[Ffmpeg-devel-irc] ffmpeg.log.20121117

[burek]

[02:09] <mishehu> greets folks.  don't know if anybody here is knowledgeable about this - but in a piece of software that I am looking at, the programmer is prepending a value of 0xB2 to every audio frame.  he says this is an flv value generated from the audio codec, bit rate, etc.  the codec is wideband speex mono.  might anybody know what the calculation is that may have gotten this value of 0xB2 ?
[02:13] <JEEB> for questions regarding the FLV container I recommend just grabbing the spec :)
[02:14] <JEEB> http://www.adobe.com/devnet/f4v.html
[02:14] <JEEB> the PDF linked here has both FLV and F4V specs
[02:14] <JEEB> the FLV part is rather short so you should be able to go through it
[02:14] <JEEB> that does sound kind of weird, but loldunno :P
[02:19] <mishehu> hopefully
[02:19] <mishehu> it's a friday and my head is already wanting to shut down for hte weekend :-)
[03:42] <retard> when specifying a start time and using -acodec copy and -vcodec copy the video gets aligned to the closest keyframe, which is fantastic, but the sound does not get aligned and uses the specified start time exactly
[03:42] <retard> is there an easy way around this?
[03:43] <retard> besides finding the exact right time
[03:55] <klaxa> not really... i think?
[03:57] <retard> is there some way of finding the precise frame number of the keyframe and specifying that as the starting point? :/
[03:58] <klaxa> well there should be... if it's matroska mkvinfo -s <file> will print out every frame and what type it is (you're looking for I-Frames)
[03:59] <klaxa> i don't know about other ways, but i'm positive there are other ways
[04:02] <retard> mmm, do you know how i can specify that frame as the start point for encoding? is there some kind of format that lets me give frame number to -ss ?
[04:03] <retard> "position may be either in seconds or in hh:mm:ss[.xxx] form."
[04:05] <klaxa> i'm afraid that ffmpeg doesn't support that, i hope someone can prove me wrong :X
[04:07] <retard> i guess i'll make do with finding as exact a time i can get and specify that
[04:07] <retard> thank you for your answers!
[04:16] <FelipeS> hey all, how in earth can I turns this .ts into an .mp4 without re encoding https://dl.dropbox.com/u/94267/fileSequence0.ts
[04:16] <FelipeS> ffmpeg complains with "pts (3000) < dts (6000) in stream 0 av_interleaved_write_frame(): Invalid argument"
[04:21] <klaxa> well... mplayer2 outputs a lot of warnings...
[04:23] <klaxa> it seems like there are timestamps missing or something
[04:24] <FelipeS> it's the first file in an m3u8 playlist
[04:25] <FelipeS> I don't see any warnings with vlc
[04:28] <klaxa> mplayer2, ffprobe and ffplay do though
[05:49] <elkng> can I make effect of lovering audio on some places in video ?
[07:20] <Devon> i get this error:  Can't write packet with unknown timestamp av_interleaved_write_frame(): Invalid argument
[07:21] <Devon> what does this mean
[16:53] <IsoLinCHiP> Hi, Im trying to use ffmpeg to stream to a FlashMedia Server hosted on ustream. Ive gotten it to work basically, however my video is choppy and I keep getting "rc buffer underflow" message (probably explains the choppy effect). Unfortunately I have no real idea from which angle to attack this problem and would be greatfull for any suggestions.
