[Ffmpeg-devel-irc] ffmpeg.log.20130131

[burek]

[00:18] <ska> oops. sorry.
[00:21] <sacarasc> You should be sorry!
[00:21] <sacarasc> What for?
[00:37] <K-Rich> Hi all
[00:37] <K-Rich> Cananyone tell me why this doesn't work? ffmpeg -y -acodec pcm_s16le -f alsa -i pulse -s 640x360 -qscale 1 -r 30 -vcodec rawvideo -f video4linux2 -i /dev/video0 -f yuv4mpegpipe | tee test.avi | mplayer -cache 1024 -nosound -
[00:38] <K-Rich> sorry ffmpeg -y -acodec pcm_s16le -f alsa -i pulse -s 640x360 -qscale 1 -r 30 -vcodec rawvideo -f video4linux2 -i /dev/video0 -f yuv4mpegpipe - | tee test.avi | mplayer -cache 1024 -nosound -
[00:45] <K-Rich> anyone?
[00:45] <K-Rich> sorry ffmpeg -y -acodec pcm_s16le -f alsa -i pulse -s 640x360 -qscale 1 -r 30 -vcodec rawvideo -f video4linux2 -i /dev/video0 - | tee test.avi | mplayer -cache 1024 -nosound -
[00:45] <K-Rich> gives errors
[00:47] <K-Rich> http://pastie.org/5976041
[00:57] <K-Rich> and still noone
[00:57] <K-Rich> heh
[01:10] <klaxa> K-Rich: two things
[01:11] <klaxa> actually three
[01:11] <K-Rich> ok
[01:11] <klaxa> a) you are using avconv, this is the wrong channel unfortunately
[01:11] <klaxa> b) apparently you are not allowed to access /dev/video0
[01:11] <klaxa> c) why do you bother to capture audio if you don't play it back with mplayer?
[01:11] <K-Rich> (mind you i "just" tried to change to real ffmpeg, with this result http://pastie.org/5976323 ))
[01:12] <klaxa> ah didn't see the tee pipe
[01:12] <K-Rich> i want the audio captured in the file but want to see the video, if the displayed video had sound it would cause a feedback loop with the recording
[01:13] <klaxa> i'm not sure, but i think when playing back raw audio (pcm_s16le) and raw video you will somehow have to tell mplayer that it is raw data
[01:13] <K-Rich> also i am in the group video and 'crw-rw----+ 1 root video 81, 0 Jan 30 15:18 /dev/video0'
[01:13] <klaxa> also the order of your parameters is a bit weird
[01:15] <K-Rich> i'm just learning this, trying to make an applet for screen/webcam recording
[01:15] <klaxa> i thnik general practice is ffmpeg -f <in_format> [in_options e.g. codec, filters] -i <in_stream> [-f container format] <out_file>
[01:15] <klaxa> *think
[01:17] <K-Rich> if  put a filename (capture.avi) rather than '-| tee test.avi | mplayer -' it works, just doesn't display
[01:18] <klaxa> ah wait...
[01:18] <klaxa> right...
[01:18] <klaxa> ffmpeg <all your stuff> -f avi pipe: | tee somefile.avi | mplayer -
[01:19] <K-Rich> /dev/video0: Operation not permitted   STILL ugh, even when i do it with sudo
[01:20] <klaxa> move the -i /dev/video0 before -vcodec even
[01:22] <K-Rich> http://pastie.org/5976457
[01:22] <K-Rich> same friggin access denied crap
[01:22] <K-Rich> err operation not permitted
[01:23] <K-Rich> hmmmmmmmm
[01:24] <K-Rich> one sec BRB
[01:24] <klaxa> try also adding something like -f video4linux2 before -i /dev/video0
[01:24] <K-Rich> i wonder if all the trying somehow left it open
[01:24] <K-Rich> ok
[01:35] <K-Rich> getting closer
[01:35] <K-Rich> i can see it
[01:36] <K-Rich> but the vid plays fast and the sound is empty (static only)
[01:36] <K-Rich> ffmpeg -acodec pcm_s16le -f alsa -i pulse -s 640x360 -qscale 1 -r 30 -f video4linux2 -i /dev/video0 -vcodec rawvideo -f avi pipe: | tee test.avi | mplayer -cache 1024 -nosound -
[01:38] <K-Rich> klaxa: you still here?
[01:38] <klaxa> yeah
[01:38] <klaxa> um... about the audio thing, have you selected the correct recording device in pulse?
[01:39] <klaxa> with the video thing, i'm not sure rawvideo has data about framerates
[01:40] <K-Rich> okay using mjpeg rather than rawvideo works sorta, way delayed and still no sound
[01:41] <klaxa> open pulseaudio volume control and check you have the correct device set for the recording stream
[01:45] <K-Rich> okay lol, got sound......
[01:45] <K-Rich> how do i know what encoders i have, i need one with yuv
[01:45] <K-Rich> still have the delay though
[01:56] <klaxa> K-Rich: ffmpeg -codecs
[02:04] <K-Rich> klaxa: woo hoo (sorta)
[02:04] <K-Rich> ffmpeg -acodec pcm_s16le -f alsa -i pulse -s 640x360 -qscale 1 -r 30 -f video4linux2 -i /dev/video0 -vcodec huffyuv -f avi - | tee test.avi | mplayer -nosound -cache 1024 -
[02:05] <K-Rich> works great, except there is a delay on the video... any idea on that?
