[Ffmpeg-devel-irc] ffmpeg.log.20140301

burek burek021 at gmail.com
Sun Mar 2 02:05:01 CET 2014


[00:14] <llogan> luc4: what was the file name? sorry, i was afk when you messaged me.
[00:15] <luc4> llogan: no problem. VID_20131212_123011_5s.mp4.
[00:16] <llogan> mgeary: that's an ancient build. do yourself a favor and get a newer one.
[00:16] <mgeary> hmm. okay, thanks!
[00:17] <llogan> luc4: i have it (video of phone being plugged in). what was the problem again?
[00:17] <llogan> mgeary: or if you like to compile  http://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide
[00:17] <average> is there a build script for these static builds ?
[00:18] <average> I'd like to know more about static building and I'd like to read those static-build-scripts
[00:18] <llogan> average: you mean to make your own?
[00:18] <luc4> llogan: precisely. ffprobe was reporting a certain frame rate. Trying to transcode with ffmpeg resulted in a strange warning message reporting that frame rate was too high for the mp4 format.
[00:18] <average> I would like to study one of those scripts
[00:18] <average> ...
[00:18] <llogan> average: ask burek or relaxed
[00:18] <average> relaxed: hey, can I study your static-build-script ?
[00:19] <llogan> i know you probably provided this before but i can't remember details
[00:19] <luc4> llogan: no problem, let me collect the data.
[00:20] <average> how can I leave an offline ?
[00:20] <average> to burek
[00:20] <average> llogan: ^^ ?
[00:21] <llogan> average: you can email him. address is in the first link from fflogger
[00:23] <luc4> llogan: http://paste.kde.org/pxm0e9hy4
[00:23] <llogan> manu1083: what is avconvert?
[00:24] <luc4> llogan: also it seems to take forever to transcode a very small file.
[00:27] <llogan> luc4: try a more recent build. i don't get those messages.
[00:28] <manu1083> llogan : Avconv is a fork of FFmpeg
[00:28] <llogan> so why are you asking about it here?
[00:29] <manu1083> http://libav.org/avconv.html
[00:29] <luc4> llogan: which means 2.1.4? Also consider I'm on mac and I built using macports... maybe that is the difference...
[00:29] <llogan> manu1083: shouldn't you be asking question about the fork at the forks IRC?
[00:30] <llogan> luc4: test with one of these https://ffmpeg.org/download.html#MacOSXBuilds
[00:30] <luc4> llogan: ok, thanks
[00:31] <llogan> users should use ffmpeg from current git master. releases are generally for distributors and such.
[01:05] <mgeary> if i just do ffmpeg -i foo.mp4 -t 30 bar.mp4, will it use the same encoding settings as on the original? it doesn't look like it does. Is there a way to preserve those?
[01:07] <mgeary> does --copy work alongside -t ?
[01:10] <Hello71> --copy isn't a thing
[01:10] <llogan> mgeary: you can stream copy with -codec copy. you can use -t with it too.
[01:11] <llogan> https://ffmpeg.org/ffmpeg.html#Stream-copy
[01:11] <mgeary> yeah, i saw that, but it wasn't clear how it would like -t
[01:11] <mgeary> thanks!
[01:13] <Hello71> I think some codecs don't support it (?)
[02:04] <andrew_46> llogan: The people you meet :)
[03:17] <bparker> I recently upgraded ffmpeg to 2.1.3 and now one of my programs is crashing inside of libavformat: mux.c:833 av_freep(&s->streams[i]->priv_data);
[03:18] <bparker> I don't use/set priv_data so I have no idea why it's trying to free it
[03:21] <bparker> sorry, I used ffmpeg 2.1.4 not 2.1.3
[03:26] <bparker> I only use libavformat to mux existing h264 data to mp4 container
[04:14] <thebombzen_> bparker: did you try using the latest git-master? that bug might be fixed since 2.1.5
[07:04] <rjp421> is there a way to explicitly include line breaks (line \n or \r) in the output?
