[Ffmpeg-devel-irc] ffmpeg.log.20120314

[burek]

[00:51] <dj_who> hi what does it mean "start: 1.49479" in ffprobe output
[01:26] <danolj> http://pastebin.com/eWCR2ySE
[01:27] <danolj> I am trying to add a 'yadif' filter to an existing and working filter that overlays an image on a video
[01:27] <danolj> in doing so, ffmpeg appears to parse the yadiff parameters correctly,but then complains about "Not enough inputs specified for the "yadif" filter.
[01:27] <danolj> "
[01:27] <danolj> the pastebin link contains everything
[01:28] <cbsrobot> try to put [out] after yadif
[02:07] <danolj> okay
[02:08] <danolj> cbsrobot: thank you, seems to do the trick
[04:42] <jhford> i was wondering if enabling things like lame, vpx, etc, would replace built  in encoders/decoders or augment them?
[04:42] <lahwran> I have 25GB of wav files, containing sound effects. I'd like to compress them to something, perhaps flac. how would I use ffmpeg to do this without losing quality?
[04:42] <lahwran> ie, I want to use the same bitrate as the input file, but encode to flac
[04:42] <jhford> lahwran, any reason not to use the command line flac encoder?
[04:43] <lahwran> 'cause I'm somewhat familiar with ffmpeg but have never used that
[04:43] <lahwran> but sure, why not
[04:43] <sacarasc> If you convert to flac, you won't lose quality. But the bitrate will be lower, that is the compression side of it. :p
[04:43] <jhford> lahwran, the command line interface for flac is really simple!
[04:43] <lahwran> sacarasc: well, if I compress to flac, then uncompress, I want it to produce exactly the same audio data
[04:43] <jhford> that's what flac does
[04:43] <sacarasc> It will be exactly the same.
[04:43] <lahwran> sacarasc: which, if I have a lower sample quality, it might not
[04:44] <jhford> well, there is some information (i forget what) that flac doesn't store
[04:44] <sacarasc> Free LOSSLESS audio codec.
[04:44] <lahwran> wav is "lossless" too, but I can encode wav to wav and lose data
[04:44] <jhford> only if your encoder is busted or set to use different output options
[04:45] <jhford> really, if you want to encode wav to wav without loosing any data, i'll introduce you to the 'cp' program :P
[04:45] <lahwran> haha I'm just saying that you can encode to lossless formats whilst losing data
[04:45] <jhford> yes, if you do things to alter the stream
[04:46] <jhford> :)
[04:46] <jhford> my experience is that a wav encoded to flac then decoded back to wav has the same sha1 checksum
[04:46] <jhford> meaning that it is exceedingly unlikely to be changed
[04:47] <lahwran> alright cool
[05:17] <pasteeater> jhford: external encoders/decoders can "replace" some default encoders if you do not declare any options.
[05:17] <pasteeater> such as libx264 being used instead of mpeg4 by default for mp4 output
[05:18] <jhford> pasteeater,  in that case, would the output still be h.264?
[05:18] <pasteeater> but of course you can choose what encoder or decoder (i don't like the word "codec") you want
[05:18] <pasteeater> if you do not have libx264 enabled, then the output will not be H.264 because it is using the "mpeg4" encoder by default.
[05:18] <jhford> ohh, ok
[05:19] <jhford> so enabling encoders or decoders could change the default, but all will still be available?
[05:19] <pasteeater> yes. i see what you're asking now.
[05:19] <pasteeater> external junk won't exclude native junk
[05:20] <pasteeater> the choice will be yours.
[05:20] <jhford> aha!  excellent, thanks for the confirmation :)
[05:20] <pasteeater> we like to keep the natives happy.
[05:20] <jhford> hehe
[05:23] <jhford> it's too bad there isn't a --program-suffix option
[05:23] Action: jhford wants to rename his ffmpeg binary to allow for installing more than one ffmpeg into the same directory
[05:23] <jhford> err, same prefix
[08:59] <lahwran> there doesn't seem to be a #flac, so I'll ask here: why does the error "ERROR: input file ./Track 77.mp3 has an ID3v2 tag" occur when trying to use the flac command? googling it just showed that I need to remove the id3 tags, but why do I need to
[10:17] <luc4> Hi! Is anyone working on ARM?