[02:05] <klaxa> mmhh no idea actually, maybe huffyuv is straining your cpu? although it shouldn't really
[02:06] <K-Rich> the only one without a delay was rawvideo, but it played at like double speed, though the audio was normal speen (the image pauses half way through)
[02:15] <K-Rich> grrrrrr
[02:15] <K-Rich> updatung the ffmpeg made my deskrecord script look like ass (pixilated)
[02:15] <K-Rich> it was clear before
[02:18] <K-Rich> fow is it a command that worked with avconv doesn't work with ffmpeg ?
[02:18] <K-Rich> it's become all blocky (pixilated)
[02:19] <klaxa> sounds like a quality setting issue
[02:21] <K-Rich> ffmpeg -acodec pcm_s16le -f alsa -i pulse -s 1280x800 -qscale 1 -r 30 -vcodec rawvideo -f x11grab -i $DISPLAY -s 640x356 -vf pad=640:360:0:2:000000 ~/Desktop/Video_$(date +%F_%T).avi
[02:23] <klaxa> hmmm that should work though
[02:25] <K-Rich> it looks like it's an animated jpeg with low quality, all blocky
[02:25] <K-Rich> it worked great with ubuntus avconv ffmpeg whatever
[02:26] <K-Rich> Stream mapping:
[02:26] <K-Rich>   Stream #1:0 -> #0:0 (rawvideo -> mpeg4)
[02:26] <K-Rich>   Stream #0:0 -> #0:1 (pcm_s16le -> libmp3lame)
[02:26] <K-Rich> mpeg4 wtf?
[02:27] <K-Rich> ffmpeg version 0.10.6-6:0.10.6-0ubuntu0jon1~precise1
[02:29] <K-Rich> Mwahahahaa got it
[02:29] <K-Rich> added -sameq
[02:29] <K-Rich> works great now :)
[02:33] <K-Rich> heh... made the other command better as well
[02:33] <K-Rich> the output at least
[02:33] <K-Rich> though there is still delay :/
[02:42] <K-Rich> ffmpeg -acodec pcm_s16le -f alsa -i pulse -s 640x360 -qscale 1 -sameq -r 30 -f video4linux2 -i /dev/video0 -vcodec huffyuv  -f avi - | tee test.avi | ffplay -an -window_title Webcam\ Recording -
[02:42] <K-Rich> hmmmmmmm still a delay :/
[05:35] <Tronic> Strange problem with my own software... Decoding of MP3 and Ogg (didn't try other formats yet) produces highly distorted sound (lots of white noise etc), but the actual audio can be heard thru.
[05:36] <Tronic> Same source code has worked with libav and older ffmpeg versions. Also, ffplay, with current version, decodes audio just fine. Video decoding in my application is also working properly.
[05:37] <Tronic> Any suggestions other than trying to go thru my own code and ffplay line by line, seeking for diffferences?
[05:40] <Tronic> I was also afraid that different library versions might be mixing up but uninstalling everything but the latest didn't change anything. I am running current git version at the moment.
[05:54] <Tronic> With WAV format audio there is no distortion.
[09:34] <praveenmarkandu> hi. if a source file is already h.264 and aac and when i transcode it again and use -c:v libx264 -c:a libfaac
[09:34] <praveenmarkandu> will it be slower than just using -c:v copy
[09:37] <praveenmarkandu> what i actually want to do is reduce the bitrate and resolution of the video
[09:37] <praveenmarkandu> will copy make a speed difference?
[10:11] <cbsrobot>  praveenmarkandu: copy is faster
[10:11] <praveenmarkandu> cbsrobot, i just tried copy with my full command and the -vf scale didnt seem to work
[10:12] <cbsrobot> yeah stream copy disables the filters
[10:13] <praveenmarkandu> oh okay
[10:13] <praveenmarkandu> and -s:v is the same as -vf scale
[10:13] <praveenmarkandu> so it wont work either right?
[10:15] <praveenmarkandu> and it seems it also disables the video bitrate setting
[10:15] <praveenmarkandu> hah. okay. should have expected
[11:43] <lemonjelly> Hello. With ffmpeg, is it possible to offset all timestamps in the *output* streams by a certain amount?
[11:45] <lemonjelly> I've been experimenting with the -itsoffset and -copyts options but I can't get ffmpeg to behave as I need
[11:47] <lemonjelly> Basically I'm trying to output mov fragments of an ism stream, but I'm using the -ss <offset> option to start partway through the input stream, and I want the first fragment I output to have a frag_start value matching <offset>. But it is always 0.
[13:14] <Zeeflo> burek, are you around? You got a bug of some sort in your build
[13:14] <Zeeflo> (which has just set me back 2 days!! :/)
[13:35] <someone-noone> Hello! I'm getting crash(from time to time) in avcodec_decode_audio4()->aac_decode_frame()->av_packet_get_side_data() with next code: http://ideone.com/uxeYwz
[13:35] <someone-noone> I can't figure out what can be a problem. Sometimes problem occur and sometimes no
[13:35] <someone-noone> Any help?
[13:39] <Zeeflo> Any ideas? fontconfig: Selected font is not the requested one: 'DejaVu Sans' != 'Arial'its/s dup=1 drop=0
[13:41] <Mavrik> someone-noone: it seems something goes wrong when you're poping the packet
[13:41] <Mavrik> are you copying all fields?
