[Ffmpeg-devel-irc] ffmpeg.log.20120720

burek burek021 at gmail.com
Sat Jul 21 02:05:01 CEST 2012


[00:23] <Sashmo> is it possible to use ffmpeg to send to multiple destination?  similar to how vlc has the ability to do bith UDP and HTTP at the same time
[00:25] <saste> Sashmo: just specify different outputs
[00:25] <saste> in the same command
[00:26] <Sashmo> hmm, ok Ill give it a try
[00:32] <tibo> can I get the total number of frames in a video file (using c++) ?
[01:13] <SubJunk> I've compiled FFmpeg successfully, but it wants the third-party DLLs like libfreetype-6.dll, libcdio.dll, etc. FFmpeg was compiled as static as were the libs. Any ideas on how I can make it use the dependencies internally instead of needing the external DLLs?
[01:33] <Freakshow> anyone know off their hip which .h file I would be looking at for rtmp related structs?
[01:34] <Freakshow> or rtsp for that matter
[01:46] <flecha> Hi! I am trying to convert an Ogg file to Mp3, but I am losing the metadata. I tried "ffmpeg -i a.ogg -map_metadata 0,s0:0 a.mp3", it copies the album and artist data, but not the  comment
[01:46] <flecha> any ideias why?
[02:05] <burek> !fb flecha
[02:06] <burek> haha
[02:06] <burek> I need some sleep, obviously :)
[02:07] <burek> Sashmo, using 2 outputs in 1 ffmpeg's cmd line will create 2 encoding processes (threads) and double your cpu usage
[02:07] <burek> at least it was like that so far
[02:07] <burek> if you just need to duplicate your output (the same stream) to several points, you can use something like this
[02:08] <burek> ffmpeg -i input -vcodec ... -acodec ... -f mpegts - | ffmpeg -f mpegts -i - -vcodec copy -acodec copy output1 -vcodec copy -acodec copy output2
[02:08] <burek> the 1st ffmpeg process will only do the encoding part
[02:08] <burek> and 2nd ffmpeg will just copy that to 2 outputs
[02:30] <Sashmo1> hey everyone, is there a way to buffer the input source of a streamlike its done in vlc??
[02:39] <Sashmo> its ok, I found it, -bufsize
[02:48] <creaux> Sashmo: why do you want to buffer?
[02:49] <Sashmo> I was trying to add a buffer to the input
[02:50] <creaux> I am trying to understand which problem you will solve
[02:51] <Sashmo> well, I thought I was having a problem with the input beacause I keep getting a clipping sound on the output stream, my source is h264 video with mp3 audio, and Im trying to convert to h264 with aac audio,  I matched the audio frequency and bitrate, but I keep getting this clip, like a crackle
[02:52] <Sashmo> almost like the volume is too high
[02:52] <Sashmo> but it isnt
[02:52] <creaux> ok, I dont think it has to do with buffering for ffmpeg
[02:52] <Sashmo> lol I was trying everyting
[02:53] <creaux> are you sure the source is fine
[02:53] <Sashmo> yup
[02:54] <Sashmo> I can listen to the source
[02:54] <creaux> which aac codec do you use?
[02:54] <Sashmo> no problem
[02:54] <Sashmo> aac
[02:54] <Sashmo> and
[02:54] <Sashmo> the other
[02:54] <Sashmo> libafaac
[02:54] <Sashmo> you know......
[02:54] <Sashmo> it could be that the input is too loud
[02:54] <creaux> I have read that libfaac is buggy
[02:55] <Sashmo> is it possible to lower it at the input?
