[Ffmpeg-devel-irc] ffmpeg.log.20160627

burek burek021 at gmail.com
Tue Jun 28 02:05:01 CEST 2016


[04:25:36 CEST] <Bray90820> Can someone help me here I seem to be getting some errors
[04:25:36 CEST] <Bray90820> http://pastebin.com/W2GDHGs7
[04:26:46 CEST] <furq> [mp4 @ 0x7fbd0a01c600] Could not find tag for codec wmav2 in stream #1, codec not currently supported in container
[04:26:50 CEST] <furq> what's not clear about this error
[04:31:12 CEST] <Bray90820> Honestly I have no idea
[04:33:25 CEST] <furq> you can't mux wma audio into an mp4
[04:33:35 CEST] <furq> use a different container (mkv) or reencode the audio to aac
[05:01:43 CEST] <Bray90820> furq: Could passthrough the video and convert the audio to something else?
[05:32:21 CEST] <Bray90820> I am also getting another error repeatedly
[05:32:25 CEST] <Bray90820> Past duration 0.731682 too large
[06:31:26 CEST] <thebombzen> Bray90820: yes
[06:31:35 CEST] <thebombzen> use -c:v copy to copy the input video
[06:37:55 CEST] <Bray90820> theholyduck: prob not but could I use --c:v to do a direct copy to mp4 like without loosing quality
[06:40:35 CEST] <Bray90820> Sorry wrong ping
[07:18:28 CEST] Action: theholyduck SLAPS Bray90820
[07:38:33 CEST] <Bray90820> Haha
[10:49:39 CEST] <totokaka> I have 160 png images with size 47616x26880. Can I use ffmpeg to create a timelapse from a cropped area of them?
[10:50:23 CEST] <totokaka> If I try just the standard way, I get this error: [png @ 0xa5f6a0] [IMGUTILS @ 0x7fff397bc5d0] Picture size 47616x26880 is invalid
[11:01:18 CEST] <jkqxz> totokaka:  ffmpeg doesn't like images which have more than INT_MAX/8 pixels (i.e. 16kx16k) because of possible integer overflow issues.
[11:01:31 CEST] <BtbN> totokaka, you realize that a single image easily eats 3GB of ram?
[11:01:59 CEST] <totokaka> I'm rendering on a 12 core VPS with 48GB RAM
[11:02:10 CEST] <totokaka> jkqxz, Does that mean I'm completely out of luck?
[11:03:21 CEST] <totokaka> I though I maybe could get ffmpeg to crop and scale the images on the input side
[11:04:02 CEST] <jkqxz> You might be able to get it to work by cutting each image into 4 or 6 pieces with some other tool?  Really ffmpeg is not made to deal with images of this size, so I don't really know.
[11:04:48 CEST] <totokaka> Well, the images are already built up of small tiles. I have used another script to paste them
[11:05:24 CEST] <totokaka> So I do have a ton of really small tiles that build up the bigger images.
[11:32:48 CEST] <CaptainPrice> I am trying to make a graphical visualizer with ffmpeg. I know the basics of signal processing. Im guessing the height of bars can be calculated using Fourier Transform
[11:33:14 CEST] <CaptainPrice> Can anybody actually help me with a simple tutorial on how to achieve this using API of ffmpeg ?
[11:34:16 CEST] <durandal_1707> CaptainPrice: see showspectrum/showfreqs filter source code
[11:37:42 CEST] <CaptainPrice> durandal_1707: Is it a part of avfilter ?
[15:28:47 CEST] <P4Titan> Hello all. I changed the output bit rate of the transcode_aac.c example to 32000 and the output file it produces fails
[15:29:34 CEST] <orgasmix> Hi. Converting from flac to aiff: how to select PCM format? `ffmpeg -i source.flac -c:a pcm_s32le -f aiff out.aiff` yields `Could not write header [...] Operation not permitted`. Leaving the codec option out defaults to 16-bit.
[15:43:26 CEST] <DHE> is 32 bit supported in the file format?
[15:47:45 CEST] <orgasmix> Converting to wav also fails with -f pcm_s24le and pcm_s32le
[15:48:40 CEST] <orgasmix> A non-empty file gets created but ffmpeg cannot -i it back and VLC will think it's mp3-64k
[16:01:27 CEST] <durandal_1707> orgasmix: be variant
[16:04:29 CEST] <orgasmix> tried too. I could extract the raw PCM but couldn't mux it as neither wav not aiff using s[24|32][le|be]. ableton did it for me though, into aiff, and `ffmpeg -i that_file_from_ableton.aiff` says `Audio: pcm_s24be, 48000 Hz, 2 channels, s32 (24 bit), 2304 kb/s`.