[07:28] <juanmabc> so, are we clear this is ffmpeg channel?
[07:30] <juanmabc> oh, encoding steps (mess), go on ;)
[07:44] <rjp421> *(like \n or \r), oops
[08:38] <haspor> hello, i try to compile a shared lib for android by using 5 static ffmpeg libs + headers, i have also included test.cpp that calls av_register_all() function but i get error undefined reference to av_register_all()
[08:38] <haspor> could someone help a bit :)
[08:46] <haspor> here is the android.mk file im using: http://pastebin.com/DsCX0U12
[11:20] <jammanbo> Noob in a bit of a crisis here. I'm trying to capture a live stream from an m3u8 resource but it's not working so well, I think because the stream is flakey. It'll run for as long as I leave it, but typically only captures the first few minutes. Any tips? Command is $ ffmpeg -i "http://&" -c copy file.ts
[11:22] <jammanbo> (If it illicits any pity help I'm trying to record a wedding ceremony from the churches live feed because the couple cant afford someone to video it)
[11:48] <rsdrsdrsd> if i use yadif=0:-1:1 does it only deinterlace videos which are interlaced?
[11:52] <relaxed> rsdrsdrsd: if they're marked as interlaced
[12:08] <rsdrsdrsd> otherwise nothing happens?
[12:13] <rsdrsdrsd> if a video has different dar and sar, how can i create a screenshot with a fixed height and proportional aspect ratio
[12:14] <rsdrsdrsd> I have scale=dar*480:480
[12:14] <rsdrsdrsd> is that ok or should i use sar
[12:14] <rsdrsdrsd> using scale=-1:480 doesn't give the expected results as mentioned in the docs
[12:43] <relaxed> rsdrsdrsd: right, I guess it's possible for interlaced content to not be flagged as such.
[12:45] <relaxed> rsdrsdrsd: I think -filter:v "scale=640:ih*ow/sar/iw" will do it
[13:21] <haspor> anyone could help with the android problem i mentioned above?
[14:38] <luc4> Hello! Anyone who can do a test on mac?
[15:28] <rsdrsdrsd> Is this a good scaling option: trunc(ceil(dar*720)/2)*2:720
[15:28] <rsdrsdrsd> so that the aspect ratio reamains
[15:28] <rsdrsdrsd> have to encode to x264
[15:30] <rsdrsdrsd> anyone? I want to use it in an autonmatic processing script
[15:30] <Achilles> Hello. I'm trying to compile ffmpeg from source but the make fails on libxvid.c
[15:30] <klaxa> rsdrsdrsd: you can use -1 to automatically calculate the correct scaling to keep the aspect ratio
[15:30] <klaxa> as in -1:720
[15:30] <klaxa> however, that might result in odd numbers for the width...
[15:30] <rsdrsdrsd> but for xvid is tahs to be divisable by 2?>
[15:31] <klaxa> Achilles: pastebin your config.log
[15:31] <rsdrsdrsd> that is why i am looking for a good solutions
[15:31] <Achilles> I'm on mobile right now :(
[15:31] <klaxa> rsdrsdrsd, just try out a few resolutions
[15:31] <rsdrsdrsd> I have done that
[15:32] <klaxa> does it work reliably?
[15:32] <JEEB> I usually use sar, but that kind of seems to be OK
[15:32] <JEEB> although you probably will want to switch the 2
[15:32] <rsdrsdrsd> why is that?
[15:32] <JEEB> to width subsampling
[15:32] <JEEB> although wait...