[02:58] <creaux> dought it is volume, then it would also be too loud when you play the source
[02:59] <Sashmo> it does sound a bit loud
[02:59] <creaux> there is a AAC converter called VisualOn, there is a build for windows which uses that one
[03:00] <creaux> http://ffmpeg.zeranoe.com/builds/
[03:03] <Sashmo> wow that was it
[03:03] <Sashmo> it was the dam volume
[03:03] <Sashmo> ah nooo its there
[03:03] <creaux> :)
[03:04] <flecha> burek, hi! here is the console output: http://pastebin.com/zgtWNdeG
[03:05] <flecha> lol
[03:18] <Spice_Boy> Hi. Is is possible to feed the output of rtmpdump to ffmpeg (which feeds to ffserver), or even just directly to ffserver?
[11:32] <kristian-aalborg> cheers
[11:32] <kristian-aalborg> trying to join two images into a movie file, but it just skips by
[11:32] <kristian-aalborg> I have tried changing the -r parameter, but no luck
[11:34] <kristian-aalborg> for instance, "ffmpeg -qscale 5 -r 20 -b 9600 -i file1.jpg file2.jpg movie.avi"
[11:36] <burek> how do you expect 2 images to be read at the speed of -r 20 ?
[11:36] <burek> you dont have 40 images, right?
[11:37] <burek> also, that is not the proper command, you need -f image2
[11:37] <burek> see the tutorials section
[11:37] <burek> http://ffmpeg.org/trac/ffmpeg/wiki/Create%20a%20video%20slideshow%20from%20images
[11:38] <burek> (btw, you have overwritten your file2.jpg with your original command)
[11:48] <kristian-aalborg> thanks
[12:28] <rohankasat> hi , can we stream .pls or .m3u ( playlist files) directly from console using ffplay ?
[13:38] <rohankasat> can we stream .pls or .m3u ( playlist files) directly from console using ffplay ? or how can we parse it , and find the streaming url's or streaming link's inside the playlist file ?
[13:44] <flecha> Hi! I am trying to convert an Ogg file to Mp3, but I am losing the metadata. I tried "ffmpeg -i a.ogg -map_metadata 0:s:0 a.mp3", it copies the artist data, but not the comment. Any ideas why? http://pastebin.com/zgtWNdeG
[16:53] <juanmabc> hello, i'm very happy of the result of my opengl/openal ffmpeg binding, though it is still not cpu accelerated (sse...) and always  perform rescale and resample, if you find this work useful, you could give advice to the developer to improve to fit your needs: http://code.google.com/p/openmedialibrary (mainly sse and swscale swresample stuff)
[17:06] <juanmabc> i'd like to know the key features to be data aligned, like if i need to use DECLARE_ALIGNED, or something else (i thought using av_malloc i was aligned safe, turns out something make it not), details are in just one file src/mlMedia.c
[17:19] <juanmabc> should i go #ffmpeg-devel if it is too specific coding and not command line usage?
[17:25] <ubitux> i don't see any question :)
[17:25] <ubitux> you want to use swscale and swresample?
[17:27] <juanmabc> ubitux: key features for data alignment
[17:27] <juanmabc> s/features/usage/
[17:28] <juanmabc> i'm already using swscale and swresample
[17:28] <juanmabc> just not data aligned "which could lead (and leads) to a speedloss"
[17:29] <juanmabc> using av_malloc/av_free does not cut it
[17:30] <juanmabc> despite the purpose
[17:33] <mulicheng> Morning all.  I'm embedding the h264 decoder into our application.  It works fine but I wonder if I can tweak it to better serve our purpose.  I use it in a voip application and I need feedback from the decoder if it had errors (mainly from packet loss).
[17:34] <mulicheng> The only err_recognition flags I saw used in the h264 decoder were AV_EF_AGGRESSIVE and AV_EF_EXPLODE.
[17:34] <mulicheng> I'm not sure explode is what I want.
[17:35] <mulicheng> More like, something if EXPLODE were set, to just set a flag instead that I could check.
[17:52] <arpu> http://paste.railsbox.eu/show/628/
[17:52] <arpu> anyone an idea why librtmp is not found ?
[17:56] <mulicheng> CODEC_FLAG_PSNR?
