeg version 4.4.2-0ubuntu0.22.04.1
milkt has quit [Ping timeout: 260 seconds]
<Blacker47> vincejv, you maybe should start with configuring mpv to write as png to avoid lossy compression impact.
<vincejv> Blacker47 it doesn't matter png or jpg (yes there is quality loss), I actually did a conversion using SVT-AV1 ffmpeg, and the AV1 file that ffmpeg produced the same colors as the one from MPV
<vincejv> i mean after i converted the video to AV1, i took a screenshot with ffmpeg, same command
<vincejv> and it matched with MPV output
<Blacker47> vincejv, i would say it is not how one should debug issuses.
<vincejv> i noticed this when i was doing comparison screenshots between the original h264 file and av1, there was a color shift... so i thought it was av1's fault... turns out it was not.. i took screenshots on the same file with ffmpeg and mpv...
<vincejv> ffmpeg screenshots of the raw file was not matching with the colors from screenshots with mpv
<vincejv> for the same file
<vincejv> but the av1 file and mpv screenshots matched colors
<vincejv> Here's the full set of images https://slow.pics/c/LtAIAV0g the outlier to my eyes is Image 1/4, which was the ffmpeg screenshot of the raw file
<vincejv> i labeled them, 2/4 mpv from raw file, 3/4 ffmpeg screenshot from av1, 4/4 mpv screenshot from av1
<vincejv> i mean as you can see here images 2 and 4 are jpgs, but the colors from 2, 3, 4 are "consistent" so it's not a "lossy compression impact"
<vincejv> might be a difference on brightness, but the colors are "quite" the same to my eyes
kts has quit [Ping timeout: 252 seconds]
<vincejv> i mean i can make it all png, but that would just retain more details of the screenshot, the issue are the colors.. or the color shift
michaelni has joined #ffmpeg
Tano has quit [Quit: WeeChat 4.2.1]
snoriman has joined #ffmpeg
<snoriman> hi, when using a DeckLink Mini Recorder, does someone maybe knows if it's possible to force the input to be in yuv?
Tano has joined #ffmpeg
vlm has joined #ffmpeg
pikapika is now known as militantorc
<alien_lappy> i'm doing fluster testing for the h264_v4l2m2m, but i get an off-by-one error; and someone told me, this may be some automatic color conversion? can anyone help me out with this? is there a way to prevent automatic color conversion?
<JEEB> `-noauto_conversion_filters`
<JEEB> that stops insertion of scale or aresample filters
<JEEB> those are logged with the last log level which is not spammy, `-v verbose`
<alien_lappy> oh, thanks
<alien_lappy> i'll go double check that
<alien_lappy> ok, i think my ffmpeg is too old, it doesn't have this option, i should update this, especially for fluster testing
<alien_lappy> also, i did not see any mention of it with verbose tbh, so maybe that isn't the case, or just the ffmpeg is too old
<alien_lappy> maybe the off-by-one is due to the firmware, i'll also poke some people there
<alien_lappy> i think this version might even be too old to even have auto conversions :-)
<JEEB> nah, that is part of the standard filter chain stuff
<JEEB> if pix_fmts don't match between a filter and the next one (including the input/output things that ffmpeg CLI configures based on the decoder/encoder's wanted ones), the filter chain attempts to add scale or aresample
<JEEB> alien_lappy: `ffmpeg -h full |grep auto_` should output the non-negative of it (which is the default)
<Netham45> I've got two input streams each running through aresample and volume before amix, I can change the volume of the first stream with 'Parsed_volume_0 -1 volume <0-1>' but I can't figure out how to do the second. Anyone know how/if I can target the second volume filter, or where some docs on it are?
<Netham45> From the interactive command line
System_Error has quit [Ping timeout: 260 seconds]
<Netham45> I'm pretty sure I need to name the volume with 'volume@volume_1=1.0,' but I don't know how to format a target filter
System_Error has joined #ffmpeg
<Netham45> to target that
<alien_lappy> JEEB: https://pastebin.com/wxUJyxKy i think both are yuv420p , but this old ffmpeg version does not have this setting ; i'll have to put a newer OS on this test RPI4 to do the rest
<Netham45> Nevermind, got it.
kts has joined #ffmpeg
kts has quit [Remote host closed the connection]
<JEEB> alien_lappy: I hope you're comparing the YCbCr data only. also unless the encoding is lossless, it's not going to match 1:1
<JEEB> but yea, I don't see a scale filter there being inserted
<JEEB> so it *should* be going input-to-output without conversions
kts has joined #ffmpeg
yrc has quit [Ping timeout: 256 seconds]
xx has quit [Ping timeout: 260 seconds]
kts has quit [Ping timeout: 240 seconds]
<vincejv> okay i figured out the color shift, i added the video filter -vf scale=in_color_matrix=bt709:out_color_matrix=bt709 and it fixed the colors, i guess mpv applies this as well
<JEEB> yea, the scale filter still doesn't configure that according to input
<JEEB> you should be able to see in `-v verbose` command line output the info for the input under the video stream
<JEEB> whether the input has that metadata flagged
<vincejv> it doesn't have it flagged, i checked with mediainfo
<JEEB> is it over 1024x576 in resolution?