[16:09:59 CEST] <orgasmix> nvm ffmpeg does. `-c:a pcm_s24be out.aiff` does the thing. says "s32 (24 bit)" and vlc says 32 bit per sample also, but the file takes up less size than with pcm_s32be. no idea why. thanks for your attention
[16:13:45 CEST] <durandal_1707> That's normal, its how it is decoded
[16:42:31 CEST] <fritsch> with the new send / receive api and other deprectations with 3.1 I wonder where to get the AVCodecContext from if this is deprecated: avcodec_send_packet(m_fctx->streams[0]->codec, &pkt);
[16:42:44 CEST] <fritsch> e.g. accessing m_fctx->streams[0]->codec
[17:04:29 CEST] <Psi-Jack> Is it possible, using ffmpeg, to convert flac media to m4a media, with more thorough metadata conversion? Right now, it seems to get like bare minimums, but not very comprehensive. Generally only enough to get a rough match with musicbrainz to be able to manually fix the rest of it.
[17:15:34 CEST] <m3gab0y> anyone up for a challenge?
[17:24:56 CEST] <sza> Hello, having trouble with ffmpeg and will be glad if someone can help me.
[17:25:18 CEST] <sza> I'm trying to convert .png files into a video file.
[17:25:38 CEST] <sza> the files are located under the same directory and their names are 1.png, 2.png and so on
[17:26:03 CEST] <sza> I'm trying to use ffmpeg with the following command
[17:26:04 CEST] <sza> ffmpeg -framerate 1/5 -i %d.png -c:v libx264 -r 30 -pix_fmt yuv420p out.mp4
[17:26:13 CEST] <sza> The error I get is:
[17:26:23 CEST] <sza> Error while opening encoder for output stream #0:0 - maybe incorrect parameters such as bit_rate, rate, width or height
[17:26:31 CEST] <sza> What  am I doing wrong?
[17:27:24 CEST] <furq> sza: there should be an actual error above that
[17:28:36 CEST] <sza> http://pastebin.com/9kwN6DFs
[17:28:44 CEST] <sza> This is the entire output
[17:28:49 CEST] <furq> [libx264 @ 0x2eb1700] width not divisible by 2 (757x600)
[17:29:30 CEST] <furq> add -vf scale=758:600 or something
[17:30:10 CEST] <sza> why should it be divisible by 2?
[17:30:19 CEST] <furq> because you're using yuv420p
[17:30:27 CEST] <sza> Thank you very much, I'm trying it
[17:31:07 CEST] <furq> you can use yuv444p if you don't want to have to resize it, but a lot of players won't play the video back
[17:31:33 CEST] <sza> Just need to play it over vimeo. do you think it'll work?
[17:31:40 CEST] <furq> vimeo will reencode it anyway
[17:32:12 CEST] <sza> Thank you very much!
[17:32:19 CEST] <sza> I really appreciate it.
[17:33:13 CEST] <sza> The output file plays it very slow. How can I make it faster?
[17:33:31 CEST] <furq> change -framerate to something faster than 0.2
[17:34:17 CEST] <sza> Faster means larger number?
[17:34:32 CEST] <furq> yes
[17:34:44 CEST] <sza> Thank you
[17:35:03 CEST] <sza> again :)
[20:04:44 CEST] <pgorley> in the tool ffmpeg, when hwaccel_retrieve_data is called, does it mean that the frame is finished being decoded?
[20:08:46 CEST] <Psi-Jack> Hmmm
[20:09:12 CEST] <Psi-Jack> Is it possible, using ffmpeg, to convert flac media to m4a media, with more thorough metadata conversion? Right now, it seems to get like bare minimums, but not very comprehensive. Generally only enough to get a rough match with musicbrainz to be able to manually fix the rest of it.
[20:09:37 CEST] <Psi-Jack> I would LIKE to have it copy the musicbrainz fields over from vorbis (flac) to AAC (m4a), but it's only doing part of it.
[20:10:19 CEST] <kepstin> the musicbrainz fields in particular are kinda tricky, because you need to know the mapping in each format - it's different for each one
[20:10:33 CEST] <kepstin> i'd suggest just retagging the file with picard after transcoding
[20:11:45 CEST] <Psi-Jack> I've been doing that, but it's error prone, because from the origin format, flac, to the m4a format, picard often time picks the wrong actual metadata selection. :)
[20:13:04 CEST] <kepstin> hmm, right, the flac and m4a formats are different enough that ffmpeg's probably not copying the recording mbid field that it would need to do an exact match
[20:14:23 CEST] <kepstin> if you could use opus or vorbis instead, that would solve your issue; the tags in those are compatible with flac so it should just all copy over ;)
[20:15:02 CEST] <furq> why can't everything just use vorbis comments
[20:15:14 CEST] <Psi-Jack> Heh, here here!