[16:56] <IsoLinCHiP> Im using Windows btw. My source video is a screencapture filtered through avisynth, which crops and scales the video down. When I preview the device via gstreamer and vlc the video is smoothe as silk
[16:57] <IsoLinCHiP> just previewed it with ffplay aswell and the result is as expected too
[16:59] <IsoLinCHiP> allthough now sometimes ffplay displays real-time buffer full framed dropped sometimes
[17:00] <IsoLinCHiP> (never mind, unrelated problem, seems to only happen when I move the ffplay window, probably not multithreaded or so)
[17:20] <paule32> hello
[17:20] <paule32> having proiblem with running ffmpeg
[17:20] <paule32> http://codepad.org/3HpNi7B7
[17:29] <saste> paule32, "SVN-r0.5.9-4:0.5.9-1, Copyright (c) 2000-2009" -> ooold and deprecated, please update
[17:29] <paule32> i have download the new git
[17:29] <saste> good
[17:29] <paule32> but don't know how to translate the command to it
[17:30] <paule32> i wanna record desktop with audio/(mic)
[17:31] <saste> should work as is
[17:33] <paule32> then i get this:   http://codepad.org/NKk2b5o3
[17:43] <IsoLinCHiP> Can anyone at least tell me what the rc buffer is?
[17:45] <saste> paule32, --enable-x11grab --enable-free
[17:46] <saste> or something like that
[19:03] <jure> how do I use avidemux with a custom version of ffmpeg?
[19:08] <Heyyy> could anybody help me with command to make a video file from an image and an mp3?
[19:11] <JEEB> ffmpeg -i input.picture -i input.mp3 -c:v libx264 -crf 23 -c:a copy out.mkv
[19:12] <JEEB> not sure if you need a flag to repeat the video side of things
[19:15] <Heyyy> JEEB: I seem to remember -sameq being used in place of some of those parameters, but not sure where
[19:15] <JEEB> sameq has been removed because it hadn't done what it was supposed to do for ages
[19:15] <JEEB> and it isn't "same quality"
[19:16] <JEEB> it is "copy quants from input stream", and it should only work when your input and output are of the same format
[19:16] <JEEB> in 99% of all cases this isn't true and in 99.9% of all cases of recommendation of -sameq on the internet it has been a false result
[19:16] <Heyyy> good to know! maybe that explains a thing
[19:16] <JEEB> I mean, sameq had been broken for a couple of years for most formats
[19:16] <JEEB> before removed
[19:17] <JEEB> and that incorrect way it was working might have actually given people good quality by chance
[19:17] <JEEB> and before that it IIRC actually had relatively bad documentation
[19:17] <Heyyy> well it wasn't *good* just that it was easy for me
[19:17] <JEEB> before it was removed the documentation was added to note that THIS IS NOT SAME QUALITY, THAT IS IMPOSSIBLE
[19:18] <JEEB> anyways, -crf in case of the libx264 video codec is closest to "constant quality"
[19:18] <JEEB> lower is "lower quants, more bits, better quality" and higher is "higher quants, less bits, possibly worse quality"
[19:18] <JEEB> 23 is the default for the mode, and a good starting point
[19:18] <JEEB> because it compresses nicely
[19:20] <Heyyy> JEEB: tried, "audio.mp3: error could not find codec parameters"
[19:20] <JEEB> pastebin whole command and the terminal output onto a pastebin site of your preference
[19:21] <JEEB> and then link your pastebin here
[19:21] <Heyyy> JEEB: k
[19:22] <Heyyy> JEEB: http://pastebin.com/cwrCnaXc
[19:23] <JEEB> ok, you are using the ubuntu packaged ffmpeg binary that comes from the libav package. It is old and wasn't updated when libav rewrote parts of the ffmpeg command line application. If you want to use it, switch to the 'avconv' command instead
[19:23] <IsoLinCHiP> If I specify a bitrate with -b:v does that account for the fps? or do I need to multiply that with the -r parameter to calculate the bandwidth needed to live-stream ?