[10:27] <zap0> my right arm gets quite a workout..
[11:14] <JacobS1> Hello everyone I need some help with udp streaming performance, I am streamin the desktop capture of a windows machine to another machine via udp in 100m lan , and I am having frame glitches in the output, I noticed that ffmpeg is consuming 50% CPU ( this is a Core 2 Duo E8400 3GHz )
[11:34] <JacobS1> this is my output: http://pastebin.com/BF8FdChr
[11:41] <JacobS2> this is my output: http://pastebin.com/BF8FdChr
[12:09] <kriegerod> JacobS2: what's the bitrate of udp stream?
[12:09] <kriegerod> and first of all why you use udp
[12:10] <JacobS2> I am using udp because it is a fast network and I thought it will give the best results
[12:10] <JacobS2> I am open to other ideas
[12:11] <JacobS2> kriegerod: I didnt define a fixed bitrate in the command
[12:12] <kriegerod> i ask about actual bitrate
[12:12] <kriegerod> how much UDP data is sent per second
[12:13] <kriegerod> if you can, try tcp instead of udp
[12:13] <JacobS2> its in the output here: http://pastebin.com/BF8FdChr
[12:19] <JacobS2> kriegerod: what is the command for tcp streaming ?
[12:20] <kriegerod> most probably exactly as your current command, with replacing udp:// to tcp://
[12:32] <f0rm4t> Anyone active?
[12:34] Last message repeated 5 time(s).
[12:35] <cbreak-work> spammer.
[12:35] <f0rm4t> Not really.
[12:36] <f0rm4t> atention getter
[12:37] <JacobS2> kriegerod: this is the command I used - -f dshow -i video="screen-capture-recorder":audio="Realtek HD Audio Input" -vcodec  libx264 -preset ultrafast -tune film -r 10 -async 1 -ab 16k -ar 11025 -threads 4 -f mpegts tcp://@:1234?listen
[12:37] <microchip_> s/getter/whore/
[12:37] <f0rm4t> got time for a quick qestion on x264 baseline profile?
[12:37] <JacobS2> and here is the output: http://pastebin.com/Tcxv8xbw
[12:38] <f0rm4t> Anyone know a way to verify a x264 mp4 is baseline profile or not?
[12:39] <kriegerod> f0rm4t: ffprobe
[12:39] <kriegerod> P.S. don't play with people attention next time. and dont ask to ask
[12:41] <f0rm4t> right... got it... no playing... no aksing to ask... Do I just ask... or is that not cricket either?
[12:41] <f0rm4t> ;)
[12:41] <microchip_> just ask
[12:42] <kriegerod> JacobS2: seems problem is network bandwidth between hosts.
[12:46] <shreya> Hello all,I need some help
[12:46] <shreya> I need to stitch together images to form a video
[12:46] <shreya> using libffmpeg
[12:46] <shreya> Please guide me regarding same
[12:46] <shreya> Thanks
[12:47] <f0rm4t> dont say you need help... just say. PMSL!!!
[12:50] <shreya> Please guide me on the same
[12:52] <shreya> Its a bit urgent
[12:52] <f0rm4t> http://www.catswhocode.com/blog/19-ffmpeg-commands-for-all-needs
[12:52] <cbreak-work> with mencoder you could just use mf://\*.png and so on
[12:53] <cbreak-work> no idea how you can read image sequences with ffmpeg
[12:53] <kriegerod> -f image2
[12:53] <f0rm4t> ffmpeg -f image2 -i image%d.jpg video.mpg
[12:53] <shreya> Can this be done without shell scripting?
[12:53] <kriegerod> yes
[12:53] <shreya> I need this feature in a Qt based application
[12:53] <shreya> kriegerod, how?
[12:54] <kriegerod> also i've made an app for certain cases of your task https://github.com/krieger-od/imgs2video
[12:54] <kriegerod> shreya: well, that's just invocation of naother executable
[12:54] <cbreak-work> Qt has QProcess for that
[12:54] <kriegerod> if not acceptable - then learn ffmpeg API...
[12:54] <shreya> kriegerod, Ok thanks Ill take a look
[13:01] <shreya> kriegerod, What are the cmdline files doing?
[13:01] <kriegerod> ?
[13:07] <f0rm4t> Anyone have a trusted url to a win32 binary of ffprobe?