[20:15:41 CEST] <kepstin> Psi-Jack: btw, if you're curious, https://picard.musicbrainz.org/docs/mappings/ shows what the different musicbrainz tags have to be converted to
[20:15:43 CEST] <Psi-Jack> m4a, though, has it's own dynamic format.
[20:15:49 CEST] <kepstin> I dunno if ffmpeg even knows how to write some of those
[20:16:18 CEST] <furq> yeah mp4 metadata is scarcely better than id3v2
[20:16:29 CEST] <furq> which is of course the gold standard of metadata
[20:16:36 CEST] <Psi-Jack> Yeah, vorbis uses musicbrainz_albumid I believe, and m4a uses "musicbrainz album id" with spaces.
[20:16:59 CEST] <Psi-Jack> furq: yeah, but still better. Hell, ANYTHING is better than id3 in every itteration of it thus far.
[20:17:04 CEST] <kepstin> "----:com.apple.iTunes:MusicBrainz Track Id" is the important one
[20:17:10 CEST] <kepstin> well, that and release id
[20:17:39 CEST] <furq> maybe mutagen will do it if you don't mind running a separate command
[20:17:49 CEST] <furq> i seem to remember it handling at least some mb tags
[20:18:11 CEST] <kepstin> mutagen on its own handles some, yeah. You might be able to use it to pull out the tags on the original file, and write them to the dest file?
[20:18:25 CEST] <furq> that's what i do when i batch convert
[20:18:35 CEST] <furq> although i don't batch convert with ffmpeg because it doesn't write mp3 xing headers
[20:19:05 CEST] <Psi-Jack> furq: Hmmm. How are you batch converting? I've been writing my library management scripts to do as much automated as possible, and so far, besides this one issue of metadata itself, it works great.
[20:19:06 CEST] <kepstin> i think i saw a patch for xing headers floating around the ML recently? would be nice to have that :/
[20:19:26 CEST] <Psi-Jack> Oh. You don't use ffmpeg.  i only convert FLAC->m4a. :)
[20:19:28 CEST] <furq> Psi-Jack: lame and then a mutagen script
[20:19:47 CEST] <Psi-Jack> And I use ffmpeg's libfdk_aac
[20:19:51 CEST] <kepstin> hmm. picard code has a *lot* of special handling for mp4/m4a; i'm not sure how much will be converted by mutagen alone
[20:21:00 CEST] <furq> i'd check but i've never run it on stuff tagged by picard
[20:21:30 CEST] Action: kepstin should check, but he never encodes to aac
[20:21:36 CEST] <furq> that too
[20:21:52 CEST] <Psi-Jack> hehe
[20:21:55 CEST] <furq> although if it can handle the id3 mappings it ought to be able to figure out the m4a ones
[20:22:10 CEST] <furq> assuming you use mp3 and not opus
[20:22:54 CEST] <Psi-Jack> As a media format, I absolutely despise mp3. Quality of m4a is much much better anyway, for a smaller size.
[20:23:15 CEST] <Psi-Jack> And my iDevices and iTunes support. ;)
[20:23:37 CEST] <kepstin> well, back when I used an iPod, I ran rockbox on it. Supported opus just fine ;)
[20:23:39 CEST] <furq> i only use mp3 for distributing to other people
[20:23:52 CEST] <furq> my mp3 player is full of vorbis, but i'd probably use opus these days
[20:23:54 CEST] <kepstin> anyways, https://github.com/quodlibet/mutagen/blob/master/mutagen/easymp4.py#L246 - looks like they do have the core musicbrainz tags in the easymp4 wrapper, so you should be good
[20:25:04 CEST] <kepstin> if you want *everything*, you'd have to retag in picard, but it preserves enough that picard can just look up the correct album right away without guessing.
[20:29:43 CEST] <Psi-Jack> My entire library was originally ripped into ogg vorbis. Now since a hurricane took out my entire library I've taken to maintaining a lossless (flac), format alongside the format I actually need to use. ;)
[20:30:16 CEST] <kepstin> yeah, I do the same. I encode both formats at the same time, and tag them separately.
