[Ffmpeg-devel-irc] ffmpeg.log.20120420

[burek]

[00:00] <burek> your input stream might be damaged, or we are doing something wrong
[00:00] <burek> try ffplay /dev/dvb/adapter0/dvr0
[00:00] <burek> do you see video normally
[00:00] <taypen> burek, i suppose audio codec should be AAC and video H.264
[00:00] <burek> taypen, try typing just ffmpeg -i input.mkv
[00:01] <burek> and use pastebin to copy the output
[00:01] <burek> so we can tell you exactly
[00:01] <xandy> burek i've updated ffmpeg now and i'll try it now again with the defaults
[00:01] <burek> ok
[00:02] <loo> burek, that's a server machine, so I cant do that
[00:02] <loo> if I cat that stream
[00:02] <burek> i see
[00:02] <loo> I can open it on my windows machine
[00:02] <burek> if you cat it to a test.ts file
[00:02] <loo> burek, sample - http://beberry.lv/up/u/sample
[00:02] <burek> and then use ffmpeg -i test.ts -vcodec copy -acodec copy test2.ts
[00:02] <burek> would it be the same size?
[00:02] <burek> filesize*
[00:03] <loo> will try
[00:03] <taypen> burek, http://pastebin.com/aekXpzsY
[00:04] <burek> Stream #0.0: Video: h264 (High)
[00:04] <burek> and 2 ac3 audios
[00:04] <burek> eng/rus so I guess you'll have to convert it to aac
[00:04] <burek> but try only with that remux command
[00:04] <loo> burek, outcome is greater in size
[00:04] <burek> and see if that works
[00:05] <taypen> burek, ok thanks
[00:05] <loo> burek, were you saying that to me?
[00:05] <xandy> size= 3188416kB time=309096:19:12.00 bitrate=   0.0kbits/s
[00:05] <xandy> hm didn't worl
[00:05] <xandy> l
[00:05] <basilgohar> burek: http://paste.basilgohar.com/view/86958547
[00:05] <taypen> burek, but in case i'll have to convert the audio, how i do that?
[00:05] <burek> taypen, if it doesn't, then try: ffmpeg -i input.mkv -acodec aac -strict experimental -ar 44100 -ab 96k -ac 2 -vcodec copy -scodec copy out.mp4
[00:06] <burek> loo, if test.ts and test2.ts don't match at filesize, then test.ts is not a valid mpegts file I guess
[00:06] <burek> the test was to see if the -vcodec copy would produce the same thing (which it should)
[00:07] <burek> but if it doesn't that means something is not right with the input I guess
[00:07] <loo> burek, the test2 is greater,, Burek, can you take a look at the sample file, please?
[00:07] <burek> you can try uploading a sample to the ffmpeg's ftp
[00:07] <burek> for developers to investigate
[00:07] <burek> and submit a bug report, pointing to that sample
[00:07] <burek> read that loo
[00:08] <loo> ah, I added a ticket, and added the link to sample, but no response yet. I'm just burning, because I can't solve this for a quite  a long time..
[00:08] <xandy> burek tryed ffmpeg -i video.flv -acodec copy -vn audio.flv now and got [flv @ 0x21e39a0] Application provided invalid, non monotonically increasing dts to muxer in stream 0: 461352 >= 461306
[00:08] <loo> I assume, that you are more engaged in that community, so you could maybe post something there
[00:09] <burek> basilgohar, the problem with simple concatenation of video streams is that it works mostly with mpg streams
[00:09] <burek> all the other streams have some problems with it (if you just use cat)
[00:09] <burek> and mpegts container is such a container that it breaks the stream into packets
[00:10] <burek> so maybe that's why all the packets played back correctly, but when you use some format like mkv
[00:10] <burek> then those concatenated streams make problems
[00:10] <burek> try using some simple video editor to first concatenate those streams into one
[00:11] <burek> and then use ffmpeg to encode it (or remux) to whatever you like
[00:11] <burek> loo, I know the feeling :(
[00:11] <burek> but try to be patient
[00:11] <burek> no report is being ignored
[00:11] <burek> it's just the lack of free time to manage all of them
[00:11] <burek> also, update your report with any new investigations that you do
[00:12] <burek> which might help speed up its resolving
[00:12] <burek> xandy
[00:12] <burek> can you please use pastebin.com, to show your command line and its output?
[00:12] <xandy> ok
[00:12] <loo> burek, ok, but this is too frustrating, a feeling that no one cares.. ++ can I add as a new post, or should I edit old ones?
[00:14] <burek> dizzylizzy, if you yourself don't stretch/extend it, then you're ok with that logic :) just extract, make changes (and make sure that duration is about the same) and then just mux it back together
[00:14] <burek> loo, if it is related to an old one, then update
[00:14] <burek> and leave the link here
[00:14] <taypen> burek, is there any difference between m4v and mp4?
[00:15] <loo> burek,ok, thank you, I'm tired, have to go. thanks again!
[00:15] <burek> m4v is a video only and mp4 can contain audio too
[00:16] <taypen> hmm, the movies i downloaded for ipad only ends with m4v
[00:16] <xandy> burek ffmpeg -i video.flv -acodec copy -vn audio.flv -> http://pastebin.com/DJyEmJUj
[00:17] <burek> well then use ffmpeg -i ... out.m4v
[00:18] <taypen> ok :)
[00:18] <burek> hm, xandy try just unpacking your audio, to see if it is not broken
[00:18] <burek> like this
[00:18] <burek> ffmpeg -i video.flv out.wav
[00:20] <burek> xandy, your ffmpeg is still very old
[00:20] <xandy> :(
[00:20] <burek> 0.10.x is now something like latest
[00:20] <burek> can you add debianmultimedia repos
[00:20] <burek> and just install ffmpeg from there?
