[Ffmpeg-devel-irc] ffmpeg.log.20130112

burek burek021 at gmail.com
Sun Jan 13 02:05:02 CET 2013


[00:11] <Trashlord> now it's built again, this time with libfdk_aac
[00:15] <Trashlord> now it works
[00:17] <Trashlord> thanks for your help, llogan
[00:20] <Trashlord> now it says this for several files: [mp4 @ 0x25c1e40] track 0: could not find tag, codec not currently supported in container
[00:21] <Trashlord> it won't work for libvorbis either
[00:25] <Trashlord> right
[00:25] <Trashlord> turns out, because I want the output to be in mp4, I must convert the video as well, using libx264
[00:26] <Trashlord> so I'll leave it as AVI
[00:31] <Trashlord> well, thanks for all the help guys. got my shit sorted out
[00:37] <Navion> Anybody here a Serviio wizard?
[00:44] <llogan> fluid tripod heads suck when it is below freezing
[02:51] <Nano> is it ok to use the FFmpeg logo in my app, to visualize what its based on ?
[02:52] <llogan> Nano: you will have to ask herve.flores [at] free.fr
[02:53] <Nano> ok, why ?
[02:55] <beastd> llogan: depends probably on which logo he wants to use
[02:56] <llogan> yes, i assumed the current one on the site
[02:57] <beastd> llogan: the current one is the winter edition by muxson
[02:57] <llogan> ah...i mean the standard one
[02:57] <Nano> i'd like to use the zigzag logo
[02:59] <llogan> Nano: contact herve. the usage/license/whatever was never really finalized properly
[02:59] <Nano> ok
[03:00] <Nano> just thought that theres already a license type defined for the logo, like the CC, GPL or something...
[03:00] <Nano> but i will contact herve to be save before publishing my app
[03:01] <llogan> herve never put it under an standard license
[03:38] <savskii> hi
[03:38] <savskii> http://pastie.org/5671976
[03:54] <beastd> savskii: [ac3 @ 0x93a66e0] invalid bit rate
[03:56] <beastd> savskii: please note that the program you are using is *not* ffmpeg. it is just a program named ffmpeg from a fork. naturally we do not support it here. if you want to use ffmpeg look here http://ffmpeg.org/download.html#LinuxBuilds
[03:58] <savskii> ok thanks mate
[03:59] <savskii> is there any tutrial i can follow?
[03:59] <beastd> savskii: you should find lots of stuff
[04:00] <savskii> cheers
[04:00] <beastd> savskii:  you need to be clear on what you want to achieve. then you go on and choose proper multimedia codecs and container format
[05:51] <savskii> hell can anyone help me send a live stream please
[05:51] <savskii> ffmpeg -i Sample.mp4 -f rtsp -rtsp_transport tcp rtsp://85.17.169.209:3051/live.sdp
[05:52] <savskii> tryed that but no luck
[06:17] <barque> what is libstagefright?
[06:24] <barque> hello?
[06:33] <beastd> barque: IIRC it is hw decode acceleration from android platform.
[06:33] <barque> Hmm, so it kinda interfaces with the hidden OpenMAX?
[06:34] <barque> I mean, I've downloaded FFMpeg4Android
[06:34] <barque> haven't compiled it yet
[06:34] <barque> but it should provide HW accel on my android device?
[06:34] <beastd> maybe
[06:35] <beastd> at least hw accel on adroid device should be possible (given android can provide it for your device)
[06:36] <barque> I need to decode videos in JNI
[06:36] <barque> cause I need to support android 2.3.1
[06:36] <barque> and need the decoded stuff as textures
[06:36] <barque> I can't use mediaplayer, it does not provide byte arrays and such, and SurfaceTexture/external opengl texture stuff was added in API level 16
[06:37] <barque> MediaMetadataretriever only does thumbnails (decodes only 1 frame for every second)
[06:37] <barque> so I'm pretty much left with doing the decoding in JNI
[06:38] <barque> which pretty much only leaves ffmpeg
[06:38] <barque> and libavcodec
[06:38] <sware> does anyone here have a generated config.h from a recent windows build
[06:39] <beastd> sware: why can't you generate it yourself?