[15:33] <rsdrsdrsd> so trunc(ceil(sar*720)/2)*2:720 is better
[15:33] <JEEB> where's my similar thingamajig
[15:35] <JEEB> trunc(sar*iw/(ih/oh)/hsub)*hsub:trunc(min(ih\,272)/vsub)*vsub
[15:36] <JEEB> although that has that min thing that you prolly don't need
[15:37] <JEEB> the width part is probably most related to you
[15:37] <rsdrsdrsd> if I wnat to have a steady height
[15:37] <JEEB> just look at the width part
[15:39] <rsdrsdrsd> should this be it: trunc(sar*iw/(ih/480)/hsub)*hsub:480
[15:40] <JEEB> well, if your height is set to 720 :P
[15:40] <JEEB> then you can just use oh
[15:40] <JEEB> trunc(sar*iw/(ih/oh)/hsub)*hsub
[15:40] <JEEB> this should give you SAR-correct width with the subsampling limits kept
[15:41] <rsdrsdrsd> trunc(sar*iw/(ih/oh)/hsub)*hsub:720
[15:41] <rsdrsdrsd> don't i get a refrence error
[15:41] <JEEB> you shouldn't
[15:42] <rsdrsdrsd> ok let's try it
[15:52] <rsdrsdrsd> sar of origal is 4:3 (1440x1080) and dar is 16:9  but the encoded video is 1278x720 [SAR 1:1 DAR 71:40] shouldn't it be 1280x720 ?
[15:52] <Achilles> Here is my config log http://pastebin.com/fAy59hKK
[15:53] <Achilles> klaxa
[15:55] <klaxa> what is the error you get while compiling?
[15:57] <Achilles> Www.Pastebin.com/nN7bUkxR
[16:01] <rsdrsdrsd> isn't dar = sar*iw/ih
[16:01] <Achilles_> This is Achilles. klaxa, any idea why I'm seeing this fatal error?
[16:01] <klaxa> actually no, not really
[16:02] <klaxa> do you need xvid? if not you could just disable it
[16:02] <klaxa> from when is your source? is it the latest git?
[16:02] <Achilles> latest git. yes.
[16:03] <klaxa> is your libxvid the latest version?
[16:03] <Achilles> yes, latest libxvidcore4
[16:04] <Achilles> -rw-r--r--   1 root root  1025508 Mar  1 08:12 libxvidcore.a
[16:04] <Achilles> -rw-r--r--   1 root root   747784 Mar  1 08:12 libxvidcore.so.4.3
[16:06] <rsdrsdrsd> i think trunc(ceil(dar*oh)/hsub)*hsub is better?
[16:06] <klaxa> i'm going to try to compile with your configuration
[16:07] <Achilles> ok. I noticed something though.. I have libxvid installed to /usr/lib/
[16:08] <Achilles> whereas in my compilation command I specify --prefix=/usr/local/hgffmpeg  and --extra-ldflags=-L/usr/local/hgffmpeg/lib
[16:08] <Achilles> I have libxvidcore.so.4.2 there
[16:09] <klaxa> yeah i thought that too
[16:09] <klaxa> maybe i should have asked
[16:13] <Achilles> should I compile without --extra-ldflags ?
[16:13] <klaxa> i removed your extra flags and it seems to compile
[16:13] <Achilles> aha
[16:13] <klaxa> depends, do you have extra libraries in that directory that are needed for the ffmpeg build?
[16:14] <Achilles> I need to compile ffmpeg with some old libraries (installed with the original build) but replace some of them with newer versions (like mp3 lame, x264, libxvid)
[16:15] <klaxa> i think those should actually be taken from the extra-cflags and extra-ldflags directories
[16:16] <Achilles> i'm new to compiling from source on linux, but I'll look it up
[16:17] <klaxa> your configuration compiled on my end though, maybe try with the old libraries first to see if that works
[16:17] <Achilles> ok
[16:23] <Achilles> thanks klaxa
[17:41] <rsdrsdrsd> oh no it doesnt matter, thnx JEEB
[19:30] <geqsaw> hi, is there an easy method to convert a 3d side by side video in a video with alternating frames? The stereo3d filter only supports anaglyph and interleaved output.
[20:16] <blight> hi guys
[20:36] <bparker> I have a capture program that encodes live video and muxes it to a file. I'm using libx264 to encode and libavformat to mux it into mp4 container. When playing back the video in VLC, I get a lot of errors and the video is mostly garbled after a few seconds: http://dpaste.com/1674961/   any idea what could be wrong?