[18:10] <Belial_> I'm trying to convert an mp4 to mp3 via ffmpeg but it refuses to work.  Running Mageia 2 and when I try to do the coversion I get "Encoder (codec none) not found for output stream #0:0"  How does one get ffmpeg to convert an mp4 to an mp3?
[18:11] <spaam> what is the commandline you are using ?
[18:11] <sacarasc> Belial_: Can you paste your complete command and output to a pastebin?
[18:11] <spaam> sacarasc: \o/
[18:11] <sacarasc> HI SPAAM!
[18:12] <spaam> sacarasc: WZUP?!
[18:12] <sacarasc> I'm kinda tired and got a headache. \o/ How about with you?
[18:12] <Belial_> command: ffmpeg -b 192k -i input.mp4 output.mp3
[18:12] <spaam> im at work :D im thinking about going home at watch futurama and eat some kebab :D
[18:13] <spaam> Belial_: the output plix? :D
[18:13] <spaam> Belial_: put it on a pastebin site :)
[18:13] <Belial_> Ah...wait...nevermind.  Somewhere in my messing around just now I managed to get a "tainted" codec installed by chance that makes it work now.
[18:13] <sacarasc> Belial_: Try ffmpeg -i input.mp4 -b:a 192k output.mp3
[18:14] <spaam> Belial_: did you install ffmpeg with libmp3lame? :D
[18:15] <Belial_> I did but didn't have the "tainted" versions it seems. Now I do and it converted.
[18:15] <spaam> ffmpeg -i file.mp4 -vn -c:a libmp3lame -b:a 192k awesomeoutput.mp3 :D
[18:15] <Belial_> spaam:  I'll give that one a try. thx
[18:16] <spaam> oh what an ugly logo on ffmpeg.org ;S
[18:16] <spaam> -vn is for disable video stuff :)
[18:17] <spaam> that commandline should work(tm)
[18:17] <spaam> sacarasc: did you like the kim dotcom song? :D
[18:17] <sacarasc> I didn't click the link yet... I'll go do it now.
[18:41] <swick> I want to encode raw rgba frames to yuv4mpeg in a c programm. Is there existing code I can take a look at?
[18:45] <Belial_> Can someone explain why the ffmpeg command in an executable script but does on the CLI? Bash script fail: #!/bin/sh
[18:45] <Belial_> for file in $@ ; do name=`echo $file | sed -e "s/\....$//g"` ffmpeg -i $fname -vn -c:a libmp3lame -b:a 192k $name.mp3 done
[18:46] <Belial_> When I run it from script I get "-vn no such file or directory" Huh?
[18:49] <iive> use `bash -x scrip.sh` to find out.
[18:49] <iive> most probably $fname needs another set of \" \"
[18:50] <iive> or just "$fname"
[19:01] <beandog> it's skipping $fname
[19:01] <beandog> must be empty
[19:01] <beandog> besides, you set it to name, not fname
[19:15] <hwkxx> way of merging videos?
[19:15] <hwkxx> without transcoding them first, because its lossy
[19:16] <hwkxx> saw that bash script which merges them with mfifo
[19:16] <hwkxx> that doesnt work :|
[19:33] <juanmabc> http://fpaste.org/c2s8/
[19:42] <juanmabc>  /bin/sh is not bash features
[19:43] <juanmabc>  /bin/bash is
[20:15] <_Nike_> My program makes a TCP connection with the RTP, and I need to send the buffer that I get to ffmpeg to parse. Parse the H264 stream bytes. Because they are fragmented. I need to separate the NALS. How can I do this without FFMPEG connect in RTSP? If my program connects and receives packets.
[21:09] <passstab> hi
[21:09] <passstab> i'm trying to follow http://jeremyblythe.blogspot.co.uk/2012/05/raspberry-pi-webcam.html
[21:10] <passstab>  but i get "codec for stream 0 does not use global headers but container format requires global headers"
[21:10] <passstab> what am is wrong with the instructions?