[13:42] <Mavrik> is your buffer padded with at least FF_BUFFER_PADDING (or what's it) bytes?
[13:43] <someone-noone> Mavrik, one momen
[13:44] <someone-noone> Mavrik, http://ideone.com/VBliwr this is how I'm reading packets. I'm porting my code from old version of libav.
[13:45] <someone-noone> Mavrik, and buffer& you're talking about AVFrame's one?
[13:53] <ledil> Hello, Ive installed ffmpeg from jon-severinsson ubuntu ppa repository, but now Im getting "Unknown encoder 'libfaac'". libfaac-dev and libfaac-0 are installed ... any hints ?
[13:54] <ledil> this is my version: 0.10.6-6:0.10.6-0ubuntu0jon1~lucid2
[13:54] <ledil> when executing ffmpeg, i cant see a --enable-faac ... is faac deprecated ?
[13:54] <someone-noone> ledil, ffmpeg codecs
[13:54] <someone-noone> do you see libfaac?
[13:55] <Zeeflo> ledil, try the daily build for ubuntu
[13:55] <Zeeflo> ah you are
[13:55] <Zeeflo> hmm
[13:56] <ledil> http://pastebin.com/8xCYWaTc
[13:56] <ledil> here is the ouput
[13:56] <ledil> of ffmpeg -codecs and faac is missing ...
[13:57] <ledil> dont know what to do ? because when Im using the mediaxxx I cant install libavcodec-extras-52 because it conflicts with ffmpeg ... when using jon-severinsson package, im getting "libfaac" unknown ...
[13:57] <ledil> I cant find a good ffmpeg package, at the moment :(
[13:57] <JEEB> because faac is nonfree and thus impossible to enable if you're making a binary for others
[13:58] <ledil> how to solve this ?
[13:58] <JEEB> I think even debian stopped linking to it after that was found
[13:58] <Zeeflo> nope
[13:58] <Zeeflo> works fine here
[13:58] <Zeeflo> wheezy 3.2 kernel
[13:58] <ledil> my server is lucid
[13:59] <JEEB> not sure but in any case I think very recent builds no longer have faac enabled as it's nonfree, methinks
[13:59] <JEEB> ledil, 1) switch to fdk-aac, as that is of better quality for AAC 2) compile yourself
[13:59] <JEEB> that's the only way
[13:59] <Zeeflo> im using Bureks builds.. faac is working here
[13:59] <JEEB> both faac and fdk-aac are nonfree
[13:59] <JEEB> o_O
[13:59] <JEEB> burek has faac enabled?
[13:59] <JEEB> Zeeflo, paste ./ffmpeg output
[13:59] <Zeeflo> im using the aac libs in my encodings, so yea, id guess so!
[13:59] <JEEB> pastebin
[13:59] <Zeeflo> sure
[14:00] <JEEB> it should have the configure line too
[14:00] <someone-noone> ledil, I'd just install ffmpeg from sources
[14:00] <Zeeflo> http://pastebin.com/KnJHXz6L
[14:01] <JEEB> with fdk-aac instead of faac
[14:01] <JEEB> because fdk-aac is fraunhofer's encoder
[14:01] <Zeeflo> and this is my input: -c:a aac -strict experimental -ac 2 -b:a 128k
[14:01] <JEEB> oh
[14:01] <JEEB> that uses the internal one
[14:01] <JEEB> not faac
[14:01] <Zeeflo> ah
[14:01] <JEEB> that'd be called ffaac (ff-something is the internal thing)
[14:01] <Zeeflo> but aac is aac right ;)
[14:01] <JEEB> yes, but that is experimental and barely as good as vo-aacenc
[14:01] <JEEB> (or somewhat better)
[14:02] <Zeeflo> it works perfectly well
[14:02] <JEEB> but still worse per the same rate than faac or fdk-aac
[14:02] <JEEB> and faac isn't exactly great either
[14:02] <ledil> Zeeflo: you are using bureks builds ? what ppa repository is that ? i cant switch to fdk-aac at the moment, I need to find a repository with a ffmpeg that has faac support
[14:02] <JEEB> there are none
[14:02] <JEEB> because faac is nonfree
[14:02] <Zeeflo> ledil,  you need just the aac
[14:02] <JEEB> as I already said, Zeeflo is not using faac
[14:02] <ledil> ah,ok
[14:02] <JEEB> well, yes -- the "experimental" status mostly comes from the fact that it's not optimal, not that it doesn't work Zeeflo
[14:03] <Zeeflo> those two does the same thing
[14:03] <Zeeflo> JEEB, I am aware of that, but as long as it works, it works?
[14:03] <JEEB> -_-
[14:03] <Zeeflo> for my stereo 128k it works like I need it to
[14:03] <JEEB> as soon as you hit something that goes below the comfort level of the libavcodec encoder it will be derp
[14:03] <JEEB> at least it no longer adds random noise
[14:03] <JEEB> it used to do that like 1.5 years ago
[14:04] <Zeeflo> dunno about that :)
[14:04] <Zeeflo> It works for me :)
[14:04] <JEEB> well, if you don't care about audio quality or don't hear the difference then the internal one is OK
[14:04] <Zeeflo> you cant hear any difference at stereo level anyways
[14:04] <JEEB> you can
[14:04] <Zeeflo> u got super ears then
[14:04] <JEEB> no
[14:04] <JEEB> I've got bad ears
[14:05] <Zeeflo> well, im not trying to create THX 7.1 lossless sound here..