[19:24] <JEEB> IsoLinCHiP, most sane encoders set their rate control with accounting the frame rate. That said, you also /need/ VBV as well
[19:24] <JEEB> -maxrate and -bufsize
[19:24] <JEEB> with streaming
[19:24] <JEEB> (or well, any use case where you have limited transfer speed capabilities)
[19:25] <JEEB> maxrate being the maximum average rate over the bufsize, and bufsize is how much the player side will be buffering at first
[19:26] <IsoLinCHiP> Im jugling around with -rtbufsize -r and -maxrate and -bufsize trying to get rid of the choppyness in my stream without a real feel for what affects the real-time buffer errors and which the rc buffer underflow :-/
[19:26] <Heyyy> JEEB: http://pastebin.com/6mEe3rRG looks same as before
[19:26] <JEEB> in other words, as long as the client has a speed of at least maxrate, and it buffers a bufsize amount of bits, the client will be able to watch the stream without other cases of buffering
[19:27] <JEEB> Heyyy, join #libav and ask there -- since they upkeep the debian/ubuntu packages
[19:27] <IsoLinCHiP> JEEB:  Im actually trying to stream to ustream so client and server might have switched positions in this case ;)
[19:28] <Heyyy> JEEB: wait, I think I missed something simple. audiofile itself troubled, not sure how
[19:28] <JEEB> IsoLinCHiP, then maxrate and bufsize variables should be set to whatever ustream wants :)
[19:28] <JEEB> they should have the information somewhere
[19:28] <JEEB> those settings are needed for any kind of working streaming over limited bandwidth
[19:29] <JEEB> because -b:v just sets the average bit rate over the whole video
[19:30] <paule32> how can i installl Xext package on debian squeeze?   (ERROR: Xext not found - by configure)
[19:30] <paule32> ?
[19:31] <IsoLinCHiP> *sigh* the errors just seem so random
[19:31] <JEEB> you can't get more useful help if you don't post your command as well as the thing in pastebin or something :P
[19:32] <paule32> me?
[19:32] <JEEB> I guess it goes for both of you :P
[19:32] <IsoLinCHiP> Sure, if that helps happy to do so :)
[19:32] <JEEB> although I have more of an idea about IsoLinCHiP's problem, possibly
[19:35] <JEEB> Heyyy, yes it was possible that the file was borked as well but I was kind of expecting that you'd know that the file was OK :D
[19:36] <Heyyy> JEEB: lol basically, still playing now
[19:39] <IsoLinCHiP> JEEB:  Maybe this makes it clearer? http://pastebin.com/W1wbwn4U
[19:41] <JEEB> IsoLinCHiP, I will guess you might actually not be going fast enough
[19:41] <JEEB> you don't have a -preset set so it defaults to 'medium'
[19:41] <JEEB> http://mewiki.project357.com/wiki/X264_Settings#preset
[19:41] <JEEB> list of presets
[19:41] <IsoLinCHiP> I couldnt find any documentation about the presets
[19:41] <IsoLinCHiP> Oh, ok, your one step ahead of me :)
[19:41] <JEEB> well, it basically just maps to libx264's preset system
[19:42] <JEEB> so that link has a listing of them
[19:42] <JEEB> I see the frame rate of the encoding going under 15 too, so it's most probably just the encoding not going fast enough
[19:42] <IsoLinCHiP> ahhhh, ok, I was searching in the ffmpeg docs
[19:43] <JEEB> medium is default so you might want to poke the ones faster than it, everything but ultrafast should be "Generally usable on one level of usable"
[19:43] <JEEB> also I have no idea if your buffer size on the DShow input side is OK :D
[19:43] <JEEB> but that seemed like the obvious thing because the fps was going down on the output side
[19:45] <IsoLinCHiP> well, that makes two of us, I have no clue what the buffer size needs to be either
[19:46] <IsoLinCHiP> Ive added -preset superfast now and it doesnt change the fact
[19:46] <IsoLinCHiP> also my none of my cores is actually at 100% so I see no reason it shouldnt be encoding "fast enough"
[19:47] <JEEB> and libx264 sure uses threads too, so that's fine as well
[19:47] <JEEB> dunno then
[19:48] <JEEB> check the input buffer size I guess?
[19:48] <JEEB> or check if the avs script you seemingly are using is quick enough?