[13:08] <kriegerod> http://ffmpeg.zeranoe.com/builds/
[13:11] <f0rm4t> :)
[13:36] <oz__> kriegerod: should I preffer tcp over udp in cases when I need both reliability and small delay ?
[13:36] <oz__> or, how bad is tcp when it comes to minimizing the delay
[13:39] <kriegerod> oz__: too generic question. hard to answer. and not an ontopic here.
[13:39] <kriegerod> i just know udp transfering of streams adds problems for receiving ffmpeg instance
[13:51] <oz__> thanks anyway. I guess It's a trial and error. problem is that even in a LAN there are cases where one will outperform the other.
[14:07] <AugustePop> how can i strip all metadata while transcoding with ffmpeg now? i tried -map_metadata -1 -map_metadata:s:v -1 -map_metadata:s:a -1, but nothing was discarded at all. every metadata is copied over.
[14:51] <luc4> Hi! Can I somehow change how large each AVPacket must be? I would like to decode in smaller chunks.
[15:07] <kriegerod> luc4 never heard of such. what's the container (file format) and codecs?
[15:08] <luc4> kriegerod: I'm trying to play a wma, but it seems ffplay cannot play it because too much time is spent in the audio callback.
[15:08] <luc4> kriegerod: I see in the SDL audio callback the call that decodes the AVPacket, but in the case of wma each packet is far larger than in the mp3 case. Do you know why?
[15:09] <kriegerod> don't know
[15:09] <kriegerod> is that super-old-weak pc?
[15:09] <luc4> kriegerod: do you know who decides how large a AVPacket should be?
[15:10] <luc4> kriegerod: this is an embedded platform.
[15:10] <luc4> kriegerod: I tried on two completely different hardware and the problem is the same.
[15:10] <kriegerod> demuxer. it is not configurable for packet size, as far as i have seen. not sure about wma
[15:11] <luc4> kriegerod: I understand... I'll try to solve some way. Thanks!
[16:46] <JacobS2> hello all, what does this message mean?: real-time buffer 155% full! frame dropped! , I am getting them alot
[16:47] <JacobS2> this is my command: -f dshow -i video="screen-capture-recorder":audio="Realtek HD Audio Input" -vcodec  libx264 -preset ultrafast -tune film -r 10 -async 1 -ab 16k -ar 11025 -f mpegts udp://172.21.120.185:48552
[16:48] <JacobS2> this is the full output: http://pastebin.com/YWtZSnWi
[16:56] <Mavrik> JacobS2: it seems dshow is getting input faster than you consume it
[16:56] <Mavrik> JacobS2: consider your input is 30 fps and you're encoding at 10 that seems logical
[16:58] <JacobS2> Mavrik: can this cause glitches in the recieving side ? because I am having problems with the viewer of this stream
[16:58] <Mavrik> hmmm, it could
[16:59] <Mavrik> JacobS2: also, if you're streaming h.264 over mpeg2-ts
[16:59] <Mavrik> add -bsf:v h264_mp4toannexb
[16:59] <Mavrik> which will fix some stream parameters
[17:02] <JacobS2> Mavrik: cool, thanks I will give it a go
[17:24] <JacobS2> Mavrik: unfortunately it didnt help, I am still getting many image errors.
[17:24] <JacobS2> The funny thing is that I am testing on 3 machines (streaming machines) and the one that is giving the most image errors is the strongest one - Core 2 Duo E8400 3GHz, the others are Core 2 Duo E6550 2.33 GHz and Pentium 4 2.8 GHz.
[17:25] <Mavrik> did you try streaming at original fps?
[17:28] <JacobS2> No, I will try
[17:33] <JacobS2> I removed the -r 10 and still the same problems, also I still get alot of real-time buffer 155% full! frame dropped!, I have noticed that the cpu is at a constant 47-52% why is ffmpeg not using more if it needs more ?
[17:37] <kriegerod> JacobS2: try saving to file. If saving to file, and playing file succeeds - network bandwidth is problem and you should reduce produced bitrate.