[00:21] <burek> http://debian-multimedia.org/
[00:26] <xandy> burek thats what i did but the machine is squeeze
[00:27] <burek> xandy, did you try just extracting the audio to wav
[00:27] <xandy> yea. my tool still says it's broken but i'll try it now with vlc again
[00:27] <burek> try extracting it with ffmpeg
[00:28] <burek> and if it stops at the same point in time
[00:28] <burek> 00:07:33.31
[00:28] <burek> then the input audio is broken I guess
[00:50] <xandy> burek thanks for your help, i think ill give it a try again tomorrow
[00:50] <burek> ok :)
[00:50] <burek> :beer:
[01:34] <biscii> does anyone know the best command to use for ffmpeg (mp3) -> segments for audio only ?
[01:34] <biscii> I've found a few segmenting examples
[01:34] <biscii> but its mostly for video
[01:50] <saste> biscii: try the segmenter
[01:51] <biscii> the apple seqmenter ?
[01:51] <biscii> or ion cannon one ?
[01:52] <saste> biscii: no the native ffmpeg segmenter muxer
[01:52] <saste> also check this: http://gitorious.org/~saste/ffmpeg/sastes-ffmpeg/commits/misc-segment-fixes-20120206
[01:52] <saste> since the current implementation may be borken with audio only
[01:52] <saste> feel free to report to me personally if the branch doesn't work
[01:52] <saste> i want to merge it soon anyway
[01:52] <biscii> cool will do. Thanks
[02:07] <basilgohar> burek: Thanks for getting back to me.  I'll try with different containers and see what I find.
[03:46] <taqattack> when encoding to x264. My threads is set to 6 but I only have 4 cores. Does this hinder my encoding performance?
[04:09] <dan3son> Hi
[04:10] <dan3son> I like how your topic mentions x264 encoding, I'm at videolan's x264 codec website, can't find any documentation describing tune/preset/profile
[04:11] <dan3son> I just want to transcode all my videos so they'll play on my TV w/out on-the-fly transcoding
[04:18] <taqattack> This link helped me a lot: http://mewiki.project357.com/wiki/X264_Settings
[04:18] <dan3son> I can't seem to find the plethora of x264 presets either, such as 'hq'
[04:18] <dan3son> My copy of ffmpeg only has two, and they're for ipods
[04:22] <taqattack> Which build are you using?
[04:23] <dan3son> windows, N-39877-g4fa706a
[04:23] <dan3son> The latest windows build, essentially
[04:24] <taqattack> and you are trying to convert a file into h264 format?
[04:25] <taqattack> essentially, you'll want something like "ffmpeg -i <inputfile> -vcodec libx264 -preset <ultrafast, superfast, veryfast, faster, fast, medium, slow, slower, veryslow, placebo> -acodec <libmp3lame,libvo_aacenc> -f <mp4,flv,mkv> <outputfile>"
[05:32] <dan3son> I'm getting a bug, trying some predefined stuff
[05:32] <dan3son> -flags2 +bpyramid+wpred+mixed_refs+dct8x8
[05:33] <taypen> hey, i've got some problem converting mkv to mp4 : http://pastebin.com/U2eiAv7g
[05:33] <dan3son> http://pastebin.com/AU6uZcyp
[05:34] <dan3son> I don't understand why I'm getting the error in processing my arguments
[05:35] <taypen> taypen, well past post been marked as spam :( here's another one http://pastebin.com/RkmvwALh
[05:36] <dan3son> [NULL @ 00000000003cc570] Unable to parse option value "bpyramid+wpred+mixed_ref
[05:36] <dan3son> s+dct8x8"
[05:49] <relaxed> dan3son: use a preset
[05:52] <dan3son> If I use -s 1920x1080 it will scale of course
[05:52] <dan3son> How does it do aspect ratio conversions by default?
[05:53] <dan3son> Like if I'm using a 4:3 source what does it do?
[05:54] <relaxed> stretch
[05:55] <relaxed> ffmpeg now has video filters that have more advanced options, like keeping the aspect ratio.
[05:55] <relaxed> look at the man page or read about it on the website.
[05:57] <relaxed> taypen: 2> log
[05:58] <taypen> relaxed, nah i just wanted to use pastebin program under linux
[06:00] <taypen> relaxed, same error without "> log"
[06:00] <relaxed> taypen: oh, use .... -f mp4 Contrabanda.m4v
[06:01] <relaxed> or omit the subs
[06:01] <taypen> same
[06:01] <taypen> how do i omit them?
[06:02] <relaxed> remove -scodec copy, add -map 0.0 -map 0.1
[06:03] <taypen> it means map the to nothing?
[06:03] <relaxed> do you want english or russian audio?
[06:03] <taypen> russian
[06:03] <relaxed> it means map (use) only the streams you specify.
[06:04] <taypen> ok
[06:04] <taypen> i'm recompiling ffmpeg now, added enabled some flags
[06:05] <taypen> what i'm writing... without added * )
[06:05] <relaxed> I hope you added support ofr another aac encoder such as libfaac or libvo-aacenc
[06:06] <relaxed> you already have libvo-aacenc support
[06:06] <relaxed> use -acodec libvo_aacenc instead, remove "-strict experimental"
[06:06] <taypen> i'm adding v4l sdl theora and vorbis support
[06:07] <relaxed> you're using gentoo, aren't you?