[21:11] <passstab> what is wrong with the instructions?
[21:11] <passstab> :P
[21:22] <wizonesolutions> Is there a good way to detect whether ffmpeg needs to be passed -vpre or -preset?
[21:22] <wizonesolutions> my local needs -preset to work but the server needs -vpre :/
[21:22] <wizonesolutions> should I just try one, and if it errors out then use the other or can I ask ffmpeg about its capabilities somehow?
[21:23] <JEEB> you should be able to see if it uses internal libx264 presets or separate preset files by the version
[21:23] <JEEB> and AFAIK ffmpeg should output the version number in pretty much any case
[21:23] <wizonesolutions> JEEB: what if one of them is compiled from Git?
[21:23] <JEEB> I still think they all have version numbers
[21:23] <wizonesolutions> it just says git-2012-<etc>
[21:23] <JEEB> oh
[21:24] <JEEB> it doesn't say 0.11 or anything?
[21:24] <wizonesolutions> JEEB: no, just ffmpeg version git-2012-07-18-c66978e Copyright (c) 2000-2012 the FFmpeg developers
[21:24] <JEEB> :|
[21:24] <JEEB> I guess either version, or a git date
[21:24] <wizonesolutions> cuz I'm on Lucid, which has a pretty old ver
[21:25] <wizonesolutions> dunno if there is a PPA with a newer one
[21:25] <JEEB> it's better to build a newer one by yourself
[21:25] <wizonesolutions> the server has 0.7.1
[21:25] <wizonesolutions> or rather 0.7
[21:25] <wizonesolutions> yeah
[21:25] <wizonesolutions> that's why I wound up doing so
[21:27] <JEEB> I think ubuntu/debian now use libav, and the more up-to-date binary on those systems is 'avconv', because libav deprecated the ffmpeg command line app
[21:27] <JEEB> (ffmpeg just merges the updates made into avconv into the ffmpeg command line app)
[21:27] <passstab> is there a codec other then msmpeg4 that i should use for asf streaming?
[21:28] <JEEB> passstab, the wmv ones? But I really don't know how good the libavcodec implementations of those formats are
[21:29] <JEEB> wizonesolutions, avconv 0.7.x should have the libx264 presets via -preset, too
[21:29] <JEEB> (the 11.10 one)
[21:30] <JEEB> (and newer)
[21:30] <JEEB> at least I think so
[21:30] <wizonesolutions> JEEB: Yeah, but the server says it doesn't understand -preset. It's CentOS, whatever DreamHost is running
[21:30] <JEEB> oh, centos
[21:30] <JEEB> ok
[21:30] <JEEB> that's AncientAncient
[21:30] <wizonesolutions> heh, yeah
[21:30] <wizonesolutions> I'm just glad they have something
[21:31] <wizonesolutions> and it produces good enough mp4 that the client is happy, so...
[21:31] <wizonesolutions> I think I'll just have the script look and see if ffmpeg complains about unrecognized option, trying -vpre first, and if so then I run the alternative
[21:31] <wizonesolutions> shouldn't be too intensive
[21:31] <wizonesolutions> do you happen to know the switch to force "yes"?
[21:31] <wizonesolutions> to prompts
[21:31] <wizonesolutions> is it just like -y?
[21:32] <JEEB> I think it's something like that, yeah
[21:33] <passstab> how about mjpeg, what is the best codec for that?