[06:39] <barque> possibly cause he doesn't have cygwin
[06:39] <sware> not sure the best way
[06:40] <barque> automake etc.
[06:40] <barque> autoconfig
[06:40] <sware> I just need it to see if any changes were made in the latest snapshot
[06:41] <beastd> barque: not sure if i can help you. but using ffmpeg libs on android should be possible. there were also some mails on ffmpeg-devel from some people making a dedicated project for using ffmpeg libs on android. maybe looking at their code may help you
[06:41] <beastd> barque: there is no autotools involved in ffmpeg
[06:41] <sware> beastd: I don't need to build, just need a config
[06:42] <beastd> sware: i still don't understand. also there are at least 4 different ways to build for windows
[06:42] <sware> beastd: it's for msvc version
[06:42] <beastd> sware: the config.h  is an internal file needed for the build and never exported directly. so i can't follow. what are you trying to accomplish?
[06:43] <beastd> sware: do you have a problem with your current build?
[06:43] <sware> beastd: about every 6 months i port ffmpeg to msvc, wanted to get a recent config.h to check for changes
[06:44] <beastd> sware: port? it doesn't work out of the box for you?
[06:44] <barque> beastd, yeah I'm currently using ffmpeg on android
[06:44] <barque> actual ffmpeg.so
[06:44] <barque> with swscaler.a
[06:44] <barque> it's just that, it's software decoding
[06:44] <barque> and it's rather slow-ish
[06:44] <barque> it can decode theora the fastest
[06:44] <barque> but it's very lackluster
[06:44] <beastd> barque: do you have a build with libstagefright enabled?
[06:44] <barque> 480p video at 21 FPS
[06:44] <barque> beastd, sadly no
[06:44] <barque> or I don't know actually
[06:44] <barque> but what do I need to check/ensure?
[06:45] <sware> beastd: I make it so ffmpeg can be built with visual studio
[06:45] <barque> unfortunately my tablet is also Tegra2
[06:45] <barque> which lacks NEON simd
[06:45] <barque> which makes it all the more... ARRRRGH-ish
[06:46] <beastd> sware: should work out of the box since a few months. please try it
[06:46] <sware> beastd: what do you mean ?
[06:46] <barque> thinking back, you gotta love every iPad/iPod platform, all come with NEON SIMD on the A4-5 chips
[06:46] <barque> so beastd, does libstagefright normally compile along with ffmpeg?
[06:46] <barque> or do I need to do something special?
[06:46] <beastd> sware: http://ffmpeg.org/platform.html#Microsoft-Visual-C_002b_002b
[06:47] <beastd> barque: did you cross-compile ffmpeg yourself for android?
[06:47] <barque> I downloaded one available on the net
[06:47] <barque> I can't find the link
[06:47] <barque> the guy had it all compiled already
[06:47] <barque> I tried compiling myself... I could not (for the love of me) get the configure script right
[06:47] <barque> nor could I use anyone else's
[06:48] <sware> ya I don't care about that tool heh
[06:48] <barque> something was up with my cygwin ndk compiler
[06:48] <barque> g++
[06:48] <beastd> barque: huh. ok :(
[06:48] <beastd> sware: you roll your own? with lots of code modifications?
[06:49] <sware> yes
[06:50] <barque> beastd, if libstagefright.cpp was included in libavcodec/ folder... chances are, I should have it compiled along with what was provided?
[06:50] <beastd> sware: ok. so you also know how to adress config.h to suit your needs and just need current one so you can work from that?
[06:50] <barque> if that is the case, this should be wicked sick
[06:50] <sware> beastd: yes
[06:52] <beastd> barque: you can try to find the decoder with a call to libavcodec
[06:53] <sware> beastd: the last build I did was june 18th. I'm assuming something have changed in the config since
[06:54] <beastd> barque: these are the other people that posted they work on ffmpeg for android: http://www.insignal.co.kr/wiki/index.php/FFmpeg_on_Android    (I do not know anything about there project, and it will maybe not be helpful to your problem.)
[06:54] <beastd> sware: for sure. lots of things
[06:58] <sware> oh well, guess I'll have to figure it out
[06:58] <sware> I have never been able to build on a windows system heh
[06:58] <sware> with mingw or anything else
[06:59] <beastd> sware: wait
[06:59] <sware> other than my ports, the last time purplepeter provided me one
[07:00] <barque> wait, when was OpenMAX and stagefright exposed as an API?