[06:07] <taypen> yes
[06:07] <relaxed> gentoo's default use flags for ffmpeg are fucking retarded.
[06:08] <relaxed> you would be better off compiling ti by hand.
[06:08] <taypen> :)
[06:08] <relaxed> it*
[06:08] <taypen> maybe if it wont work with this flags
[06:10] <taypen> and the latest portage tree have old version of ffmpeg
[06:10] <relaxed> One time a guy was in here and he had the "-encode" use flag set.
[06:12] <taypen> yes but gentoo is not so bad :)
[06:13] <relaxed> Yes it is. If it allows you to compile a butchered version of ffmpeg, how else has it allowed you to shoot yourself in the foot.
[06:14] <relaxed> ?
[06:14] <taypen> gentoo rely on user
[06:14] <taypen> what linux are you using?
[06:14] <relaxed> I'm aware of that. I'll stop trolling your choice of distro.
[06:15] <relaxed> Debian Sid
[06:18] <taypen> ok "use -acodec libvo_aacenc instead, remove "-strict experimental"" this line helped
[06:19] <taypen> or maybe it's because i removed subs
[06:20] <taypen> yes it's because i removed subs and added maps
[06:20] <relaxed> both helped
[06:21] <taypen> no i tried now with old acodec and with experimental
[06:21] <relaxed> you'll get better quality with libvo_aacenc
[06:23] <taypen> isn't those should be the same aac
[06:24] <taypen> well changed to libvo_aacenc )
[06:24] <relaxed> different encoders of the same codec. libvo_aacenc > aac
[08:37] <neonoe> Hi, howto mix tow video streams by incrusting one of them as reduced image overlaid on the other kept at full size ?
[08:37] <neonoe> I have seen old topics on this subject on the web. I would know if you have some tricks for such purpose.
[12:51] <loo> Any ideas how to change the buffer size of ffmpeg? http://ffmpeg.org/trac/ffmpeg/ticket/1224 Or is there a different way how to pass that stream?
[14:57] <loo> what's the format of this sample http://beberry.lv/up/u/sample ?
[15:00] <loo> burek, I don't know those problems were related to buffer, because once  I got the ffmpeg to convert 25 seconds.. which was 2x the usual size..
[15:09] <basilgohar> burek: Returning to my issue, I noticed that the Matroska I made from remuxing the AVCHD files appears to have slightly different frame rate properties than the originals...namely, the tbc value is 59.92 instead of what should probably be 59.94 (i.e., 60000/1001).
[15:09] <basilgohar> So, I think the issue I am observing has to do with the fact that Matroska has a limit precision when it comes to frame time lengths.
[15:09] <basilgohar> *limited
[15:09] <basilgohar> This is my suspicion, because I'd hear of this from another channel.
[15:10] <basilgohar> It might also explain why I get two dropped frames.  I wonder if I can specify the framerate when muxing, and if that would make any difference...
[15:11] <basilgohar> I could also just be completely clueless and talking nonsense. :)
[15:13] <Mavrik> basilgohar: I've noticed dropped frames when concating AVCHD files as wel
[15:13] <Mavrik> l
[15:13] <Mavrik> basilgohar: right on the file boundaries
[15:14] <stowelly> Hi guys, does anyone know much about the Dirac / Schrodinger formats at all?
[15:15] <Mavrik> basilgohar: my solution was to use Premiere to put them together and remux them to mp4
[15:15] <Mavrik> basilgohar: it seems that those other index files around MTS files were required for our Canon camera
[15:16] <stowelly> im trying to assess for our game wether we could get a smaller filesize than Bink without sacrificing on quality, but in all my tests using ffmpeg the file sizes have been quite huge (however possibly im using the wrong encode params)
[15:17] <loo> Mavrik, can you please take a look at this http://ffmpeg.org/trac/ffmpeg/ticket/1224 ?
[15:22] <basilgohar> Mavrik: Can you suggest any other way to remux AVCHD files without using closed software?  I never noticed glitches when playing back files muxed in this manner, and I do it all the time...
[15:23] <basilgohar> I guess one dropped frame might slip by me, though.
[15:23] <Mavrik> basilgohar: I didn't find any
[15:23] <basilgohar> stowelly: There's a #dirac channel, and the implementer of libschroedinger (ds) is usually there.
[15:24] <stowelly> thank you basilgohar
[15:24] <basilgohar> If you just want to avoid licensing, then you might consider using VP8/WebM.
[15:24] <basilgohar> I mean, licensing costs.
[15:26] <stowelly> our legal team arent happy with that, particulary with the legal sh*t storm looming, think they estimated we would have to pay MPEG LA about 0.4 gbp per unit if they win
[15:27] <loo> anyone?
[15:39] <iive> stowelly: how about just using mpeg1 and good encoder?
[15:40] <stowelly> dont you have to pay royalties for distributing a mpeg1 decoder?
[15:41] <iive> i think I don't, and even if you do, it would soon become 20 years from the publishing of the standard.
[15:42] <iive> "The standard was finished with the 6 November 1992 meeting"
[15:47] <stowelly> ahh apparently after speaking to my lead, that mpeg 1 wasnt good enough for what we need, i.e file size to quality ratio
[15:49] <iive> well... this is why I said "good encoder". One of big problems with mpeg1/2 is their inexact IDCT definition. This forces it to use frequent keyframes that ruin the compression.