[14:05] <JEEB> I have good eyesight for video compression artifacts and bad ears, and I can still easily hear that when you happen to drop below the comfort level of libavcodec's aac encoder, it's worse than fdk-aac here
[14:05] <JEEB> and the comfort level is probably somewhere around 128k
[14:06] <JEEB> although when I had to use the libavcodec encoder I kept to 192k just in case
[14:06] <Zeeflo> you couldnt hear the difference in 128k to 160 or 192k
[14:06] <Zeeflo> no one can
[14:06] <JEEB> oh yes you can with the libavcodec encoder
[14:06] <Zeeflo> dude come on..
[14:06] <JEEB> it's better than the libavcodec vorbis encoder, that's for sure
[14:07] <JEEB> probably better than the libavcodec wma encoder, too
[14:07] <JEEB> I know that there are encoders where at those rates you really shouldn't be able to distinguish stuff, but the libavcodec's encoder is not one of those
[14:08] <Zeeflo> then you probablay got Monster cables with 24k gold all the way through, with oxydized this and that priced at $1000 pr inch on a $200.000 home stereo
[14:08] <JEEB> ...
[14:08] <Zeeflo> the human ear cannot hear the difference
[14:08] <ledil> JEEB, fdk-aac, do I need an additional package to install ?
[14:08] <JEEB> yes, you would have to compile it as well
[14:08] <Zeeflo> you gotta compile it
[14:08] <ledil> or is it only "-acodec fdk-aac" ?
[14:09] <JEEB> it is nonfree as well because of the crappy license fraunhofer put on the source
[14:09] <ledil> JEEB, is there binary package ?
[14:09] <JEEB> so it cannot be distributed either
[14:09] <JEEB> in binary
[14:09] <JEEB> (linked with (L)GPL at least)
[14:09] <Zeeflo> Whats your suggestion instead of bureks aac then?
[14:09] <Zeeflo> ill make a sound test
[14:09] <Zeeflo> a clinic sound test
[14:10] <JEEB> fdk-aac with afterburner on, start around 96k and go up from there with various samples
[14:10] <Zeeflo> @ constant bitrate
[14:10] <Zeeflo> @128k
[14:10] <Zeeflo> vs aac
[14:10] <Zeeflo> ill do that
[14:11] <JEEB> that's a bit on the high'ish side for fdk-aac probably, but with many sources the libavcodec aac encoder should start showing its skin there
[14:11] <Zeeflo> and ill show you the results
[14:12] <Zeeflo> crap its cold here
[14:12] <JEEB> also please do not compare me with the so-called "audiophiles", I'm all for valid tests and such, and I am just saying that the libavcodec encoder just sucks at its job since no-one finished it >_>
[14:12] <Zeeflo> im freezing
[14:12] <JEEB> if someone would have the time to put into the libavcodec encoder it'd be great
[14:12] <JEEB> unfortunately, it's simpler in most cases for people to just compile fdk-aac and use it locally
[14:13] <Zeeflo> people probably just think like me: It works well enough, leave it at that.
[14:13] <Zeeflo> also
[14:13] <Zeeflo> its the downside of GPL
[14:13] <Zeeflo> no-one can force any-one to finish :D
[14:13] <JEEB> more like open source
[14:13] <JEEB> not specific to GPL in any way
[14:14] <JEEB> or all software to be honest
[14:14] <Zeeflo> i stand corrected, but you know what I mean
[14:14] <JEEB> even with payware there are cases where a developer has stopped developing the app and no matter how much the clients derp he won't finish it
[14:14] <JEEB> ("finish" as in fix the bugs etc.)
[14:14] <JEEB> also, if you really want a laugh
[14:15] <JEEB> try the "ogg" encoder
[14:15] <JEEB> not libvorbis, make sure it uses the internal one
[14:15] <JEEB> I guarantee you will have a laugh :)
[14:15] <JEEB> so compared to that, the aac encoder is much better
[14:18] <Zeeflo> h264 wont work with ogg would it?
[14:19] <Zeeflo> in a mp4 container
[14:19] <JEEB> yes, because it's a container that doesn't specify a way to mux vorbis in it. It has nothing to do with H.264 :)
[14:19] <Zeeflo> i use libx264 with aac for mp4 containers, for streaming (rtmp)
[14:19] <Zeeflo> flash and html5 fallback
[14:20] <Zeeflo> i dont think ogg would work with that?
[14:20] <JEEB> I'm not saying you should use it
[14:20] <Zeeflo> would flash eat that?
[14:20] <JEEB> I just said that if you were going to try out aac and fdk-aac, you might as well quickly encode the same thing with the vorbis encoder
[14:20] <Zeeflo> well. that would depend on the player..
[14:20] <JEEB> just for laughs :P
[14:20] <JEEB> also rtmp(e) is flv
[14:20] <JEEB> not mp4
[14:21] <Zeeflo> nah, im using mp4
[14:21] <JEEB> of course your rtmp(e) service can take in already encoded mp4
[14:21] <JEEB> and remux that on the fly
[14:21] <JEEB> into flv
[14:21] <JEEB> just that live streaming doesn't fly with that
[14:21] <JEEB> unless your mp4 muxer suddenly supports movie fragments
[14:21] <Zeeflo> it does
[14:21] <Zeeflo> -movflags
[14:22] <Zeeflo> +faststart
[14:22] <JEEB> uhhh
[14:22] <JEEB> that's not it
[14:22] <JEEB> that's just putting the index in the front /after/ encoding
[14:22] <Zeeflo> do you mean somthing like qtindexing then?