[17:41] <JacobS2> I will try that, how do I reduce bitrate ? what value do you recomend ? (100mb lan)
[17:43] <JacobS2> kriegerod: can "real-time buffer 155% full! frame dropped!" be related to network bandwidth, because from what I can tell this message is the main difference between the problamatic machine and the working one
[18:31] <JacobS2> kriegerod: I am trying to save to file to check if it is a network problem, what command do I need to create a file instead of udp stream to best immulate my original command ?
[18:48] <JacobS2> kriegerod: I saved to a file, and there were no glitches, you mentioned I need to set the bitrate. what whould be a reasonable value ? also the documentation mentions I need to set the bufsize, can you please give me a good example
[18:52] <Craig`> anyone active?
[19:20] <pasteeater> Craig`: i am now. ugh...damn that scotch.
[20:03] <dj_who> hi all, what does it mean "start:(some value)"  in ffprobe output?
[20:10] <Skaag> I have a problem with lipsync, here's a mediainfo dump for the input file: http://pastebin.com/NphTjFDd
[20:13] <Skaag> this is how I transcode it: ffmpeg -y -threads $THREADS -i "$file_source" -s "$dimensions" -aspect "$aspect_ratio" -vb "$video_bitrate"k -vcodec libx264 -vpre slow -vpre "$preset" -bufsize "$max_bitrate"k -maxrate "$max_bitrate"k -ab "$audio_bitrate"k "$file_target"
[20:16] <Skaag> And here's a mediainfo dump of the output: http://pastebin.com/5aaDQ2ZR
[20:17] <relaxed> Skaag: which version are you using?
[20:18] <Skaag> ffmpeg version N-32611-gd55b06b, Copyright (c) 2000-2011 the FFmpeg developers
[20:19] <Skaag> here's a full dump of -version: http://pastebin.com/5sdMh5PB
[20:19] <Skaag> lip sync issues start around 30 seconds into playback
[20:23] <relaxed> Updating to a recent version may fix it.
[20:24] <Skaag> interesting
[20:24] <Skaag> this is a version we built ourselves from source
[20:25] <Skaag> (6 month ago)
[21:21] <dericed> what types of audio codec are possible in mpeg-ts?
[21:22] <dericed> by possible, i mean supported by the container format
[21:52] <TACPILOT> found some documentation referring to an issue with how libavcodec handles h.264 interlaced content. I dont know how old the info is?
[21:52] <TACPILOT> http://pastebin.com/e6iTmFYY
[21:52] <TACPILOT> has this issue been resolved with current versions ??
[21:53] <TACPILOT> the (odd and even) jumping around is of concern to me
[22:07] <deni> can anyone estimate how long would it take to convert a 22Gb .dv file to an .avi file with ffmpeg?
[22:08] <deni> 2 cores.... 2Ghz proc
[22:08] <deni> i need a rough estimate....it doesn't need to be correct to the minute i just want to know are we talking days, hours or minutes?
[22:13] <balrog_phone> deni: Start the conversion, see how fast it's going, calculate
[22:15] <TACPILOT> time may differ greatly by the settings u choose .. doing as balrog_phone says will most likely be your only option
[22:15] <deni> balrog_phone: about 5 seconds after i asked the question the process finished
[22:15] <deni> :D
[22:25] <chao> Does anyone know if the common unit for bit rate in video streams is MB/s or MiB/s (base 10 or base 2)?
[23:06] <jblack> Hello! I have a script that splices a handful of .flvs into one .wav file.  It seems that padding is being introduced in the process.
[23:07] <jblack> It happens rarely, but when it does, it can be off by as much as 5 seconds on a minute and a half of audio
[23:09] <pasteeater> jblack: are you using ffmpeg?
[23:09] <jblack> I'm not quite sure how to take it past that point.
[23:09] <jblack> Yeah.         os.system("ffmpeg -i "+source+" -ar 44100 -ab 128k -ac 2 "+destination)
[23:09] <pasteeater> use a pasetebin service to show your script
[23:10] <jblack> pardon, that's the wrong wrong call.
[23:11] <jblack> I just found the actual function that's doing the actual work.
[23:11] <jblack> It's a mess.
[23:11] <jblack>  should be able to ask a better question in a little bit though
[23:25] <jblack> NEver mind. I dont' think this has anything to do with ffmpeg itself. sorry.
[23:37] <arbin> chao, in my experience it's always powers of 10
[23:37] <chao> Okay thank you
[00:00] --- Thu Mar 15 2012