[07:00] <barque> Android 4.0?
[07:02] <beastd> sware: maybe this helps  http://pastebin.com/Y5Vr0qKi
[07:02] <beastd> sware: i generated it on mingw with --disable-yasm
[07:03] <beastd> barque: i have no special knowledge about android
[07:03] <sware> thanks beastd. I will cross reference that thanks
[07:10] <beastd> barque: you can e.g. test sth like this "codec = avcodec_find_decoder(AV_CODEC_ID_H264);" then look into codec->name if it is not "libstagefright_h264" then you are not using libstagefright
[11:25] <Trashlord> hey. just wanted to say I ended up ditching the AVI container. the sound was all choppy on my TV, unwatchable. for the past 9 and a half hours I've been converting all the videos to mp4 container format, with H264 video and AAC audio, that one works
[11:25] <Trashlord> it's gonna take at least another 10 - 12 hours to convert the rest of the files
[12:28] <elkng> is there difference of having option before or after option "-i file" ?
[12:32] <JEEB> elkng, yes
[12:32] <JEEB> before -i you set input/decoding settings
[12:32] <JEEB> after -i you set output/encoding settings
[12:34] <elkng> is that information in man page ?
[12:35] <elkng> its confusing, is there any other console software where order of options is matter ?
[12:35] <JEEB> I think there are, but don't ask me for an example right now :P generally you'd then have -input-options and -output-option separately
[12:36] <JEEB> and yes, it's in the manpage/documentation as far as i know
[12:36] <JEEB> because I think the possible manpage and friends are made of the same source as http://ffmpeg.org/ffmpeg.html#Synopsis
[12:36] <elkng> and how to recognize what are options depend on order ? like option -ss and others
[12:36] <JEEB> everything that you set before -i is gonna be done before decoding
[12:37] <JEEB> if it is a compressed stream
[12:37] <JEEB> for example -ss and -t set before -i will try to cut the stream so you can copy it over
[12:37] <JEEB> at GOP/frame lengths
[12:37] <elkng> and what that command could mean: "-ss 10:0 -f file.mp4 -ss 20:0 output.avi" ?
[12:38] <Fjorgynn> no
[12:38] <Fjorgynn> isn't it -t output?
[12:38] <Fjorgynn> aha
[12:39] <JEEB> elkng, that seems invalid to me
[12:39] <JEEB> -f mp4 is valid, -f file.mp4 is not valid because -f is format
[12:40] <JEEB> and that most probably would be /after/ -i
[12:40] <JEEB> it would seek to whatever you mean with 10:0, set output format to mp4, set ss again (!?) and then write mp4 container stuff into a file output.avi
[12:50] <elkng> JEEB: I mean "-i" not "-f"
[12:51] <elkng> the combination "-ss 10:0 -i file.mp4 -ss 20:0"
[12:51] <elkng> or "-ss 20:0 -i file.mp4 -ss 10:0"
[12:52] <JEEB> so it would first seek as well as possible to whatever you mean with 20:0 by the audio frames and video GOPs, then it would start decoding and seeking to the point that you note as 20:0
[12:53] <elkng> its tricky
[12:53] <JEEB> as long as you know that coded audio consists of audio frames and that coded video has GOPs
[12:54] <JEEB> you will expect that if you have both video and audio, that before decoding starts seek won't be exact
[12:54] <JEEB> with just audio, the audio frames are small enough that you usually get close enough
[12:54] <elkng> more convinient was to have something like these options: "-i file.mp4 --start-position 0:20:0 --end-position 0:30:0"
[12:54] <JEEB> of course, lossless formats like FLAC or wavpack might disagree
[12:55] <JEEB> uhhh
[12:55] <JEEB> you didn't do that
[12:55] <JEEB> and if you want that you just put -ss and -t /after/ -i
[12:55] <JEEB> -ss is seek, and -t is how long you want to keep decoding in timestamps
[12:56] <elkng> JEEB: "/after/ -i" that way instead of seeking in file it will decode input video until that mark and it will be much longer
[12:56] <JEEB> naturally
[12:57] <JEEB> but do remember that unless you actually checked where the GOPs start in the file
[12:57] <JEEB> you can't use before -i either too easily :P
[13:01] <Fjorgynn> what does -absf aac_adtstoasc do?