[20:30:17 CEST] <furq> yeah disk space is cheap
[20:30:33 CEST] <furq> i only bother encoding if it's going on a portable device or something
[20:30:53 CEST] <Psi-Jack> I maintain both, so my tools helps synchronize everything. :)
[20:32:20 CEST] <Psi-Jack> I rip a new album to flac, run picard on it, merge it into my library, then use my tools to convert it to m4a, bring that m4a into my iTunes for Match, then I can access that song from anywhere, then my m4a's get merged into their own library, which Subsonic provides, because when I travel overseas, itunes match may not be available, sadly.
[20:33:03 CEST] Action: kepstin just puts his entire music collection in lossy formats on an sd card in his phone.
[20:33:19 CEST] <Psi-Jack> Heh.
[20:33:34 CEST] <furq> i just have it all on google play music
[20:33:42 CEST] <kepstin> nice that sandisk has that one 200gb card so everything actually fits :/
[20:33:56 CEST] <Psi-Jack> I have my phone download most of the time, since I get, with the Match service, AAC 256kbit, directly from Apple or their "Cloud" depending on if the match matched or uploaded.
[20:34:16 CEST] <kepstin> streaming/cloud stuff kinda sucks in canada, since all our cell phone data plans are expensive and capped :/
[20:35:01 CEST] <furq> generally if i don't have wifi then i'm on a train or something
[20:35:12 CEST] <furq> so my sd card is full of tv shows
[20:36:45 CEST] <furq> you'd think google would be happy to store 128k opus for users who know they can play it back, but nope
[20:36:50 CEST] <furq> it's all 320k cbr mp3
[20:37:46 CEST] <furq> not that i'm complaining about unlimited storage for my music of questionable provenance
[20:43:55 CEST] <Psi-Jack> kepstin: Yeah, mine's unlimited, even when I go to Japan, it's still unlimited data.
[20:46:56 CEST] <kepstin> Psi-Jack: in case you just need a simple tool to copy tags in your workflow, https://gist.github.com/kepstin/2c23f3ecc0a60f188273931527de5c12 should work from flac (or indeed any format using vorbistags) to aac
[20:47:21 CEST] <Psi-Jack> VEry nice.
[20:47:28 CEST] <Psi-Jack> I will check this out and see if it'll work for me.
[20:47:55 CEST] <Psi-Jack> And, thank you! :)
[20:48:21 CEST] <furq> #! /usr/bin/env python3
[20:48:22 CEST] <furq> pls
[20:51:43 CEST] <kepstin> if you insist :/
[20:55:12 CEST] <c_14> What about those 2 historical unixes that had env in /bin?
[20:55:44 CEST] <kepstin> if you manage to get python3 running on them, you already know how to fix the issue.
[20:58:57 CEST] <Psi-Jack> But that's not the issue. :)LOL
[21:00:38 CEST] <c_14> If you can't find enough problems in your life, sometimes you just have to make more. ;p
[21:05:54 CEST] <Psi-Jack> Heh, I should probably rewrite my library management scripts in python. Right now they're pure bash. ;)
[21:12:59 CEST] <Psi-Jack> Ooh, and python-mutagen is fairly current in EPEL for EL7. :)
[21:16:00 CEST] <Psi-Jack> But, of course, not Python 3.x
[21:17:39 CEST] <furq> it should work fine with python 2
[21:17:53 CEST] <furq> just get rid of the parens in print
[21:18:07 CEST] <Psi-Jack> It didn't, because of line 9. :)
[21:18:12 CEST] <furq> ^
[21:18:14 CEST] <Psi-Jack> Or that yes. hehe
[21:19:33 CEST] <kepstin> oh, just remove that ',file=...' bit I think, and it'll just print to stdout instead of err?
[21:19:42 CEST] <Psi-Jack> Correct.
[21:19:42 CEST] Action: kepstin doesn't do much python2 stuff now :/
[21:20:20 CEST] <Psi-Jack> print "Usage: {} src.flac dest.m4a".format(sys.argv[0]) , syntax error pointing at that " after m4a. grrr.
[21:20:57 CEST] <furq> replace {} with %s
[21:21:08 CEST] <furq> i think. i try my best to avoid python
[21:21:43 CEST] <Psi-Jack> {} is almost right, but {0} is correct for 2.
[21:21:51 CEST] <Psi-Jack> {} works too.
[21:22:00 CEST] <furq> oh yeah it's python with its ten million ways of doing string formatting
[21:22:06 CEST] <Psi-Jack> heh
[21:22:35 CEST] <kepstin> I rewrote https://gist.github.com/kepstin/2c23f3ecc0a60f188273931527de5c12 in python2 for you, and even tested it.
[21:23:07 CEST] <Psi-Jack> Heh that works too, %s, with % format method.
