[Ffmpeg-devel-irc] ffmpeg.log.20121003

burek burek021 at gmail.com
Thu Oct 4 02:05:01 CEST 2012


[00:10] <manizzle> hey guys, i am trying to convert a .wmv to avi
[00:10] <jotham> lol
[00:11] <manizzle> is that voodoo
[00:11] <klaxa> if you elaborate the specifics a bit mo-- wait are you trolling?
[00:11] <jotham> manizzle you want to convert the streams inside the containers (they are both containers)
[00:12] <jotham> but afaik an avi takes the same streams wmv containers do
[00:12] <klaxa> my guess would be he wants XviD/mp3 in the end
[00:13] <manizzle> someone said the mss2 can be converted, but the wmavoice needs to be run in windows
[00:13] <manizzle> stripped and rejoined to the mss2->xvid
[00:13] <manizzle> is that the only solution?
[00:13] <manizzle> WMAPro-in-WMAVoice is fucked up yeah?
[00:15] <klaxa> can ffmpeg decode that?
[00:15] <klaxa> wait what
[00:16] <klaxa> how did you get a codec into a codec
[00:18] <klaxa> what did you try so far manizzle?
[00:18] <klaxa> because from the command line help it looks like ffmpeg should be able to decode all the streams you named
[00:19] <manizzle>  http://sprunge.us/ceAU
[00:20] <klaxa> hmm... try the git version, i doubt it will work though, otherwise try contacting the devs, either here or mailing list or something
[00:22] <manizzle> any devs here?
[00:22] <iive> he have 1.0 that is quite recent.
[00:23] <klaxa> yeah therefore the git version might not be a lot different
[00:23] <manizzle> so which ffmpeg- mailing list should i email?
[00:23] <klaxa> manizzle: does mplayer play it?
[00:23] <manizzle> http://ffmpeg.org/contact.html
[00:23] <manizzle> lemme try
[00:23] <klaxa> you could try to dump raw pcm and use that instead
[00:24] <iive> manizzle: there may be another solution, while you are waiting for this to be implemented. and if it requires RE, it may take a while.
[00:24] <manizzle> k
[00:25] <iive> 32 bit mplayer and mencoder can load win32 binaries. It may be a lot trickier to build and install needed files.
[00:25] <manizzle> um, so i play with mplayer i can hear audio
[00:25] <manizzle> but i cant see the video
[00:25] <iive> xine should also have win32 loader, not sure about vlc (and can't help with them).
[00:26] <manizzle> mplayer says no video
[00:26] <manizzle> is it maybe the video stream thats fucked up then?
[00:26] <iive> hum.. let me check.
[00:27] <manizzle> http://fitzcarraldoblog.wordpress.com/2011/07/13/how-to-play-mss2-codec-windows-media-video-9-screen-wmv-files-in-64-bit-linux/
[00:27] <manizzle> maybe the mss2 shit is the fucked up crap
[00:27] <iive> mplayer uses ffmpeg mss2 decoder
[00:27] <iive> you just need a recent version (1.1 ?)
[00:28] <manizzle> so the git version would help?
[00:28] <manizzle> i have 1.0
[00:28] <manizzle> ffmpeg version 1.0 Copyright (c) 2000-2012 the FFmpeg developers
[00:28] <manizzle>   built on Sep 29 2012 11:22:50 with gcc 4.7.1 (GCC) 20120721 (prerelease)
[00:28] <iive> i mean, mplayer 1.1
[00:29] <manizzle> MPlayer SVN-r35014-4.7.1 (C) 2000-2012 MPlayer Team
[00:29] <klaxa> manizzle you *could* try this: https://gist.github.com/ee9bdf4d7fcffa6dec29
[00:29] <klaxa> if it's not the mss2 thing fucking up
[00:31] <manizzle> k mplayer2 works
[00:32] <iive> 35014 is from jully 6
[00:33] <manizzle> MPlayer2 UNKNOWN (C) 2000-2012 MPlayer Team
[00:33] <manizzle> yeee
[00:33] <manizzle> it works
[00:33] <iive> the mss2 support have been added few days ago.