<alien_lappy> JEEB: also, it seems i misunderstood, it may be the RPI4 config that has the firmware not do color conversion (instead of ffmpeg doing it)
<vincejv> 1080p
<JEEB> yea, mpv will expect unflagged 1080p to be BT.709
<JEEB> best is to flag the content, of course
yrc has joined #ffmpeg
blb has quit [Ping timeout: 246 seconds]
blb has joined #ffmpeg
kts has joined #ffmpeg
milkt has joined #ffmpeg
frankplow has quit [Quit: Goodbye]
frankplow has joined #ffmpeg
kts has quit [Ping timeout: 264 seconds]
ivanich has quit [Ping timeout: 256 seconds]
klaxa has quit [Quit: Quit.]
klaxa has joined #ffmpeg
e^pi-1 has quit [Quit: WeeChat 4.2.2]
Ox7C5 has joined #ffmpeg
Offspring has quit [Quit: ZNC - https://znc.in]
Offspring has joined #ffmpeg
minimal has joined #ffmpeg
waleee has joined #ffmpeg
SystemError has joined #ffmpeg
System_Error has quit [Remote host closed the connection]
waleee has quit [Quit: WeeChat 4.1.2]
rv1sr has joined #ffmpeg
CheckMyBrain has quit [Quit: Connection closed for inactivity]
riotz has joined #ffmpeg
riotz has quit [Client Quit]
Marth64 has joined #ffmpeg
Tinos has joined #ffmpeg
Marth64 has quit [Remote host closed the connection]
<yrc> Hi! I’m back with my duration issue. It seems I was once again mistaken in my assumption.
<yrc> Short reminder of context: I rip individual chapters into MKV files using `ffmpeg -f dvdvideo`. To know what to rip with which characteristics, I use `ffprobe -f dvdvideo` first on each title. For instance, I have 12 chapters.
AbleBacon has joined #ffmpeg
<yrc> After ripping, each chapter’s MKV played with VLC shows the exact length reported by ffprobe (±1s). There is one exception, though: chapter 5. ffprobe reports length 00:11:46.080000, but VLC shows 08:35, which is consistent with what ffprobe reports on the ripped MKV (e.g. 00:08:34.840000000 for the video stream).
<yrc> So either ffprobe -f dvdvideo reported a wrong information, or ffmpeg truncated streams early.
<yrc> After viewing the chaining of chapters 5 and 6, I would lean on the former, i.e. ffmpeg truncated the chapter while ripping. Any idea why?
<yrc> In case that can help, here is the full command that rips this chapter:
<yrc> 'ffmpeg' '-hide_banner' '-f' 'dvdvideo' '-preindex' '1' '-trim' 'false' '-title' '1' '-chapter_start' '5' '-chapter_end' '5' '-i' '/home/yves/AV/.iso' '-i' '/home/yves/AV/tmp.empty_a.stereo.48000.192000.706.080000.ac3' '-i' '/home/yves/AV/tmp.empty_a.5.1(side).48000.448000.706.080000.ac3' '-map_chapters' '-1' '-map' '0:0' '-map' '0:1' '-map' '1:a:0' '-map' '0:2' '-map' '1:a:0' '-map' '2:a:0' '-map' '0:3'
<yrc> '-map' '0:4' '-map' '0:5' '-map' '0:6' '-map' '0:7' '-map' '0:8' '-sws_flags' 'lanczos+accurate_rnd' '-filter:v:0' 'bwdif,crop=x=0:y=72:w=720:h=432:exact=1,scale=w=1024:h=432:force_original_aspect_ratio=disable,setsar=1/1' '-enc_time_base:v:0' 'demux' '-c:v:0' 'libx265' '-x265-params:v:0' 'profile=main10:preset=slower:crf=18:sar=1:videoformat=pal:rc-lookahead=120:bframes=12:ref=6:subme=7:aq-mode=3' '-c:a:0'
<yrc> 'copy' '-map_metadata:s:a:0' '0:s:1' '-metadata:s:a:0' 'language=fre' '-metadata:s:a:0' 'title=1: FRE 6🕩 5.1(side)' '-c:a:1' 'copy' '-metadata:s:a:1' 'language=fre' '-metadata:s:a:1' 'title=2: FRE 2🕩 stereo' '-c:a:2' 'copy' '-map_metadata:s:a:2' '0:s:2' '-metadata:s:a:2' 'language=eng' '-metadata:s:a:2' 'title=3: ENG 6🕩 5.1(side)' '-c:a:3' 'copy' '-metadata:s:a:3' 'language=eng' '-metadata:s:a:3'