[19:48] <IsoLinCHiP> Ive tried increasing it, but it only seemed to delay the problem, not solve it
[19:49] <IsoLinCHiP> As far as I can tell in gstreamer and vlc the avs is not dropping frames
[19:49] <JEEB> anyways, away for now
[19:49] <IsoLinCHiP> ffplay does drop frames though
[19:53] <IsoLinCHiP> duh, and a minute later or so it isnt again, this is really frustrating
[20:03] <paule32> http://codepad.org/Dy1YRfKP
[20:03] <paule32> don't have changing include files
[20:24] <Francexi> Hi there
[20:25] <Francexi> i've a problem during the compile of ffmpeg, anyone can help me?
[20:28] <Francexi> is someone alive?
[20:49] <burek> Francexi not until you provide some more info
[20:50] <Francexi> i'm running under slacware
[20:51] <Francexi> the error is libavcodec/libavcodec.so: undefined reference to `x264_encoder_open_128'
[20:51] <Francexi> libx264 is correctly installed
[20:51] <Francexi> other?
[20:51] <burek> paule32, apt-cache search ext
[20:51] <Francexi> burek slackware*
[20:51] <burek> or xext or something
[20:51] <burek> and when you find what you need, apt-get install ...
[20:52] <burek> if it was correctly installed it wouldn't throw that error
[20:52] <Francexi> the library exists
[20:52] <burek> what version?
[20:53] <Francexi> hold on, i see
[20:53] <paule32> burek: this dependces can't be solved:   http://codepad.org/EwSlfdoG
[20:53] <Francexi> 0
[20:53] <Francexi> 0.128.x
[20:53] <burek> ?
[20:53] <burek> where did you see that
[20:54] <Francexi> x264 --version
[20:54] <burek> paule32, i dont speak german unfortunatelly
[20:54] <Francexi> mmm, maybe is better trying to use pkg-config to get the version
[20:54] <burek> what exactly are you trying to do
[20:54] <burek> Francexi, for god's sake..
[20:54] <burek> you don't need binary
[20:54] <burek> you need a lib
[20:55] <Francexi> the with lib you mean
[20:55] <paule32> burek: i try to compile ffmpeg under debian 6 squeeze
[20:55] <Francexi> libx264?
[20:55] <burek> Francexi, how did you install your libx264?
[20:55] <Francexi> hold on (again)
[20:55] <paule32> wir x11 support for screen casts
[20:55] <burek> paule32, did you check our Wiki
[20:56] <Francexi> libx264 installed use a SlackBuild script
[20:57] <Francexi> ./configure \   --prefix=/usr \   --libdir=/usr/lib${LIBDIRSUFFIX} \   --enable-visualize \   --enable-shared \   --enable-pic \   --enable-strip
[20:57] <Francexi> this configure is used on x264 program (i think that --enabled-shared install the library)
[20:58] <burek> where exactly is your libx264 installed
[20:59] <Francexi> according to find /usr | grep libx264
[20:59] <Francexi> in /usr/lib
[20:59] <burek> btw, why did you use --libdir
[20:59] <burek> is there any specific reason for that
[20:59] <Francexi> no
[20:59] <Francexi> the slackbuild use it
[20:59] <burek> then why?
[21:00] <Francexi> i copy/paste it from it
[21:00] <burek> could you just type ./configure
[21:00] <Francexi> http://slackbuilds.org/slackbuilds/14.0/multimedia/x264/x264.SlackBuild
[21:00] <Francexi> used it
[21:00] <burek> try to build it normally
[21:00] <burek> undo what you did so far
[21:00] <burek> and just go into the git source dir for x264
[21:01] <burek> and type ./configure --enable-shared --enable-static
[21:01] <burek> and that'll do
[21:01] <Francexi> ok, i'll give a try
[21:06] <paule32> burek: i get this errors:   http://codepad.org/0Z8wqmJ4
[21:51] <burek> paule32, probably you mixed several times compiling results
[21:51] <burek> try to do a clean: make distclean
[21:51] <burek> and do ./configure again
[21:52] <paule32> ok
[00:00] --- Sun Nov 18 2012