[21:34] <wizonesolutions> lol, so weird, -preset veryslow looks ungodly awful whereas the 0.7 -vpre veryslow looks great
[21:35] <wizonesolutions> converting from FLV to MP4
[21:35] <wizonesolutions> doesn't really matter, the server one is good which is what matters
[21:36] <JEEB> they really shouldn't be that different tho
[21:36] <JEEB> I have no idea what exactly the ffpreset file sets tho
[21:37] <JEEB> with -preset veryslow you get otherwise defaults (crf rate control, 23)
[21:37] <JEEB> and the libx264's internal veryslow preset
[21:37] <JEEB> I kind of have a feeling that you're setting rate control by yourself in one of the cases
[21:42] <wizonesolutions> JEEB: maybe so. I've just been googling...just need to understand the minimum to get this working. I'm using "ffmpeg -i <input file> -vpre veryslow <output file>.m4v" on the server.
[21:42] <wizonesolutions> locally, same except I used -preset veryslow, but the result was rather scary.
[21:42] <wizonesolutions> it was like the Halloween edition. Well, I think the audio was OK, but the video was like that one on YouTube where he transcodes it 1000 times.
[21:44] <JEEB> I think ffmpeg might select H.264 with "m4v" in the output file name by default, but generally it's good to manually set video and audio codecs you want
[21:46] <JEEB> ffmpeg -i input.derp -vcodec libx264 -vpre veryslow/-preset veryslow -<ratecontrol mode> <ratecontrol value> <if streaming, or sending through a limited pipe overall, possible extra vbv-related values> -acodec faac/aac -ab <audio bitrate> output.m4v
[21:47] <JEEB> this would be my minimum, more or less
[21:47] <wizonesolutions> JEEB: Oh yeah I was also using -vcodec libx264
[21:47] <wizonesolutions> will keep that in mind. is there any way to figure out what ffmpeg went with based on what I did give it?
[21:48] <JEEB> the terminal output
[21:48] <wizonesolutions> I don't want the server quality to change if I make the command more explicit
[21:48] <wizonesolutions> does it show me the exact options or do I have to interpret it?
[21:48] <JEEB> interpret
[21:48] <JEEB> and I have no idea if it would tell you what rate control mode etc. it set
[21:49] <JEEB> with a newer setup the default for libx264 is crf 23, an old ffmpeg would just set bitrate 200 or so
[21:49] <JEEB> crf = more or less quality based VBR
[21:50] <JEEB> (crf 23 has been the actual default of libx264 the library for quite a while now)
[21:50] <JEEB> also, if both used libx264 looking at the ffmpeg's terminal output
[21:51] <JEEB> you could do 'strings output.m4v |grep "x264"'
[21:51] <JEEB> should output the piece of text libx264 saves into the bitstream
[21:51] <JEEB> that contains the settings
[21:52] <JEEB> also, personally I would just build the same version of libx264/ffmpeg etc. on all machines
[21:52] <JEEB> simpler to handle
[21:54] <passstab> can anyone help me with the message "codec for stream 0 does not use global headers but container format requires global headers" ?
[21:55] <passstab> in ffserver
[21:55] <passstab> when i try to connect to it
[22:04] <wizonesolutions> JEEB: DreamHost is shared hosting :( however, the Medibuntu version of Ubuntu's ffmpeg can take the same command if I manually copy the "max" preset (which is similar to the server veryslow) to libx264-veryslow.ffpreset
[22:04] <wizonesolutions> so I'm going to run with that
[22:04] <wizonesolutions> and just tell the client to ignore the quality on my local. they're happy with the server one anyway, I just don't want to be accounting for my local's differences when writing the script
[22:04] <wizonesolutions> I don't care about the local output really
[22:04] <wizonesolutions> so, workaround in place :)
[22:04] <wizonesolutions> much appreciated on the replies...is there a karma system here, can I ++ you somehow? :)
[22:09] <JEEB> nope, not as far as I know :D
[22:51] <Digit> is there a way i could use ffmpeg on the command line to batch convert a load of .flv to whatever format is hiding in the flv container, without attempting any conversion of audio or video quality, frequency, rate, ratio etc.
[22:52] <Mavrik> um
[22:52] <Mavrik> you'd like to convert videos packed into .flv container to what exactly?