[15:50] <iive> mpeg4 main improvement was more strictier (i)dct constraints. h264 for example uses a bit-exact one.
[15:51] <iive> I remember an iso working on mpeg1/2/4 "standard" like idct that is also defined to be bitexact. It may be viable "workaround".
[15:54] <basilgohar> stowelly: You mean your legal team isn't happy with using WebM?
[15:54] <taqattack> I have a question about licencing ffmpeg. I'm writing an GUI application for FFmpeg. I intend to keep the GUI code closed source. So does that mean I can't distribute FFmpeg with software?
[15:55] <iive> basilgohar: webm/vp8 is known to use some things that are very similar to patented stuff in h264.
[15:55] <Mavrik> taqattack: you can't even link to it.
[15:56] <Mavrik> unless you make sure it's LGPL and your licence is LGPL compliant
[15:56] <iive> Mavrik: are you kidding?
[15:56] <Mavrik> (not all en/decoder licenses are LGPL compliant so you'll have to build your ffmpeg without those)
[15:57] <Mavrik> you can, of course, invoke it as a process
[15:57] <iive> (obviously I wrote my question after seeing only your first line).
[15:57] <Mavrik> basilgohar: last time I checked MPEG LA was claiming patent royalties for WebM as well ;)
[15:58] <taqattack> So if I build LPGL version of ffmpeg and create dlls. I can distribute it with my software?
[15:58] <Mavrik> yes
[15:59] <iive> imho (not legal) but lgpl should allow you to distribute the unmodified ffmpeg binary with any closed source program. As long as you follow the lgpl rules (provide offer to the source etc...).
[15:59] <Mavrik> iive: you can always distribute the binary
[15:59] <Mavrik> even if it's GPL
[15:59] <Mavrik> (unless it's specifically not distributable)
[15:59] <Mavrik> difference is if you can link against ffmpeg (libav*) libraries with your closed-source program
[16:00] <Mavrik> calling ffmpeg.exe from your closed-source program does not violate GPL
[16:00] <taqattack> Oh really
[16:00] <iive> with gpl it may be a little bit more tricky, depending of interpretation of the "linking to". aka is the usage generic or specific.
[16:00] <basilgohar> Mavrik: Nope.  MPEG-LA made a public call for people interested in forming a pool around VP8.  Nothing else followed, and that was over a year ago.
[16:00] <taqattack> But I was wondering if I could redistribute the ffmpeg.exe with GUI application as package
[16:00] <iive> but it should still allow you to distribute them separately.
[16:00] <Mavrik> basilgohar: that and they made public claims VP8 infringes their patents
[16:01] <Mavrik> basilgohar: as of yet that was not tested in court
[16:01] <basilgohar> iive: Yes, very similar, and very obviously worked around issues known to be patented.
[16:01] <basilgohar> Mavrik: Very few things are tested in court.
[16:01] <Mavrik> well, it's enough to make our legal team nervous
[16:01] <stowelly> either way, if it gets tried in court and loses it will cost us half a million!
[16:01] <basilgohar> Yes, they've made plenty of FUD-like claims around VP8, because an unencumbered codec directly threatens their business model.
[16:01] <iive> basilgohar: legal matters are like quantum physics.
[16:02] <Mavrik> it'd be annoying if you'd have to pay ton of royalties for "free" WebM when MPEG-LA starts trolling around
[16:02] <Mavrik> basilgohar: of course they did
[16:02] <basilgohar> I'm not saying depend on Google to defend you, but a lot of companies have already adopted VP8 and have built software and hardware around it.
[16:02] <Mavrik> basilgohar: but with US legal system being what it is
[16:02] <Mavrik> trusting your business to the "logic" of non-infringement of VP8 is... stupid
[16:03] <Mavrik> at least H.264 is a known evil
[16:03] <iive> well, even if you pay mpeg-la, you could still be sued by 3'd parity.
[16:03] <Mavrik> iive: cue Motorola vs. Microsoft :D
[16:03] <iive> I think there was some such "claim" not long ago.
[16:04] <Mavrik> yeah, Motorola came and sued Microsoft for H.264 - MPEG LA doesn't own whole H.264 patent pool :D
[16:04] <Mavrik> Microsoft moved their datacenters from Germany to Nederlands because of it
[16:04] <iive> Mavrik: to be precise, mpeg-la doesn't own the whole patent pool :)
[16:05] <basilgohar> Mavrik: MPEG-LA doesn't provide indemnification, either.
[16:05] <basilgohar> And, as pointed out, it doesn't provide protection from lawsuits outside of their pool either.
[16:05] <basilgohar> VP8 was designed to avoid patented techniques, and it's backed and being actively used by Google and many software and hardware vendors.
[16:06] <Mavrik> the most reasonable option is to just not do business with US if at all possible
[16:06] <iive> the problem is WIPO
[16:06] <Mavrik> opening your business to patent trolling can just bring too much damage to the company
[16:06] <basilgohar> Mavrik: Even though I live in the US, I find your statement very reasonable!
[16:06] <Mavrik> I just hope that part of law system gets revised asap :(
[16:06] <loo> http://ffmpeg.org/trac/ffmpeg/ticket/1224
[16:07] <basilgohar> The US is trying to move to a more "IP"-oriented economy, because it's easier to make money off of something you don't need a factor to make and trains to ship.
[16:07] <iive> Mavrik: they just made one patent reform. Don't you feel the improvements they brought?
[16:09] <Mavrik> iive: oh? when was that?