[00:33] <manizzle> noice
[00:33] <manizzle> yeah vlc doesnt have mss2 support yet
[00:33] <manizzle> oh well
[00:34] <manizzle> mplayer for now
[00:35] <manizzle> http://git.mplayer2.org/mplayer2/log/?qt=grep&q=mss2
[00:35] <manizzle> >
[00:35] <manizzle> ?
[00:37] <iive> mplayer2 is fork of mplayer
[00:40] <iive> hum, you are right, mplayer should also support it with the win32 loader. can you try mplayer and put the text output in a pastebin site and give the url to me?
[00:42] <manizzle> ill install mplayer svn
[00:55] <manizzle> yeah mplayer-svn works too
[00:55] <manizzle> MPlayer SVN-r35226-4.7.1 (C) 2000-2012 MPlayer Team
[00:55] <manizzle> 198 audio & 412 video codecs
[00:55] <manizzle> cool thanks
[00:56] <manizzle> but i think i like mplayer2 better, so switching back to mplayer2-git
[05:22] <Philos> Does anyone know where to get mingw-w32api?
[05:22] <Philos> I need it for a cross-compile but it's not listed on the repository.
[07:13] <Tesseract433> Hi
[07:17] <Tesseract433> I'm having some trouble compiling ffmpeg on a x86_64 Fedora 11 machine
[07:19] <Tesseract433> It fails when attempting to link ffmpeg_g with repeated undefined reference to `av_lfg_init'
[07:20] <Tesseract433> I'm guessing I'm doing something stupid, but I can't figure out what
[07:24] <burek> Tesseract433, did you checkout from git recently?
[07:25] <Tesseract433> yeah
[07:25] <burek> there were some massive compile errors on fate, and I guess they are fixed now
[07:25] <burek> so try to git pull/checkout again
[07:26] <Tesseract433> I'm on the latest commit
[07:30] <burek> still having same issues?
[07:31] <burek> can you pastebin the config.log
[07:32] <Tesseract433> Yeah, still have the same issue
[07:32] <Tesseract433> I just tried the 1.0 release and it linked fine
[07:44] <rugrug234546> hello, can someone help me out with this ffmpeg error (http://pastebin.com/XYXj6WYq). The error is "Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument". I have just pulled the trunk version of ffmpeg from git and compiled it. I need to use the m3u8 live segmenting that was added a few weeks ago. I'm not an ffmpeg expert, so not sure where to go from here. Thanks.
[07:51] <rugrug234546> sorry i got disconnected. I'm goign to ask my question again. \
[07:52] <rugrug234546> i'm getting the error "Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument". You can see the entire output here: http://pastebin.com/XYXj6WYq - I'm cloned from git the latest dev version, because I need to use the m3u8 live playlist generation. But I don't know what the error i'm getting is and how to go about fixing that issue.
[07:52] <rugrug234546> any help is appreciated
[08:06] <rugrug234546> i think i've forgotten to include codecs... maybe...
[08:06] <rugrug234546> that's when i ./configure'd the src
[08:11] <symtab> hi
[08:11] <symtab> how can i watermark a video with the movie filter and keep the rest exactly the same (same bitrate...)?
[08:12] <symtab> /usr/local/bin/ffmpeg -i file.mp4 -sameq -vf "movie=watermark2.png [watermark]; [in][watermark] overlay=main_w-overlay_w-10:main_h-overlay_h-10 [out]" file_out.mp4 changes the bitrate from 900something to 3600something
[08:12] <klaxa> you can't without re-encoding unless you make it as subtitles which are removable though
[08:13] <klaxa> because of the -sameq flag
[08:13] <klaxa> it keeps the same quality, therefore uses high bitrates
[08:15] <symtab> so i have to re-encode with the exactly the same birate and add the watermarking?