[22:53] <Digit> to.. whatever they already are.  i presume some will want to be avi, some mp4... and whatever else.  idk.  the whole flv thing kinda throws me.
[22:54] <Digit> if i could convert them all to .ogv without loss of quality, that'd be ideal. ^_^
[22:54] <Mavrik> um
[22:54] <Mavrik> you're mixing up containers and video/audio formats
[22:55] <Mavrik> your flvs want to be flvs as much as they "want" to be anything else
[22:56] <Mavrik> so what you really have to do is decide which container would you like your video/audio tracks in :)
[22:56] <Mavrik> and of course, make sure said container supports your video/audio formats
[22:56] <Digit> k.  :P   they're all already a format or codec or whatever under the flv bit though right?   sorry, i'm not perfect on the lingo.
[22:57] <Mavrik> Digit, yeah, basically "flv", "avi", "mp4", etc. is just a box where you put your audio and video in
[22:57] <Mavrik> it's called a "container"
[22:57] <Mavrik> in the container you have one or more streams, which can be audio, video, subtitles, data, etc
[22:58] <Digit> would .ogv fall into that "etc"?
[22:58] <Mavrik> yeah
[22:58] <Mavrik> in .ogv you usually find one video and one audio stream
[22:58] <Mavrik> video stream being encoded in Theora format
[22:58] <Digit> so i could have all my flv converted to ogv, without loss of quality?  :O
[22:58] <Mavrik> and audio in Ogg Vorbis
[22:59] <Mavrik> Digit, probably not, I think ogv can't take anything except theora video / vorbis audio :\
[22:59] <Mavrik> and flv can't take theora video / vorbis audio so you have a problem ;)
[23:00] <Mavrik> Digit, what's your actual goal btw? What are you trying to do?
[23:00] <Digit> oh well, i'll settle for avi or mp4 then i suppose.  it'd be better than flv which i discovered are useless for a lot of my friends.
[23:00] <Digit> i'm trying to get rid of the flv-ness.  ^_^   for many reasons.
[23:00] <Mavrik> ah
[23:00] <Mavrik> Digit, mp4 is something most devices/OSes support out of the box
[23:01] <Mavrik> you can remux .flvs that have H.264 video / AAC audio without loss of quality
[23:01] <Mavrik> ffmpeg -i <your flv>.flv -codec copy output.mp4
[23:01] <Digit> my imediate goal is trying to stop the philosophy discussion and get to the practical cl suggestions. ^_^  wanna find out the options which prevent any converting or meddling with the actual audio and video
[23:01] <Mavrik> that would be the "copy" codecs
[23:02] <Mavrik> those will just copy the underlying audio/video streams
[23:02] <Digit> sweet.  thnx.
[23:03] <Digit> would that work as a batch command just using *.flv and *.mp4... or would i need to wrap that in a clever script to achieve that?
[23:03] <Mavrik> you'll have to do a clever script
[23:15] <Digit> hrmm, does ffmpeg have a way to be smart about the output name, so it uses the input file's name preceding any file-type suffix... or do i have to be clever with my script for that too?
[23:47] <Sashmo> im curious, has anyone had good success streaming to wowza or flash media server?  I'm using DVB sources and I cant see to get a good enough stable output to both FMS and Wowza.  I've even resorted to building a separate segmenter to handle the HLS side of things.  That has smoothed what Im sending to wowza, but I still think its highly inefficient to use another segmenter when wowza works perfectly fine on the same machine, when sending analog from Flash med
[23:50] <Sashmo> Im willing to pay some guru to solve the problem for me&. Any takers?
[23:55] <beandog> For something as specific as that, I'd start hitting some forums
[23:55] <Sashmo> been there, done it
[23:55] <Sashmo> 6 months
[23:55] <Sashmo> something with my encoding techniques is causing it, im sure, but I cant seem to solve it
[00:00] --- Sat Jul 21 2012


More information about the Ffmpeg-devel-irc mailing list