[16:09] <Mavrik> (we mostly focused on EU lately)
[16:09] <Mavrik> ugh, gotta go, see ya guys :)
[16:09] <iive> 1-2 years ago. They moved from invented first to filled for patent first (like it is in EU). And changed few minor other things.
[16:35] <basilgohar> Okay, so I think I was able to remux the file from an h264 elementary stream using gstreamer.  I have to play with the files now, but at least I have files!
[17:02] <someone-noone> I want to synchronize audio to external clock(yes, yes i know, look at ffplay.c). I understand that i need to update my audio_clock every time i read audio packet and every time i read some audio samples. In my program i don't operate double values, instead i'm using uint64_t which represents timestamp in milliseconds. So my external clock is just current timestamp. The problem is: how to correct calculate audio_clock. Here is what 
[17:02] <someone-noone> 1) when program starts i make audio_clock = gettimeofday.
[17:03] <someone-noone> 2) later, when i'm trying to decode and play audio data i need to update my audio_clock
[17:04] <someone-noone> in "GetPacket" i do:
[17:04] <someone-noone> if (_audio_pkt.pts != AV_NOPTS_VALUE)
[17:04] <someone-noone> 		_audio_clock += _audio_stream->time_base.num*_audio_pkt.pts/_audio_stream->time_base.den;
[17:04] <someone-noone> in "GetSamples" i do:
[17:04] <someone-noone> n = 2 * _audio_stream->codec->channels;
[17:04] <someone-noone> _audio_clock += _audio_chunk_size / (n * _audio_stream->codec->sample_rate*1000);
[17:06] <someone-noone> the problem is i don't understand if i'm doing it correctly and if no, how should i do it? I don't understand what exactly "pts" represents, can someone explain?
[17:06] <iive> presetnation timestamps.
[17:06] <iive> aka the time when the sample should be presented.
[17:07] <iive> however with audio the things are a little bit complicated.
[17:07] <someone-noone> iive: in seconds, right? and it should be converted to human value with time_bas
[17:07] <someone-noone> ?
[17:07] <iive> the problem is that when you have 44100 samples per second, you can assume that playing 44100 would advance the clock with exactly 1 second.
[17:08] <iive> however if you use wall clock (like get time of the day) you may find that it sometimes gets less than second, and sometimes more than a second.
[17:09] <iive> some of the audiocards use separate oscilator thus they playback it slower of faster.
[17:10] <iive> so for example with mpeg streams up to 3 clocks can be encoded. video, audio and system (of the system supposedly captured the a/v).
[17:10] <iive> so, if there is audio only, you can do 2 things.
[17:11] <iive> first use only the audio as clock. rely entirely on the samples played.
[17:11] <iive> second... monitor both clocks and resample the input to compensate for the drift.
[17:12] <iive> well.
[17:12] <iive> audio clock should be relative to the audio stream, not the time of the day.
[17:13] <iive> same goes for the usual timestamps.
[17:15] <someone-noone> so, if i understand right, it should be 0 at beginning and should grow each time i get packet\decode frame, but what should be external clock than? If i make audio clock to gettimeofday()("in milliseconds") and than will add pts time to this value and use external clock as simple gettimeofday() ?
[17:17] <iive> someone-noone: why you need external clock?
[17:17] <rly> Is there any combination of ffmpeg + open-source viewer which allows me to create a screencast?
[17:18] <rly> All ways I have tried result in excuses being thrown by ffmpeg or really bad output or segfaults when I watch it with mplayer or vlc.
[17:18] <rly> It seems rather terrible that mplayer and vlc crash in both cases.
[17:18] <rly> Somehow once rich media is involved it seems OK to crash.
[17:19] <rly> Really, really questionable practices.
[17:19] <someone-noone> iive: i need to synchronize to smth :) when i decode some avi file and want to play just audio it plays very fast.
[17:19] <someone-noone> rly: you my try ffplay ?
[17:19] <someone-noone> may*
[17:19] <someone-noone> rly: and with which software are you capturing screen? platform?
[17:19] <rly> someone-noone: ffmpeg -f oss -i /dev/audio -f x11grab -r 30 -s 1645x579 -i :0.0 -sameq screengrab.avi
[17:19] <rly> someone-noone: I have tried that command for example.
[17:20] <rly> someone-noone: either the output is wrong or all of the viewers don't work.
[17:20] <rly> Or both.
[17:20] <rly> someone-noone: Linux.
[17:20] <someone-noone> rly: ffplay plays that screengrab? Is there any errors while encoding?
[17:20] <iive> someone-noone:  if you have to synchronize to something you don't have control over, then I guess you would have to resample of pitch the audio faster/slower to compensate for the sound card drift.
[17:21] <rly> someone-noone: WARNING: library configuration mismatch
[17:21] <iive> however "audio plays very fast" indicates there is some kind of another problem, not just clock drift.
[17:21] <someone-noone> rly: try to rebuild your ffmpeg
[17:21] <rly> someone-noone: I already did.
[17:22] <rly> someone-noone: that's what I got when I did this.
[17:22] <rly> What I don't get is why that error message exists in the first place.
[17:22] <rly> It shouldn't be possible to get in this state in the first place.
[17:22] <iive> rly: why are you scaling to odd dimensions?
[17:22] <someone-noone> rly: clean old version of ffmpeg, download new version -> configure & make & make install
[17:23] <rly> someone-noone: I have git ffmeg, define 'clean'.
[17:23] <rly> someone-noone: make clean?