[08:15] <klaxa> i think that would reduce quality maybe
[08:16] <klaxa> but try it and see if the result is still of reasonably good quality
[08:17] <symtab> just removing -sameq works
[08:17] <symtab> just tested
[08:20] <symtab> thanks for helping
[08:22] <rugrug234546> ok i ./configured and compiled with encoders and i used them in my ffmpeg command. Now i'm getting Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument
[08:24] <klaxa> complete log?
[08:24] <klaxa> i think there was this...
[08:30] <klaxa> oh sorry you already posted it... you're using a lot of functions i'm not familiar with... um... my guess would be (and this is probably wrong) to move -map 0 before the -vcodec copy -acodec copy
[08:30] <klaxa> actually that makes no sense
[08:30] <klaxa> disregard that
[12:16] <King_Rat> hey all
[12:16] <King_Rat> i'm trying to do a screen capture with x11grab and alsa
[12:18] <King_Rat> but I can't get alsa input working
[12:19] <King_Rat> it says the device or resource is busy, and no idea where to
[12:19] <King_Rat> start (at fixing it)
[12:20] <King_Rat> http://pastebin.com/MK1x5ZQV - here's the command and its output
[12:23] <King_Rat> http://pastebin.com/v8ihNW2U - here's my .asoundrc, if that's of any use
[13:18] <King_Rat> well i managed to get it to start by killing everything sound related
[13:19] <King_Rat> then the video just contained random background noise, and when i tried playing something, it went silent
[14:59] <ibsk8in31> Hi all:  I'm converting an .flv recorded from FMS 4.0 w/Nellymoser to wav in order to add audio back to AviSynth and attach to video.  When I convert the duration changes from 1:28 to 0:23.  Any way I can grab the whole thing?  Thanks
[15:05] <tuxhat> hello may i ask why when i do a screencast with video and audio i get alsa.buffer.xrun ?
[15:06] <dericed> I have 23.98 fps progressive material that I need to convert to 29.97 for broadcast. Does ffmpeg have a filter or process for converting through pulldown?
[15:38] <ibsk8in31> I did some research and it looks like it can't handle the silence.  The conversion process only reads the audio when the user's microphone was active.
[15:38] <ibsk8in31> Does anyone know have to handle silence in Nellymoser audio conversion?  It's changing the duration of my audio.
[15:40] <burek> King_Rat http://ffmpeg.org/trac/ffmpeg/wiki/Capturing%20audio%20with%20FFmpeg%20and%20ALSA
[15:40] <burek> also http://ffmpeg.org/trac/ffmpeg/wiki/How%20to%20grab%20the%20desktop%20(screen)%20with%20FFmpeg
[15:41] <burek> ibsk8in31 can you please use a pastebin site (like www.pastebin.com) to show your ffmpeg command and the complete console output?
[15:41] <burek> tuxhat, sync issues probably
[15:42] <burek> dericed, what is a "pulldown" ?
[15:43] <dericed> http://en.wikipedia.org/wiki/Three-two_pull_down
[15:43] <JEEB> it's basically the act of making extra fields to make content be 30000/1001 instead of 24000/1001fps
[15:43] <dericed> it adds frames by repeating fields rather than frames.
[15:44] <JEEB> yup
[15:44] <dericed> burek: When i use 'ffmpeg input -r ntsc output' to convert from 23.98 to 29.97 the result looks like it stutters
[15:44] <JEEB> it's because it's doing rate conversion and not "proper" pulldown
[15:44] <JEEB> also yes, it should look like stuttering to be honest
[15:45] <JEEB> because you get extra pictures that shouldn't be there
[15:45] <dericed> I was trying to mimic pulldown by using yadif to convert frames to fields, then the fps filter (to dup or drop to convert the rate), then convert fields back to frames. is this sane?
[15:45] <burek> dericed, maybe asking in #ffmpeg-devel would help?