[14:23] <Zeeflo> moov atoms?
[14:23] <JEEB> no, I mean like something like L-SMASH movie fragments feature. the libavformat demuxer supports it, but muxer AFAIK doesn't
[14:23] <JEEB> moov atom moving is moving the index
[14:23] <JEEB> can only be done /after/ encoding
[14:23] <Zeeflo> i dont know about movie fragtments..
[14:23] <Zeeflo> My setup works..
[14:23] <JEEB> yes, because your thing most probably just reads the mp4 and remuxes it to flv on the fly when feeding it to rtmp(e)
[14:23] <Zeeflo> MP4's on amazons CDN (cloudfront) played with JW player 6.1
[14:24] <JEEB> I'm just telling you that rtmp(e) by itself is flv
[14:24] <muken> JEEB: ? libavformat muxer supports movie fragments
[14:24] <JEEB> muken, oh?
[14:24] <JEEB> I've just never seen them used
[14:24] <JEEB> lol
[14:32] <JEEB> Zeeflo, btw if you really want to see what the CDN is doing to your content, you can point ffmpeg to the rtmp(e) URL the player reads (ffmpeg -i rtmp://url:port ) and see what format etc. it sees coming towards you
[14:32] <JEEB> I would guess it just remuxes your mp4 files into flv and transports them like that
[14:33] <Zeeflo> i would have to make the file public in the bucket then wouldnt i?
[14:33] <JEEB> nope
[14:33] <JEEB> at least I have no idea what you're talking about
[14:33] <Zeeflo> heh
[14:33] <Zeeflo> i know I would have to
[14:33] <Zeeflo> :)
[14:33] <JEEB> if you are talking about pastebinning, then no -- you can clear out the rtmp(e) url or whatever from the command line and so forth
[14:34] <JEEB> and the rtmpe handshake
[14:34] <Zeeflo> nono, I am using signed and timed urls for my content
[14:35] <JEEB> anyways, just trying to tell you that it pretty surely isn't mp4 when it gets pushed through rtmp(e)
[14:35] <Zeeflo> ill make a file public and see :)
[14:36] <JEEB> sure, ffmpeg should be able to read what's coming up from the tube with an rtmp:// url as input
[14:36] <Zeeflo> BTW
[14:37] <Zeeflo> bureks build has a bug
[14:37] <Zeeflo> how do I submit it?
[14:37] <JEEB> depends on if it's a bug in the build or bug in ffmpeg itself
[14:37] <Zeeflo> i dont know..
[14:37] <JEEB> if you still get the bug with building the same revision yourself f.ex. then you poke bug burek
[14:38] <JEEB> otherwise your poke the ffmpeg trac issue tracker
[14:38] <Zeeflo> when you select subtitles=blabla.srt, if theres no fonts on the server, it doesnt warn you
[14:38] <Zeeflo> server=system
[14:39] <Zeeflo> it did that on my old server as well
[14:39] <Zeeflo> I had to get the ms core fonts
[14:39] <Zeeflo> and install fontconfig
[14:39] <Zeeflo> and then I had subs in my stream
[14:39] <JEEB> so if the system lacks any fonts of any description it just fails to render anything instead of failing with an error?
[14:39] <Zeeflo> before I did that, it acted like it was creating a new stream with subs in it, but it wasnt
[14:39] <JEEB> uhh
[14:40] <Zeeflo> an no warning/error messages
[14:41] <Zeeflo> with everything else, ffmpeg is pretty good at warning you or even halting
[14:41] <Zeeflo> just not with this!
[14:41] <Zeeflo> atleast for the debian build
[14:43] <JEEB> that sounds like a ffmpeg bug, but you might want to check how it handles itself in case you build a build with current ffmpeg with libass etc. yourself, and then try it out like that. Although I feel in most cases like that the build would then depend on shared libraries and would actually derp at you for trying to start it up without f.ex. fontconfig to begin with (Before ffmpeg's code actually runs)
[14:44] <JEEB> burek's build should have everything in it static methinks, so even if you don't have the stuff around, it would still kind of go forward
[14:44] <JEEB> naturally lacking any font cache or anything you wouldn't really get anything
[14:45] <JEEB> so I guess that some checks might be looked into, but depending on how "dumb" the checks would have to be done, or where they would have to be done (ffmpeg? or fontconfig? or something else?) it could probably take time to see it get completely "fixed"
[14:46] <Zeeflo> yea
[14:46] <JEEB> for example, if it's just a case of lax checking in fontconfig
[14:46] <JEEB> even if you fix it right now
[14:46] <JEEB> it won't fix machines already out there
[14:46] <Zeeflo> one just gotta remember to install fonts on his/her new server if they want subs in their streams! Otherwise they might gotta start all over!
[14:46] <Zeeflo> <--- like he had to!