[17:23] <rly> someone-noone: make uninstall?
[17:23] <someone-noone> rly: you shouldn't get this warning
[17:23] <rly> someone-noone: and yet someone made a mistake such that I do get it.
[17:23] <rly> Sorry for being the messenger.
[17:24] <iive> rly: that commands works for me. I think i have the last release installed.
[17:24] <rly> I don't even get the error message.
[17:24] <someone-noone> iive: what kind of problem it can be? I'm using WinMM to play LPCM, if i load some uncompressed wav file to it and play it in loop it works fine
[17:24] <rly> A mismatch is always between two things.
[17:24] <rly> It should tell me exactly what doesn't match.
[17:24] <rly> And then it should also tell me why that is a problem.
[17:25] <rly> Finally, it should tell me a sequence of steps that will solve that problem.
[17:25] <rly> This is basic software engineering.
[17:25] <rly> Software engineering shouldn't involve having to fix other peoples' broken software.
[17:25] <rly> If everyone would have to do that, nothing would help.
[17:26] <iive> rly: it said library. ffmpeg is probably build with shared libraries, and they have been overwritten.
[17:26] <iive> it is likely ld.so error.
[17:26] <rly> Ok, I see it.
[17:27] <rly> ldconfig wasn't run.
[17:27] <rly> But then I still get another error.
[17:27] <rly> SDL_OpenAudio:
[17:27] <rly> X Error of failed request:  BadValue (integer parameter out of range for operation)
[17:28] <iive> try without sound -an i think.
[17:28] <rly> Is recording sound (possible with Windows 95) too advanced?
[17:28] <iive> depends on libsdl.
[17:30] <rly> X Error of failed request:  BadValue (integer parameter out of range for operation)
[17:30] <rly>   Major opcode of failed request:  133 (XVideo)
[17:30] <rly>   Minor opcode of failed request:  19 ()
[17:30] <rly>   Value in failed request:  0x90
[17:30] <rly>   Serial number of failed request:  27
[17:30] <rly>   Current serial number in output stream:  28
[17:31] <rly> iive: similar issue.
[17:31] <rly> iive: this is without sound.
[17:31] <taqattack> Ok so from what I gather, I can call ffmpeg.exe from my GUI application which I intend to be closed source. But I cannot build and redistribute the ffmpeg.exe with my application. Is this corect interpretation of the licensing?
[17:31] <rly> taqattack: what's the ffmeg licence? LGPL?
[17:31] <rly> (some parts are GPL)
[17:31] <taqattack> I'm using the zeranoe build which is GPL
[17:33] <iive> rly: are you running under X server? have you escalated your privilages (su) ?
[17:33] <rly> iive: yes and no.
[17:33] <iive> can you e.g. start a new terminal from the cli?
[17:34] <rly> iive: I can run xterm from a terminal emulator.
[17:34] <rly> iive: what do you mean?
[17:35] <iive> just checking if the programs have correct display link.
[17:35] <rly> iive: the problem lies with epic failure on the side of ffmpeg.
[17:35] <iive> as I said... works for me.
[17:36] <iive> so the problem is somewhere on your end.
[17:37] <rly> iive: the software exits with an X error.
[17:37] <iive> X means Xorg server.
[17:37] <rly> iive: it doesn't say 'we tried to call this and that X function, which we know must work because of reasons X, Y and Z', but it didn't work.
[17:38] <rly> In short, I have no reason to assume this isn't a programming error.
[17:38] <rly> Because otherwise it would have worked.
[17:38] <iive> rly: actually this is exactly what is says. X returned error the program didn't expect.
[17:39] <rly> iive: that's an X error message.
[17:39] <iive> try without the -s
[17:40] <rly> [mpeg4 @ 0x141a080] Error, Invalid timestamp=40, last=40
[17:40] <rly> Video encoding failed
[17:40] <rly> Yeah, that's much better.
[17:40] <rly> ...
[17:40] <rly> Also, it just says 'failed'.
[17:41] <rly> It doesn't say 'byte 90' expected.
[17:41] <rly> Or 'you have byte 90 10 30, but the specification says it has to be byte 90 10 31' or whatever.
[17:41] <rly> This is just amateuristic crap software.
[17:42] <iive> whatever.
[17:42] <rly> I am not the only one who has these issues with ffmpeg.
[17:42] <iive> try at #libav
[17:49] <rly> iive: it does play with mplayer, but it is full of graphical bugs.
[17:49] <rly> iive: so, that leads to the conclusion that not only ffplay stinks, but the recording part too.
[17:50] <iive> I told you the last release works fine.
[17:50] <rly> iive: so, you are saying that git is broken?
[17:50] <rly> iive: fine, I will compile the latest release to prove you wrong.
[17:50] <rly> iive: compilation options for ffmpeg please.
[17:50] <rly> iive: it must be the exact same thing.
[17:50] <iive> probably. one of the prices of using developer version software.
[17:50] <rly> iive: otherwise it is still comparing apples to oranges.
[17:51] <iive> rly: same options.
[17:51] <rly> iive: no, I mean compile time options.
[17:51] <iive> you mean, the ones I've used?
[17:51] <rly> iive: ffmeg --version | someparstecommand
[17:51] <rly> iive: you say it works. I say it doesn't. I would have to use the same options as you did to make it work.
[17:52] <rly> iive: the Debian Stable version _also_ doesn't work, btw.