<yrc> 'title=4: ENG 2🕩 stereo' '-c:a:4' 'copy' '-metadata:s:a:4' 'language=jpn' '-metadata:s:a:4' 'title=5: JPN 6🕩 5.1(side)' '-c:s:0' 'copy' '-map_metadata:s:s:0' '0:s:3' '-metadata:s:s:0' 'language=fre' '-metadata:s:s:0' 'title=1: FRE' '-c:s:1' 'copy' '-map_metadata:s:s:1' '0:s:4' '-metadata:s:s:1' 'language=fre' '-metadata:s:s:1' 'title=2: FRE²' '-c:s:2' 'copy' '-map_metadata:s:s:2' '0:s:5'
<yrc> '-metadata:s:s:2' 'language=fre' '-metadata:s:s:2' 'title=3: FRE³' '-c:s:3' 'copy' '-map_metadata:s:s:3' '0:s:6' '-metadata:s:s:3' 'language=fre' '-metadata:s:s:3' 'title=4: FRE⁴' '-c:s:4' 'copy' '-map_metadata:s:s:4' '0:s:7' '-metadata:s:s:4' 'language=fre' '-metadata:s:s:4' 'title=5: FRE⁵' '-c:s:5' 'copy' '-map_metadata:s:s:5' '0:s:8' '-metadata:s:s:5' 'language=fre' '-metadata:s:s:5' 'title=6: FRE⁶'
<yrc> '/home/yves/AV/t.1.ch.5.mkv'
<microchip_> dude, use a pastebin
<yrc> Yeah… sorry. I did not realize it was that long compared to the IRC line-size, for a single command :-o
xx has joined #ffmpeg
jarthur has joined #ffmpeg
milkt has quit [Quit: leaving]
<yrc> Interestingly, if I _remove_ the `-preindex 1` option from the command, I get a chapter file that is indeed 11min long instead of 8min, but both VLC and MPV abort playback at 08:35 as if the file was still truncated :-(
MisterMinister has joined #ffmpeg
Narrat has joined #ffmpeg
Snuupy has joined #ffmpeg
Snuupy has left #ffmpeg [leaving]
vlm has quit [Remote host closed the connection]
sebonirc has joined #ffmpeg
<sebonirc> hi! is it possible to continue a failed ffmpeg conversion from where it left off? the command i ran was a simple `ffmpeg -i input.mp4 output.mp4`, and it was able to convert almost the entire thing before running out of disk space. i freed up disk space, and if possible i'd like to start where it left off (or close to it) rather than restarting the entire thing
<sebonirc> if it helps, i know the frame that it failed on
<BtbN> not sensibly
<BtbN> also, the defaults for such a conversion are horrible
<BtbN> you'll want to be much more specific in terms of what output format and quality you expect
Livio has joined #ffmpeg
<sebonirc> i don't care too much about the specifics of the output. the input file is 11gb, which, is a hell of a lot bigger than it should be, so really i just wanted to get it to a reasonable-ish size. i assumed that the defaults would be fine if i don't want to think too much about the exact output format and quality
<sebonirc> what do you mean by "not sensibly"?
<BtbN> You could in theory manually find out where it broke, and seek there. But getting a clean cut is next to impossible
<BtbN> so much easier to just start over, and use proper settings this time
<BtbN> The defaults are very poor
Narrat has quit [Quit: They say a little knowledge is a dangerous thing, but it's not one half so bad as a lot of ignorance.]
<sebonirc> what settings would you recommend, given that i don't know much about audio/video encoding and just want something that Works for most things? i don't want, like, noticeable loss in quality, but also i'm fine if it's a lossy conversion if it's not really noticeable when watching it
<sebonirc> and a clean cut is impossible even if i know the frame it stopped on?