[14:47] <Zeeflo> :D
[14:47] <JEEB> I generally build my stuff myself so I generally grab all the related libraries for that
[14:47] <Zeeflo> i tried that as well
[14:48] <Zeeflo> but I couldnt get it to support SRT subs
[14:48] <Zeeflo> I had to convert everything to ass
[14:48] <Zeeflo> untill I got Bureks build
[14:48] <JEEB> sure you were building everything new enough?
[14:48] <Zeeflo> Besides, I dont think building it myself would make it any bit better!
[14:48] <Zeeflo> a wise man said, if it works dont fuck with it
[14:49] <Zeeflo> and his build works
[14:49] <Zeeflo> atleast for me
[14:50] <JEEB> well, it would at least make you not forget fontconfig, fonts you'd still have to grab somewhere but I'd guess in most cases those might be grabbed as a dependency?
[14:50] <JEEB> also I'm just used to building ffmpeg and friends by now :V
[14:51] <JEEB> I can see why burek's builds are quite useful for many people tho
[14:51] <JEEB> they offer an easy way out for those who don't know their ways to compile the stuff they need
[14:51] <Zeeflo> i just install it directly
[14:52] <Zeeflo> fontconfig isnt a dependency of msttcorefonts
[14:52] <Zeeflo> but its needed for ffmpeg
[14:52] <JEEB> I know it isn't :|
[14:53] <JEEB> also I was talking of it the other way, generally grabbing fontconfig/freetype might end you up with some fonts too in many cases -- I might be completely incorrect tho naturally
[14:54] <Zeeflo> im still missing fonts though
[14:54] <Zeeflo> ffmpeg is still complaining, but atleast it can fallback to arial
[14:54] <Zeeflo> it want Deva Vu
[14:55] <Zeeflo> im perfectly fine with arial
[14:59] <Zeeflo> JEEB, is there also a command to level the volume fixed?
[14:59] <Zeeflo> ive noticed some video files has high volume
[14:59] <Zeeflo> some has moderate
[14:59] <Zeeflo> some almost has no volume :D
[15:04] <durandal_1707> volume , volumedetect filter
[15:05] <durandal_1707> they are all listed in filter documentation available on web
[15:07] <Zeeflo> i must have missed something i wrote earlier
[15:07] <Zeeflo> ;)
[15:11] <ubitux> ebur128 is likely more effective than volumedetect
[15:12] <ubitux> (but slower)
[15:28] <_kud> Hello
[15:29] <_kud> I'm trying to do a video with some png; i've got this script= https://gist.github.com/1e0fdf400961e7e5d779
[15:29] <_kud> but it seems not to loop it 20 times
[15:29] <_kud> do you have any idea how to make a loop properly?
[15:51] <undercash> hello
[15:52] <undercash> i m trying to compile ffmpeg on ubuntu precise 64 and i get a ERROR: libx264 not found
[15:52] <undercash> which didnt happen for a very long time  :)
[15:52] <undercash> oops 32bit sorry, testing on laptop
[15:53] <JEEB> libx264-dev package is not installed?
[15:53] <JEEB> if you want to use the package that is
[15:53] <JEEB> if you built libx264 yourself then make sure you had either --enable-static or --enable-shared set
[15:53] <undercash> yea it is installed
[15:54] <JEEB> check config.log then?
[15:54] <undercash> ok let's put it
[15:54] <hackeron> hey, I'm trying to record from a granstream video server, I'm doing: ffmpeg -i rtsp://admin:admin@192.168.0.214:554 -analyzeduration 0 -codec:v copy -codec:a libfaac -flags +qscale -global_quality 1 -afterburner 1 test.mkv --- that gives me broken metalic sound. If I switch libfaac to libmp3lame then I get good audio but it is not in sync with the video. Any ideas?
[15:54] <JEEB> lol
[15:54] <JEEB> so which is it?
[15:54] <undercash> the --enable-static  i mean
[15:54] <undercash> but note that it wasn't necessary like 2 weeks ago
[15:55] <JEEB> undercash, so which is it -- are you building with a pre-built binary package of libx264 or are you building it yourself?
[15:55] <undercash> myself
[15:55] <JEEB> if you aren't building it yourself you don't add that
[15:55] <JEEB> oh
[15:55] <undercash> following the well known topic, i want to add libass
[15:55] <JEEB> so you had neither --enable-static or --enable-shared on your x264 configure line?
[15:55] <undercash> seems interesting
[15:55] <undercash> neither yes, but before it wasn't a problem
[15:55] <undercash> if i remember well
[15:55] <JEEB> that was quite a long time ago then
[15:55] <JEEB> not weeks
[15:56] <undercash> 2 weeks ago
[15:56] <JEEB> no, that wasn't the up-to-date x264 then
[15:56] <JEEB> I know for at least a year if not more x264 has needed one of those
[15:56] <JEEB> or it would not install the libraries
[15:56] <undercash> i guess it was since i had to update yasm too ;)
[15:56] <JEEB> no
[15:56] <JEEB> yasm update to 1.2 was new
[15:56] <JEEB> that was added for the possible addition of haswell asm
[15:56] <undercash> anyway thx you i got to recompile x264
[15:57] <undercash> not really useful to debate if it s a recent update or not..
[15:57] <JEEB> you might as well do --disable-cli too
[15:57] <XATRIX> Guys, i need some advices... I'm trying to catenate a few mp3 files and, result file is not readable on some players... I'm interested in JS players, and chrome player. Can someone help me to debug ?