[21:23:08 CEST] <kepstin> oh, dropped the /usr/bin/env thing, oh well
[21:25:09 CEST] <Psi-Jack> So wierd...
[21:25:30 CEST] <Psi-Jack> So I changed it on my side, that still didn't work. Used the exact code you just gisted, worked. *shakes fist*
[21:25:54 CEST] <Psi-Jack> Oh, because I forgot to change python3 to python2
[21:26:15 CEST] <Psi-Jack> EPEL apparently has python 3.4. Heh
[21:26:37 CEST] <kepstin> yeah, i guess they just don't have the mutagen libraries for python3?
[21:26:46 CEST] <Psi-Jack> Nope.
[21:26:53 CEST] <Psi-Jack> And apparently no pip for it either. :/
[21:27:05 CEST] <Psi-Jack> Oh wait, there it is.
[21:27:11 CEST] <Psi-Jack> python3-pip, not python34-pip
[21:27:17 CEST] <furq> it should work fine with python 2 anyway
[21:27:23 CEST] <Psi-Jack> Yep.
[21:27:39 CEST] <Psi-Jack> I still write for python 2, but need to eventually upgrade my knowledge to 3.
[21:27:54 CEST] <Psi-Jack> but, maintain 2 obviously since even EL7 still has python 2
[21:28:17 CEST] <kepstin> I kinda like how the native asyncio stuff (added in 3.3, i think?) works.
[21:30:01 CEST] <Psi-Jack> Heh
[21:30:21 CEST] <Psi-Jack> And if I re-write my library management stuff in python, I can probably extract metadata information faster.
[21:30:31 CEST] <Psi-Jack> Right now, I'm extracting it using ffprobe and sed looping.
[22:03:48 CEST] <imperio_> hi, I followed instructions here: https://launchpad.net/~mc3man/+archive/ubuntu/trusty-media to add ppa
[22:03:59 CEST] <imperio_> however, it says it cannot find the address
[22:04:02 CEST] <imperio_> is it normal?
[22:05:05 CEST] <imperio_> my goal was to try to get 3.1 packages directly for travis
[23:29:32 CEST] <imperio_> to make it simple: how could I install ffmpeg 3.x on travis without building it please?
[23:32:40 CEST] <furq> imperio_: http://johnvansickle.com/ffmpeg/
[23:34:06 CEST] <imperio_> furq: Thanks
[23:34:21 CEST] <imperio_> however I have questions: it is compiled as static lib, right?
[23:34:24 CEST] <imperio_> (.a?)
[23:37:57 CEST] <imperio_> furq: ah sorry, you just provide the executable, I'm only interested in libavcodec, libavutil and libavformat :-/
[23:38:05 CEST] <imperio_> I should have precised
[23:38:40 CEST] <imperio_> but you have a libav link!
[23:38:44 CEST] <imperio_> yay \o/
[23:40:16 CEST] <imperio_> arf, doesn't provide what I'm looking for
[23:40:19 CEST] <imperio_> too bad :-/
[23:40:28 CEST] <imperio_> well, I'll just provide the one I have I guess
[23:45:40 CEST] <DHE> does your distribution already provide something like ffmpeg-dev ?
[23:46:31 CEST] <furq> i assume it's an old ubuntu
[23:46:52 CEST] <DHE> travis is the buildbot thing, right?
[23:47:17 CEST] <furq> something like that
[23:47:50 CEST] <furq> apparently it's locked to 12.04, so that's no use
[23:49:30 CEST] <llogan> imperio_: why can't you compile it?
[23:49:56 CEST] <imperio_> llogan: the goal is to provide it into the new mozilla browser engine
[23:50:15 CEST] <imperio_> their build process is already huge enough
[23:50:55 CEST] <imperio_> so i can totally compile it, but I don't want to
[23:56:15 CEST] <imperio_> btw, why the 3.x packages are only available on yakkety?
[23:56:19 CEST] <imperio_> that seems strange
[23:56:43 CEST] <furq> you'd have to ask the ubuntu packagers
[23:56:50 CEST] <imperio_> damn
[23:56:51 CEST] <furq> once they're done chasing the ball around the yard
[23:56:56 CEST] <imperio_> XD
[23:57:06 CEST] <imperio_> I thought I had to fight only with debian ones
[23:57:23 CEST] <imperio_> this is a damn curse
[23:57:46 CEST] <imperio_> I just thought: if i repackage ffmpeg myself, that would do the trick!
[00:00:00 CEST] --- Tue Jun 28 2016


More information about the Ffmpeg-devel-irc mailing list