[15:45] <JEEB> dericed, sounds like an overcomplicated way of doing it :s and possibly incorrect
[15:46] <dericed> JEEB: this was occurring to me
[15:46] <dericed> JEEB: maybe I need to pipe to mplayer and back
[15:46] <JEEB> I know avisynth has the capabilities to do both pulldown and pullup, but that you'd have to learn, too :s
[15:47] <burek> I would rather decompress everything to deinterlaced raw video and encoded it back to normal video I need
[15:47] <dericed> burek: ?
[15:47] <JEEB> burek, it's the opposite
[15:47] <JEEB> he has progressive
[15:47] <JEEB> 24000/1001 content
[15:47] <burek> oh i see
[15:47] <JEEB> he needs to make it 30000/1001 interlaced
[15:47] <burek> well, anyway you have some information missing
[15:48] <burek> so artifacts are expected
[15:48] <burek> like scaling a small image to a bigger one
[15:48] <ibsk8in31> http://pastebin.com/MUwcX7DG
[15:49] <JEEB> burek, you don't really lose information with the thing by itself as you're just separating to fields and adding extra fields to make it 30000/1001
[15:49] <burek> I never said you loose anything
[15:49] <JEEB> > you have some information missing, so artifacts are expected
[15:50] <burek> yes, you have missing frames
[15:50] <JEEB> what?
[15:50] <JEEB> I'm pretty sure that when you put those fields back together when doing inverse telecine, you will get the original chroma data unless you do lossy compression in the middle, but of course we were just talking about the processes of pullup and pulldown
[15:50] <burek> <dericed> burek: When i use 'ffmpeg input -r ntsc output' to convert from 23.98 to 29.97 the result looks like it stutters
[15:50] <JEEB> and you don't lose any frames
[15:50] <burek> 24 <> 30
[15:51] <JEEB> yes, you get extra frames, and that's why it stutters because fluid motion isn't fluid any more
[15:51] <burek> exactly what I said
[15:51] <JEEB> it's not "information missing"
[15:51] <burek> :)
[15:51] <burek> and why "extra frames" if nothing is missing
[15:51] <JEEB> because you're converting from a lower frame rate to a higher?
[15:52] <JEEB> instead of adjusting the video speed
[15:52] <burek> exactly my point, again
[15:52] <JEEB> what?
[15:52] <dericed> burek: because I need to achieve 29.97 fps for compatability with a broadcast system
[15:52] <burek> missing frames
[15:52] <JEEB> what?
[15:52] <JEEB> I guess you are very badly saying that you're not having 24->48
[15:52] <JEEB> and thus you lack the "everything's double" to get fluid motion
[15:52] <burek> it's pretty obvious that 30-24 = 6 frames missing each second
[15:53] <burek> I don't see the point in this arguing at all
[15:53] <ibsk8in31> Nellymoser losing duration of audio issue:  Do I use -async to force packets even when there is silence?  http://pastebin.com/MUwcX7DG
[15:53] <JEEB> burek, it's 24->30
[15:53] <JEEB> not 30->24
[15:53] <JEEB> thus you get 6 /extra/ frames
[15:54] <burek> again.. "<dericed> burek: When i use 'ffmpeg input -r ntsc output' to convert from 23.98 to 29.97 the result looks like it stutters"
[15:54] <JEEB> yes
[15:54] <JEEB> -r ntsc is 30000/1001
[15:54] <JEEB> not 24000/1001
[15:54] <burek> "from 23.98 to 29.97"
[15:54] <JEEB> from 24000/1001 to 30000/1001
[15:54] <JEEB> yes
[15:54] <burek> do you know how to read at all?
[15:54] <JEEB> yes
[15:54] <JEEB> and it looks like 30-24 not 24-30 extra frames
[15:54] <JEEB> see the from and to
[15:55] <JEEB> you are converting to something bigger than the input
[15:55] <dericed> dericed: now that everything is clarified, any suggestion to improve on -r. My next guess is to pipe to mencoder and back.