[15:57] <JEEB> because you only need the library for ffmpeg
[15:57] <JEEB> not the command line encoder
[15:58] <undercash> i came here coz i thought maybe something was broken in the ffmpeg git
[15:59] <undercash> ok JEEB thx, i m recompiling
[15:59] <undercash> yes i never use x264 as standalone
[16:00] <hackeron> My input audio is pcm_mulaw, 8000 Hz, mono, s16, 64 kb/s from RTSP - any ideas why libfaac would generate broken, metalic sounding audio while libmp3lame sounds fine? - Anything I can do in ffmpeg like maybe force a higher sample rate for the output audio?
[16:02] <Mavrik> hackeron: it's quite possible libfaac doesn't support that sample rate
[16:02] <Mavrik> hackeron: you can try using libfdk_aac, which is pretty much currently the best ffmpeg supported AAC encoder
[16:02] <Mavrik> hackeron: or try resampling the audio to 22050 or some such
[16:04] <JEEB> undercash, http://git.videolan.org/?p=x264.git;a=commit;h=c1e60b9032196d204db8dce77051360e403a1d2f
[16:04] <JEEB> it was may 2011
[16:04] <JEEB> where it changed
[16:04] <JEEB> everything from this commit onwards needed --enable-static or --enable-shared
[16:06] <hackeron> Mavrik: I tried: ffmpeg -i rtsp://admin:admin@192.168.0.214:554 -analyzeduration 0 -ar 44100 -codec:v copy -codec:a libfdk_aac -async 1 -flags +qscale -global_quality 1 -afterburner 1 test.mkv -- now it sounds OK but it is out of sync with the video
[16:06] <hackeron> Mavrik: it stays in sync if I use libmp3lame though :/
[16:07] <Mavrik> *grumbl* :)
[16:07] <Mavrik> hackeron: try without resampling when u sing fdk_aac
[16:07] <hackeron> Mavrik: then it dies saying: [libfdk_aac @ 0x7f8ce88b2c00] Unable to encode frame: Encoding error
[16:07] <hackeron> Audio encoding failed (avcodec_encode_audio2)
[16:08] <undercash> did the job, thx jeeb
[16:11] <JEEB> np
[16:14] <antonello_> Can I obtain an decoder without reading the file header?
[16:14] <hackeron> Mavrik: any ideas?
[16:14] <Mavrik> not really, shouldn't go out of sync :\
[16:15] <Mavrik> does your input get probed properly (your analyzeduration is awfully short?)
[16:15] <Mavrik> antonello_: yep
[16:16] <hackeron> Mavrik: hmm, not sure - but changing libfdk_aac to libmp3lame in the above command I get audio/video sync
[16:16] <hackeron> and changing the libfaac I get broken metalic sound
[16:17] <Mavrik> antonello_: just call avcodec_find_encoder with right encoder ID
[16:18] <antonello_> i do this ...
[16:19] <Mavrik> good.
[16:20] <antonello_> I obtain a AVCodecContext ... and after i setting  the codec parameters
[16:21] <antonello_> channels sample_rate bit_rate sample_fmt channel_layoyt
[16:21] <antonello_> my error is
[16:21] <antonello_> [aac @ 0x10bd800] Not evaluating a further program_config_element as this construct is dubious at best.
[16:21] <antonello_> [aac @ 0x10bd800] channel element 1.7 is not allocated
[16:22] <antonello_> I don't understand where i miss .
[16:23] <hackeron> Mavrik: same story with libvo_aacenc - it seems to be about 1.5 seconds out of sync just like libfdk_aac - but perfect sync with libmp3lame. Does ffmpeg calculate duration different with libmp3lame?
[16:23] <Mavrik> hackeron: hmm... wouldn't know, but duration has little to do with it :\
[16:24] <antonello_> this is my code http://pastebin.com/iUGmCq6J line  184 -190
[16:25] <Mavrik> antonello_: um
[16:26] <hackeron> Mavrik: any ideas why libmp3lame would be in sync and libfdk_aac and libvo_aacenc audio is around 1.5 seconds behind the video?
[16:26] <Mavrik> antonello_: AVFormatContext won't have its fields filled out if you don't read the file header
[16:26] <Mavrik> antonello_: which also means your ids and other data will be wrong
[16:26] <Mavrik> hackeron: no idea really... that shouldn't happen :\
[16:26] <Mavrik> hackeron: did you try using -async 22050 or something like that
[16:28] <antonello_> then.... what should I use?
[16:29] <Mavrik> either probe the file to fill out those fields
[16:30] <Mavrik> or don't read them at all
[16:30] <Macey> Hi All, who looks after the API?
[16:31] <Mavrik> O.o
[16:31] <hackeron> Mavrik: just tried: ffmpeg -i rtsp://admin:admin@192.168.0.214:554 -analyzeduration 0 -ar 44100 -async 44100 -codec:v copy -codec:a libfdk_aac -flags +qscale -global_quality 1 -afterburner 1 -y test.mkv -- no change, audio is the same amount behind video - libmp3lame still in sync :/
[16:34] <Diogo> hi i need to capture sound from a blackmagic card using ffmpeg, i don't have sound input...anyone with the same problem?
[16:46] <antonello_> Can I obtain an header to add an head of my streeam ?