[17:52] <iive>  configuration: --prefix=/usr --enable-gpl --enable-version3 --enable-nonfree --enable-postproc --enable-pthreads --enable-static --enable-shared --enable-libfaac --enable-libtheora --enable-x11grab --enable-libmp3lame --enable-libvorbis --enable-libx264 --enable-libxvid --cpu=core2 --enable-debug=3 --disable-stripping
[17:52] <rly> iive: so, it is not as if stable releases mean _anything_.
[17:52] <iive> rly: debian doesn't even package ffmpeg.
[17:52] <rly> iive: Debian Multimedia does.
[17:53] <iive> 0.10 is the one that works.
[17:54] <iive> and yes, the current git does flood my console with "invalid dropping st:0"
[17:54] <iive> the resulting file still plays fine.
[17:55] <rly> iive: I have an Intel Core cpu, how is that named in CPU?
[17:55] <rly> (that's yet another bug in the configure --help output)
[17:55] <rly> It should list all valid values for CPU.
[17:55] <iive> it should query them from the gcc -march
[18:00] <rly> iive: that release doesn't even come with --enable-gpl.
[18:01] <rly> Hmm, or perhaps it does.
[18:01] <rly> Odd
[18:01] <iive> i just thought you have connected from parallel universe. Was going to ask for presidents names and such :E
[18:02] <rly> Ok, compiling...
[18:06] <rly> [swscaler @ 0xa2d2e0] 0x0 -> 640x480 is invalid scaling dimension
[18:06] <rly> [swscaler @ 0xa2d2e0] 0x0 -> 640x240 is invalid scaling dimension
[18:06] <rly> [swscaler @ 0xa4aec0] 0x0 -> 640x240 is invalid scaling dimension
[18:06] <rly> Failed to inject frame into filter network
[18:06] <rly> iive: then we get that.
[18:06] <rly> iive: with command: ffmpeg -f x11grab -r 30  -i :0.0 -sameq screengrab.avi
[18:09] <rly> Do you even ever test your software?
[18:10] <rly> I think there have been 4 issues reported for this in the past 3 years and all of them didn't fix it.
[18:10] <rly> Who reviews the patches? Santa Clause?
[18:12] <iive> michaelni, so you have patches that are not applied?
[18:12] <iive> actually elvis presley. :E
[18:13] <relaxed> rly: you have to specify the frame size before the input and don't use -sameq
[18:14] <rly> relaxed: before the input?
[18:14] <rly> relaxed: just show a complete command that works.
[18:14] <rly> relaxed: otherwise it's just comparing apples and oranges all over again.
[18:14] <relaxed> ffmpeg -s WxH ... -i INPUT ...
[18:15] <rly> relaxed: ffmpeg -f x11grab -r 30 -s 897x579 -i :0.0 -sameq screengrab.avi
[18:15] <rly> relaxed: results in epic failure similar to above.
[18:16] <relaxed> which version?
[18:17] <rly> ffmpeg version 0.10.2 Copyright (c) 2000-2012 the FFmpeg developers
[18:17] <rly> But I do get  WARNING: library configuration mismatch _again_.
[18:19] <relaxed> remove the old libs
[18:19] <rly> relaxed: which ones?
[18:19] <rly> relaxed: the error message should tell me what to do.
[18:20] <rly> It created the problem, so it can solve it.
[18:20] <relaxed> which distro?
[18:21] <rly> relaxed: Debian, but I built it from source.
[18:21] <rly> It shouldn't even install if there is some problem.
[18:21] <rly> The theorem 'after installation it works' should always hold.
[18:22] <relaxed> try, LD_LIBRARY_PATH=/usr/local/lib ffmpeg ...
[18:22] <relaxed> stop spouting your nonsense
[18:22] <rly> It's not nonsense.
[18:22] <iive> or run ldconfig again...
[18:22] <rly> I already ran ldconfig.
[18:22] <rly> Didn't help this time.
[18:22] <rly> No idea why not.
[18:23] <iive> probably because my options are used to build my own package.
[18:23] <iive> so they ended up on a different location (prefix=/usr)
[18:23] <rly> iive: I changed the prefix.
[18:23] <relaxed> to what?
[18:24] <rly> Ok, never mind. That's the problem this time.
[18:24] <rly> Still, I don't expect it to work.
[18:24] <iive> of course not.
[18:24] <iive> these liberal hippies can't even tie their shoes correctly.
[18:25] <rly> That has been my experience, yes.
[18:25] <rly> Most of them shouldn't be allowed near to a computer.
[18:26] <iive> ~~
[18:26] <rly> Take 1000
[18:27] <rly> [buffer @ 0x1ce0740] Buffer video input changed from size:897x579 fmt:bgra to size:0x0 fmt:yuv420p
[18:27] <rly> [swscaler @ 0x1cf1020] 0x0 -> 897x579 is invalid scaling dimension
[18:27] <rly> [swscaler @ 0x1d19740] 0x0 -> 897x289 is invalid scaling dimension
[18:27] <rly> [swscaler @ 0x1d19640] 0x0 -> 897x289 is invalid scaling dimension
[18:27] <rly> Another prediction that came true.
[18:28] <rly> I should become an oracle.
[18:28] <rly> Any more excuses as to why it doesn't work?
[18:30] <rly> Perhaps you should focus on having less features and for the ones that do ship make sure that it works.
[18:31] <TimNich> me/ decides its time to start the weekend
[18:31] <TimNich> away/ its the weekend...
[18:44] <DelphiWorld> hey all
[18:44] <DelphiWorld> finaly my project is compiled now
[18:44] <DelphiWorld> but
[18:44] <DelphiWorld> how to list ffmpeg actual linked codecs?