<sebonirc> the video is around 1h25m, and the input file is over 11gb, for context. i'm not even sure why either, it's not exceptionally high quality or anything. so i just want to get the size down to something more reasonable (the conversion i did with default settings converted ~1h21m before failing, and produced a file which is 2.6gb, and like that's still bigger than it could be but it's already a huge
<sebonirc> improvement so)
iive has joined #ffmpeg
namazso has quit [Remote host closed the connection]
namazso has joined #ffmpeg
Ox7C5 has quit [Quit: Lost terminal]
aus has quit [Remote host closed the connection]
aus has joined #ffmpeg
CheckMyBrain has joined #ffmpeg
rv1sr has quit []
<yrc> A clean cut is nearly impossible. I do that for DVD-to-MKV backup purposes, and even though I’m very attentive to precision to the finest precision available through the tools I use (ffmpeg, etc.), when putting pieces back together, I still have an almost-imperceptible (but still there) video or audio pause of a few milliseconds once in a while where the stitching occurred.
vlm has joined #ffmpeg
waleee has joined #ffmpeg
<yrc> sebonirc: ^. Also, if you look some lines up in this channel, you’ll see I posted encoding parameters for H265 that are good enough to obtain a very good quality with parameters that fit most video sources, in case H265 output is OK for you instead of MP4.
<yrc> Of course, you’d need to adapt crop to your needs, and probably avoid scaling altogether.
relue has quit [Ping timeout: 264 seconds]
<sebonirc> yrc: i just joined this channel so i don't have the history, sry. could you resend the encoding parameters?
SystemError has quit [Remote host closed the connection]
SystemError has joined #ffmpeg
<yrc> Main part is: -c libx265 -x265-params profile=main10:preset=slower:crf=18:rc-lookahead=120:bframes=12:ref=6:subme=7:aq-mode=3
<yrc> Hum, I should probably have written -c:v instead of -c.
relue has joined #ffmpeg
<sebonirc> thanks :)
<yrc> np
<another|> of
Blacker47 has quit [Quit: Life is short. Get a V.90 modem fast!]
ivanich has joined #ffmpeg
lavaball has quit [Remote host closed the connection]
<yrc> Hi. I’ve finally found out that the issue I have in ripping my DVD’s title 1 chapter 5 comes from a limitation or bug in ffmpeg: running `ffplay -f dvdvideo -title 1 -chapter_start 5 .iso` plays fine until 8:35, then the video freezes even though the counter in the console is still going on. In VLC, the chapter plays fine. How could I possibly report that, considering “intelectual property”? I mean, I
<yrc> can’t just drop a 7.5GB copy-protected ISO file in a bug report, can I?
Haxxa has quit [Quit: Haxxa flies away.]
Haxxa has joined #ffmpeg
robobub has quit [Quit: Connection closed for inactivity]
kasper93 has quit [Ping timeout: 264 seconds]
vlm has quit [Quit: Leaving]
xx has quit [Ping timeout: 260 seconds]
Tinos has quit [Ping timeout: 250 seconds]
TheSashmo has joined #ffmpeg
jagannatharjun has quit [Quit: Connection closed for inactivity]
<Ingvix> I'm trying to set an overlay on a lavfi generated colour to create a png but I always end up with just the solid colour on the output
<Ingvix> my command: ffmpeg -ss 00:00:15 -i 20240413_113103.mp4 -ss 00:01:10 -i 20240413_113103.mp4 -f lavfi -i color=c=black:s=1920x1080 -frames 1 -filter_complex "[0:v][1:v]blend=all_mode=grainextract[blend];[blend]colorkey=0x8A9F9C:0.05:0.1[ck];[ck]alphaextract[a];[0:v][a]alphamerge[merged];[merged][2:v]overlay" test%01d.png
Tinos has joined #ffmpeg
<Ingvix> I can see the supposed overlay with transparent parts if I remove the overlay filter so it's not empty
Tinos has quit [Remote host closed the connection]
YuGiOhJCJ has joined #ffmpeg
Tinos has joined #ffmpeg
SuicideShow has quit [Ping timeout: 245 seconds]
SuicideShow has joined #ffmpeg
Dagger has quit [Ping timeout: 260 seconds]
e^pi-1 has joined #ffmpeg
FlorianBad has quit [Remote host closed the connection]
Dagger has joined #ffmpeg
Livio has quit [Ping timeout: 264 seconds]
Tinos has quit [Ping timeout: 250 seconds]
kasper93 has joined #ffmpeg
noobaroo has quit [Quit: WeeChat 4.2.2]
JanC has quit [Ping timeout: 268 seconds]
YuGiOhJCJ has quit [Ping timeout: 260 seconds]
YuGiOhJCJ has joined #ffmpeg
kasper93 has quit [Remote host closed the connection]
iliv has quit [Ping timeout: 246 seconds]