[15:55] <burek> ibsk8in31, did you try to convert the input, keeping the video stream too, using wav for audio stream
[15:55] <burek> just to see if it will produce the correct timings
[15:55] <burek> and then just get rid of the video
[15:55] <JEEB> dericed, mplayer /could/ do what you want but I have absolutely no experience there. I don't think ffmpeg can do pulldown (it can do ivtc nowadays tho, IIRC) :/
[15:56] <burek> geez JEEB, you didn't even read me
[15:56] <burek> I said exactly the same thing
[15:56] <JEEB> no, I did read you
[15:56] <burek> if he has got 24 frames
[15:56] <JEEB> <burek> it's pretty obvious that 30-24 = 6 frames missing each second
[15:56] <burek> and he wants 30 frames
[15:56] <ibsk8in31> I have tried that method with DirectShowSource() in Avisynth.  how can I keep the video stream in FFMPEG too?
[15:56] <burek> he's got 6 frames missing
[15:56] <burek> what's not clear in that statement
[15:56] <dericed> JEEB: no ffmpeg doesn't do ivtc. See the ticket here: https://ffmpeg.org/trac/ffmpeg/ticket/681
[15:56] <JEEB> oh you meant it that way
[15:56] <JEEB> that was sure a backwards way of saying it
[15:57] <burek> well, he surely doesn't have 6 MORE frames, does he? :)
[15:57] <JEEB> you should've just said that the added six frames cause the stuttering
[15:57] <ibsk8in31> End goal is to stitch two .flv's recorded by Flash Media Server together side-by-side w/audio.  However, I lose audio packets ;(
[15:57] <JEEB> because when you talk about missing frames
[15:57] <JEEB> it usually means they are getting lost
[15:57] <burek> his input has less information than it's needed to create desired output, that's the correct statement?
[15:57] <JEEB> yes, but it isn't being lost
[15:57] <JEEB> it just never was there
[15:58] <JEEB> the idea of losing information usually means that you had it to begin with
[15:58] <burek> [15:50:23] <burek> I never said you loose anything
[15:58] <burek> you misread it somehow
[15:58] <JEEB> <burek> he's got 6 frames missing
[15:58] <JEEB> I guess I misread this as "lost"
[15:58] <burek> ok :)
[15:59] <JEEB> but let's just say that you didn't put it in the usual way this stuff is conversed about :D
[15:59] <JEEB> if something's missing, you've usually lost it
[15:59] <JEEB> etc. etc.
[15:59] <burek> ibsk8in31, try something like
[15:59] <JEEB> dericed, ok -- too bad
[15:59] <dericed> I'll try to be more clear. I have 24000/1001 but need to convert to 30000/1001. Thus I need 6 frames ADDED. Right now ffmpeg makes new frames by duplication which causes a stuttery look. In telecine work, there is a technique called 3:2 pulldown to duplicate fields rather than frames for a less stuttery look. Can ffmpeg do something like 3:2 pulldown.
[16:00] <JEEB> dericed, yes I know that
[16:00] <JEEB> I'm just telling you that ffmpeg itself probably doesn't have pulldown
[16:00] <JEEB> so you'd have to use that mplayer you mention, if that works, or avisynth
[16:00] <JEEB> for that
[16:00] <JEEB> I only have avisynth experience mysefl
[16:00] <burek> ffmpeg -i input.flv -acodec pcm_s16le -ar 8K -ac 1 output.flv
[16:00] <JEEB> *myself
[16:00] <burek> or avi
[16:00] <ibsk8in31> Ok thank you.  Will try now
[16:00] <burek> output.avi
[16:01] <Stefff> I have a self captured mpg File from Composite In and Audio In. Audio and Video have a little delay which I'd like to correct. I tried it with following cli (+ Log) http://pastebin.com/0mXKbwLJ  But when I play this file with vlc e.g. I get no timecode at all it stays at 00:00. And ffmpeg -i shows me "0 channels" at the Audio Stream. What I'm doing wrong?  Or could I even get the capture at the capturing process sync?