[13:02] <Abhijit> hey guys
[13:02] <JEEB> audio bit stream filter that converts ADTS AAC to ASC
[13:02] <Abhijit> is libavg.de down for for everyone or just me?
[13:02] <Fjorgynn> JEEB: which means what+
[13:02] <JEEB> Fjorgynn, there are two types of handling AAC inside things?
[13:02] <JEEB> just like there's that mp4 type H.264 and Annex B H.264
[13:02] <Fjorgynn> okay
[13:03] <JEEB> I don't think those changes should always be visible to the user tho, at times you just have to add the bit stream filter manually when using ffmpeg :P
[13:04] <Fjorgynn> JEEB: a service I use to "save videos" has that command
[13:10] <Fjorgynn> [mp4 @ 04B95620] malformated aac bitstream, use -absf aac_adtstoasc
[13:10] <Fjorgynn> av_interleaved_write_frame(): Operation not permitted
[13:10] <Fjorgynn> that's why
[13:15] <JEEB> yes, so mp4 container wants ASC
[13:16] <JEEB> and your input is ADTS
[13:16] <JEEB> don't ask me why ffmpeg doesn't do this automagically :D
[13:44] <elkng> thats an interesting question why ffmpeg doesn't do this automagically
[13:45] <Mavrik_> iirc there's no common logic to turn on required bitstream filters
[13:46] <Mavrik> so muxer dies :)
[13:48] <JEEB> and it isn't really an interesting task so it's mostly a case of "watches pelcome"
[13:53] <Mavrik> indeed
[14:06] <Abhijit> hey guys. the freecal game is just disappeared from the internet. i can not find any reference anywhere? am i missing something?
[14:06] <Abhijit> sorry wrong channnel
[14:55] <Fjorgynn> added vlc to path <3
[16:45] <espr3ss0> hi
[16:45] Action: espr3ss0 waves
[16:45] <espr3ss0> anyone know how to set / fix, "stream 0, timescale not set" ?
[16:46] <espr3ss0> is there a "timescale" flag ?
[16:46] <espr3ss0> although it might not be the problem :-/
[16:47] <espr3ss0> I have 2 mp4's and they both give the "timescale not set" problem using ffmpeg, but one can't skip (when played) further the about 10 mins
[16:48] <espr3ss0> i'm counting on ffmpeg coz it should be able to do it ! ;-)
[16:51] <JEEB> can you compile L-SMASH and see what boxdumper --timestamp that_file.mp4 > timestamps.txt gives?
[16:51] <JEEB> http://code.google.com/p/l-smash/
[16:51] <JEEB> l-smash
[16:51] <espr3ss0> i could do !
[16:51] <JEEB> l-smash should be simple to compile, ./configure and then make
[16:52] <espr3ss0> after that using ffmpeg how would one fix it ? o_O
[16:52] <JEEB> uhh
[16:52] <JEEB> boxdumper is to get info on how the file is created
[16:53] <JEEB> if you would do boxdumper file.mp4 > all_box_data.txt you would get a file full of the whole file's structure
[16:53] <JEEB> in a more readable format
[16:53] <JEEB> if boxdumper can find a timescale for the file, and can output timestamps on that timescale, then ffmpeg has a problem somewhere
[16:53] <JEEB> if boxdumper can't find it either
[16:53] <espr3ss0> hmm, might have fixed it via an mkv version
[16:53] <JEEB> then we have a problem
[16:54] <JEEB> well, the problem is still interesting :P
[16:55] <espr3ss0> :-)
[16:57] <JEEB> because if there really is no timescale then there's really no way of finding the timestamps for that track
[16:57] <JEEB> because timestamps are X/Y
[16:57] <JEEB> where Y is timescale
[16:57] <JEEB> timescale = ticks for one second
[16:58] <espr3ss0> thanks for the heads-up
[16:58] <JEEB> how does the boxdumper --timestamp output look?