[20:37] <bparker> geqsaw: why would you want to do that though? frame sequential is not a commonly supported input method for players/displays
[20:37] <bparker> plus you're just losing resolution by scaling it again
[20:40] <Mavrik> bparker, your input buffer gets overrun?
[20:41] <bparker> Mavrik: why do you say that?
[20:42] <bparker> pretty sure it doesn't
[20:43] <bparker> it was working fine on 1080p60 source I tested earlier, then I changed to 1080i60 and started having problems
[20:44] <blight> did you run valgrind on it? ;)
[20:44] <Mavrik> bparker, because that's usually what causes your issue
[20:44] <blight> someone around who knows libswresample? i think i have an unusual use case for it and i wonder how to make it work ;)
[20:52] <bparker> blight: asan
[21:04] <TylerMGF> hey guys, I have a quick question. When encoding from 5.1 to 2.0, I get a muffled sound, every time. I assume that's because it gets 2 audio sources and discards the rest. how can I make ffmpeg combine all the streams into 2?
[21:36] <klaxa> TylerMGF look for downmixing for example here: http://trac.ffmpeg.org/wiki/AudioChannelManipulation or here: http://www.ffmpeg.org/ffmpeg-filters.html#channelmap
[21:38] <TylerMGF> thank you, klaxa. running a build of ffmpeg without libfdk-aac because I'm on windows and too lazy to compile, so I just encoded it with vorbis @ 600kbps (stereo, yeah, I know) and it's not muffled anymore
[21:38] <klaxa> oh yeah to downmix you have to re-encode
[21:39] <klaxa> you can't combine audio-tracks without encoding
[23:32] <geqsaw> bparker: my TV does not support the higher resolution of side by side video.
[23:43] <cbsrobot> geqsaw: can't you combine stereo3d and framepack ?
[23:46] <geqsaw> cbsrobot: thanks, I'll try it out
[23:49] <cbsrobot> maybe: -vf [0:v]stereo3d=in=sbs1:out=ml[left],[0:v]stereo3d=in=sbs1:out=mr[right],[left][right]framepack=frameseq
[23:49] <cbsrobot> ^not tested
[23:50] <cbsrobot> geqsaw: wait
[23:50] <cbsrobot> I just read on the stereo3d manual:
[23:51] <cbsrobot> for the out parameter:
[23:51] <cbsrobot> Available values for output image formats are all the input formats as well as:
[23:51] <cbsrobot> so try: -vf stereo3d=in=sbs1:out=al
[23:52] <geqsaw> someone should update the official documentation in this point
[23:53] <cbsrobot> it's there:
[23:53] <cbsrobot> http://www.ffmpeg.org/ffmpeg-all.html#stereo3d
[23:53] <cbsrobot> just 2 lines below 'out'
[23:55] <orangey> hello all!
[23:55] <geqsaw> oh
[23:56] <geqsaw> I'm so blind...
[23:56] <orangey> I'm trying to fix a video file that I'm pulling off of an SD card (recovery attempt).
[23:56] <orangey> however, I try avconv -i broken.m4v fixed.mp4 and get: http://pastebin.com/yRHYCkrh
[23:56] <orangey> it appears that the "header" is correctly picked up
[23:56] <orangey> as is the size, creation time, etc.
[23:56] <klaxa> for avconv support please visit #libav
[23:56] <orangey> klaxa:
[23:56] <orangey> aha.
[23:57] <orangey> apologies. I thought they were interchangeable
[23:57] <orangey> For what it's worth, I tried ffmpeg first and it didn't work either, failing in exactly the same awy
[23:57] <iive> that's expected. try ffplay and seeking...
[23:58] <orangey> iive: Sounds like I don't even have ffmpeg installed like I thought.
[23:59] <iive> you may also try to make a gzip-ed copy of the file. video is already compressed so it cannot be shrinked further. but lots of zeros compress a lot more better.
[23:59] <orangey> iive: is that to me?
[00:00] --- Sun Mar  2 2014


More information about the Ffmpeg-devel-irc mailing list