[18:47] <relaxed> rly: try libav
[18:47] <rly> relaxed: don't you ship that too?
[18:48] <rly> relaxed: and obligatory: which version?
[18:48] <rly> It also appears that all X capture programs stink.
[18:48] <rly> E.g. they show large white blocks which are not visible when using it normally.
[18:48] <rly> For example when there is scrolling terminal output.
[18:54] <relaxed> rly: libav is the ffmpeg fork. libav.org
[19:01] <rly> and what is this?  [VD_FFMPEG] DRI failure.
[19:01] <rly> This is not demonstrating DRI failure.
[19:01] <rly> What it should so is output a test case for something which fails.
[19:02] <rly> A minimal case at that.
[19:02] <rly> No, but someone decided 'let's just scream it failed'.
[19:06] <iive> DelphiWorld: -formats, the external codecs should start with lib*
[19:06] <rly> relaxed: I built libav, now what?
[19:07] <rly> relaxed: i.e. can libav also record?
[19:07] <relaxed> rly: yes, libav has avconv which behaves just like ffmpeg.
[19:09] <DelphiWorld> iive: thx
[19:09] <rly> relaxed: which returns Unknown input format: 'x11grab'
[19:09] <rly> relaxed: and yes I had --enable-x11grab
[19:51] <RobertNagy> could someone explain the purpose of the sinkbuffer?
[19:51] <RobertNagy> aren't frames always pulled in a filter graph?
[19:51] <RobertNagy> a buffersink would only be useful if frames were push through the chain?
[20:42] <ptrkmj> I would like to use ffmpeg to download bunch of wmv files served via mmsh. I don't want to bother with stream mappings - is there an option to just dump those streams without remuxing?
[20:50] <brontosaurusrex> what would i need to do to compile ffplay as well? (osx)
[21:08] <cbreak> brontosaurusrex: sdl? :)
[21:08] <brontosaurusrex> cbreak: how?
[21:08] <cbreak> what do you want ffplay for? Both VLC and mplayer are probably more capable
[21:08] <cbreak> what how?
[21:09] <brontosaurusrex> mplayer is not really working as it should
[21:09] <cbreak> of course it is.
[21:09] <brontosaurusrex> cbreak: do you have a bin i can try?
[21:09] <cbreak> no. Compile it yourself.
[21:09] <brontosaurusrex> i did
[21:12] <taqattack> Okay I'm having some trouble with RTMP streaming with FFmpeg
[21:12] <taqattack> http://pastebin.com/gVw3EdZ1
[21:16] <Mista_D> What lib is used to decode/encode s302m (BSSD) please?
[23:03] <stf> i have a .ts file from a tv-recording which have different audio streams, i want to cut the file by now i try it this way:  'ffmpeg -ss 192 -t 6778 -vcodec copy -acodec copy -map 0.1:0.1 -map 0.3:0.2 -map 0.6:0.3 -i "test.ts" "test_1.ts"' i get the error "Number of stream maps must match number of output streams" < what does this means?
[23:07] <stf> hello
[23:07] <stf> ?
[23:07] <stf> anyone who can help?
[23:15] <sokak> Hello everyone. Im trying to experiment to convert a dual audio mkv with subtitles to .avi, to be played on a cheap dvd player (it plays avis, i want to try to see if it can play also double audio avis for fun) I think i messed it up pretty weird, there is the sample output of this  ffmpeg -i input.mkv -map 0:0 -map 0:1 -map 0:2 -map 0:3 -map 0:4 -vcodec copy -acodec copy -acodec copy -scodec copy -scodec copy test.avi  ht
[23:15] <sokak> tp://pastebin.com/Z7c3JVAe
[23:16] <iive> stf: counting begins from zero
[23:16] <stf> ah
[23:17] <stf> live: 0.0 is the video stream, or not?
[23:17] <sokak> yes it is
[23:17] <iive> stf: yep
[23:17] <sokak> (ops sorry x3)
[23:18] <stf> live: ehm but where is my failure?
[23:18] <stf> do i need to -map 0.0:0.0 too?
[23:19] <stf> live: ?
[23:19] <iive> stf: you need to map something to the first output stream.
[23:20] <stf> okay
[23:20] <stf> live: could you give me an example?
[23:22] <stf> ah wait i have to map 0.0:0.0
[23:24] <stf> what do i need to do if i only want to have streams 0.0, 0.1, 0.3 and 0.6 to the new file?
[23:32] <stf> live: ?
[23:33] <sacarasc> It's iive, not live. Double lower case I.
[23:34] <stf> ups
[23:34] <stf> iive: ?
[23:35] <iive> i thought you had it figured out? doesn't it work?
[23:36] <iive> btw, i'm not sure if you can copy subtitles into avi.
[23:36] <iive> hum.. that was in the sokak, example.
[23:36] <sokak> hehe yeah sorry i mixed up :x
[23:37] <stf> ehm i want to copy from ts to ts
[23:38] <stf> iive: i want to copy from ts to ts
[23:42] <iive> so map 0.0:0.0 doesn'twork?
[23:44] <stf> This works: "-map 0.0:0.0 -map 0.1:0.1" But do i add: "-map 0.2:0.2" therefore it is worried me
[23:44] <stf> ehm no it do not worried me but i do not understand the problem
[23:46] <DelphiWorld> iive: aac come by default with ffmpeg?
[00:00] --- Sat Apr 21 2012