[16:58] <JEEB> it should output timestamps for all tracks in the file
[16:59] <espr3ss0> i will get round to installing it ! ;-p have stuff to do :-(
[17:00] <JEEB> no need to install it really
[17:00] <JEEB> but yeah, compile it when you have time :)
[17:00] <JEEB> it's a nice mp4-style container implementation
[19:49] <cha0s_> is anyone available for a little help in troubleshooting and possiblly correcting an ffmpeg/winff issue?
[20:43] <skifreak> hey all, having some trouble with a bluray file, specifically decoding with ffmpeg while playing. When I ripped the file via passthrough I didn't receive any errors. I think my problem due to the block code look-up failure, just trying to figure out where to start http://dpaste.com/868741/   current ffmpeg version 1.0.1
[21:32] <CStanners> Does anyone know what "Marker bit missing in complexity estimation part 1" is about?
[22:11] <elkng> I noticed many videos of films of good quality and in size of less than 1GB but when I try to achieve that size I end up with quiet bad quality, what options those videso could have been encoded with ?
[22:14] <lectrick> Anyone know a strategy to avoid "rb buffer underflow" and general buffer underflow errors when transcoding a 720p MKV file to a dvd vob?
[22:20] <lectrick> The other thing is, I am trying a multipass strategy in order to perhaps help mitigate the underflow issue. The first pass creates the pass log file but writes 0 bytes to it when it's done :O
[22:20] <cbsrobot> lectrick: what version of ffmpeg are you using ?
[22:20] <cbsrobot> there was a bug from sept. till nov. 2012 &.
[22:21] <cbsrobot> env.
[22:21] <cbsrobot> elkng: depends on what codec, size etc. you want to use
[22:21] <cbsrobot> it's kinda hard to help you with so little informations
[22:24] <elkng> cbsrobot: it should depend on codec and not on some tricky ffmpeg's options ?
[22:26] <lectrick> cbsrobot: ffmpeg version 1.0 Copyright (c) 2000-2012 the FFmpeg developers
[22:26] <cbsrobot> lectrick: try to upgrade
[22:26] <cbsrobot> September, 28, 2012, FFmpeg 1.0
[22:27] <lectrick> cbsrobot: I am using a managed tool maintainer (homebrew on OS X in this case), and someone else is responsible for updating the repo with later ffmpeg versions. I could build it manually I suppose but this is probably a pain
[22:27] <cbsrobot> lectrick: you have xtools installed ?
[22:27] <lectrick> you know what? I haven't updated homebrew in a while, let me see if it has updated repos
[22:28] <cbsrobot> sounds rasonable
[22:28] <lectrick> oh shit, it's going to update to 1.1
[22:28] <cbsrobot> else: http://ffmpeg.org/trac/ffmpeg/wiki/MacOSXCompilationGuide
[22:28] <lectrick> well that's a prerequisite I can do lol
[22:28] <lectrick> stand by.
[22:29] <cbsrobot> elkng: well it depends on a lot of things ...
[22:29] <elkng> cbsrobot: wheather, pressure ?
[22:30] <cbsrobot> elkng: sure use -weather and -pressure options to set them correctly
[22:31] <elkng> it was pun
[22:31] <cbsrobot> just make sure to play them in similar conditions or the files will have bad quality & :-)
[23:08] <lectrick> cbsrobot: after an extensive set of updates and ultimately upgrading ffmpeg from 1.0 to 1.1, and retrying the same command, the buffer underflow errors seem to be gone :O
[23:09] <lectrick> i guess my last question is, how do I create a dvd ISO given a .vob file? :)
[23:09] <sacarasc> Use something like dvdauthor.
[23:10] <lectrick> yup, a/v plays fine in my 7 minute test output sample :)
[23:12] <lectrick> sacarasc: i have dvdauthor, i will look into how to use it, but if anyone knew the command which just took a .vob file and spit out a menuless .iso that would be wonderful haha
[23:19] <cbsrobot> I think you'll need mkisofs or similar to create the iso
[23:25] <lectrick> yeah i'm installing that now too... fun afternoon figuring out how to do all this lol
[00:00] --- Sun Jan 13 2013


More information about the Ffmpeg-devel-irc mailing list