[16:49] <hackeron> I'm reading audio/video from an RTSP source. I have -ar 44100 -async 1 and -flags +qscale - when I use -codec:a libmp3lame - audio and video are in sync, however when I switch to libfdk_aac or libvo_aacenc the video is about a second ahead of the audio. Any ideas?
[16:53] <hackeron> ah wait, I tried to create an mp4 file and now the libmp3lame is out of sync by the same amount, hmm
[17:06] <hackeron> ah, I had to move -async 1 to be after -codec:a libmp3lame - then I get audio/video sync with libmp3lame - but not with libfdk_aac or libvo_aacenc -- is -async 1 not supported with those?
[17:30] <Youka> I'm a bit confused about avcodec_decode_video2 and the auto allocated bitmap inside AVFrame. In the examples, AVFrame bitmap is never freed after decoding, so is it a memory leak (which is ignored because it's small and the program exits after it)?
[17:30] <teratorn> Youka: run some tests under valgrind
[17:31] <jeje34> hi to all ;-)
[17:33] <jeje34> I'm using the last release version of FFMPEG (1.1.1) and it seems the function av_lockmgr_addref and av_lockmgr_release doesn't exist anymore. Before, I was using AVCodec 5.2, and these functions were existing
[17:35] <jeje34> sorry I was disconnected
[17:35] <jeje34> I use ffmpeg to decode h264 video in a multithread appication
[17:40] <jeje34> and in time, I've the error : Insufficient thread locking around avcodec_open/close() whith a crash of my application
[17:43] <jeje34> when I was using avcodec52.dll, i call if (av_lockmgr_addref() == 1)
[17:43] <jeje34>             { av_lockmgr_register(&ff_lockmgr);
[17:43] <jeje34>             }
[17:44] <jeje34> with static int ff_lockmgr(void **mutex, enum AVLockOp op) and casting mutex to CRITICAL_SECTION (I'm on windows envirronment)
[17:45] <jeje34> but with the last version, the av_lockmgr_addref doesn't exist
[18:03] <jeje34> what happen if several thread call av_lockmgr_register(&ff_lockmgr)
[18:03] <jeje34> ?
[19:19] <JoeyJoeJo> I have an mkv file that won't play and I'm not sure why. Can ffmpeg repair the file or at least let me know what the problem is?
[19:20] <klaxa> probably, try running it through ffprobe
[19:20] <klaxa> also some more information on the exact way you can't play it would be great
[19:20] <klaxa> i.e. what player and what error message
[19:21] <JoeyJoeJo> ffprobe seemed to work. It said "Invalid data found when processing input"
[19:21] <klaxa> if you torrented the file, verify your local data, maybe the header is corrupted, i have similar problems from time to time
[19:22] <klaxa> mplayer2/libavformat interprets the file as DV Video and starts playing back garbage
[19:22] <JoeyJoeJo> I'm trying to play it with VLC and I don't see any picture or hear any sound. The seek bar doesn't even count. And I can't find any errors either
[19:23] <JoeyJoeJo> I did torrent it. so I try verifying my local data
[21:10] <llogan> burek: perhaps the IMPORTANT note on gusari forum should be even bigger.
[21:26] <bparker> is there a way to embed exact pixel colors at certain positions within a compressed video? For example I have a compressed video and I want the first pixel of the video to be 100% blue. It doesn't matter if the video has to be re-encoded, that's fine. Any idea if that's possible?
[21:27] <bparker> I'm worried about the compression messing with the color values themselves
[21:28] <klaxa> depends on the codec i guess, but yeah compression will probably mess with that stuff
[21:29] <klaxa> lossly compression that is
[21:35] <bparker> klaxa: right, that's what I'm trying to work with unfortunately
[21:35] <bparker> I need some markers placed on each corner of a compressed video
[21:35] <bparker> with specific pixel values because I'm essentially encoding auxillary data there
[21:36] <klaxa> can't you work with approximate values?
[21:36] <bparker> sortof like steganography, but really small bits of data
[21:36] <bparker> it's possible, but I'm worried about false positives
[21:36] <bparker> if the video happens to contain a color in the range I'm looking for, that's bad
[21:37] <klaxa> mh yeah
[21:37] <bparker> and I don't want to visibly alter the video enough that it's distracting to users
[21:37] <bparker> becaue the displays can see 100% of the pixels
[21:37] <bparker> what I'm basically doing is
[21:38] <bparker> taking a video signal in from a device, capturing it and doing image analysis on it
[21:38] <bparker> then based on that analysis, doing different kinds of post-processing on the image and outputting it to a TV
[21:38] <bparker> analysis being some bits I want set that tell me what to do with the video
[21:39] <bparker> and the only way I have to communicate that is with the actual video data (pixels) itself
[21:40] <klaxa> i don't think you'll be able to achieve that unless you use lossless codecs or settle with approximation
[21:40] <klaxa> or establish a second independent data channel of some sort
[21:40] <bparker> that's what I thought
[21:41] <bparker> unless I can come up with some kind of scheme/algorithm that is resistant to the compression, that's really my only option
[21:42] <klaxa> and you have to use the videostream? why not try to code it in binary at the top or bottom and analyze that within some tolerance range?
[21:42] <klaxa> like black and white pixels
[21:42] <klaxa> you should be able to differenciate black and white even when compressed
[21:43] <klaxa> although it may put strain on the codec because hard edges are hard to compress
[00:00] --- Fri Feb  1 2013