[16:01] <dericed> burek: https://ffmpeg.org/trac/ffmpeg/ticket/1782 :)
[16:02] <burek> +1 :)
[16:02] <dericed> and it's cousin request: https://ffmpeg.org/trac/ffmpeg/ticket/681
[16:02] <burek> however, I'm not sure how nowadays video editors handle frame duplication process.. can it be specified which algorithm to use or not..
[16:03] <burek> so it might be a good idea to use a video editor for such things, if it supports such specification
[16:04] <burek> Stefff, with -itsoffset it does matter what -i comes first
[16:04] <burek> it defines what will get "moved" in time
[16:04] <burek> which*
[16:05] <burek> i.e. avoid using negative values for -itsoffset
[16:05] <burek> as it can have unpredictable results
[16:06] <Stefff> ok, I try the other way around
[16:06] <burek> btw, what are these "buffer underflow" errors
[16:06] <Stefff> I dont know.
[16:06] <burek> try without -vcodec copy -acodec copy, try re-encoding, to see if it works
[16:08] <burek> Stefff, one more thing, if you plan to play around with recorded material, try to capture it to something lossless
[16:08] <burek> for the future usage, I mean
[16:10] <Stefff> yeah, maybe we talked about it a while ago. I tried a little with lossless, but it would need to much space to archive it and the reencode later some other outputs from that.
[16:11] <burek> well it all depends what is your goal
[16:11] <burek> if the goal is the best image quality, the fastest encoding speed and the best compression ration, then you are in trouble :)
[16:11] <burek> raio*
[16:11] <burek> ...
[16:11] <burek> ratio :)
[16:11] <JEEB> dericed, an example of what you want to do is available on http://avisynth.org/mediawiki/Select#SelectEvery
[16:11] <JEEB> but if you think mplayer can do it, too, great :)
[16:13] <ibsk8in31> burek: I tried ffmpeg -i input.flv -acodec pcm_s16le -ar 8K -ac 1 output.flv -- It produced an .flv of the correct length.
[16:14] <burek> now just do: ffmpeg -i output.flv -vn output.wav
[16:14] <ibsk8in31> When I converted that .flv to .wav, I lost packets again!  The duration of the final file was only 0:35
[16:14] <burek> or: ffmpeg -i output.flv -vn -acodec copy output.wav
[16:15] <burek> I see
[16:15] <ibsk8in31> ok sweet.  what does -vn do?
[16:15] <burek> did you try playing output.flv
[16:15] <Stefff> indeed. My goal is to archive VHS Cassettes on a 3ghz pentium4 (i think). So I need fast encoding and the (nearly) best image quality I can get with it. So my choice was mpeg2 because the hardware is capable to handle the realtime encoding without an framedrop.
[16:15] <JEEB> try ffvhuff or something
[16:15] <JEEB> should be rather quick :)
[16:15] <burek> -vn means "no video"
[16:15] <burek> it removes video stream
[16:15] <JEEB> naturally lossless formats are also relatively bigger
[16:16] <Stefff> which means 50-100 times bigger :)
[16:16] <ibsk8in31> Still 0:35 when using -vn.  Trying -acodec copy
[16:16] <burek> Stefff, so you need 1) the best image quality, 2) fast encoding and 3) good compression rate (since you are archiving)
[16:16] <burek> those 3 never go together :)
[16:17] <Stefff> yepp :)
[16:17] <burek> 2/3 you can always manage at the cost of the 3rd :)
[16:17] <Stefff> yepp
[16:17] <ibsk8in31> Burek, still 0:35 :(
[16:18] <Stefff> so I think mpeg2 with 10Mbit Bitrate is an 80% Solution.
[16:21] <burek> Stefff, you could file a bug report maybe
[16:22] <burek> and provide a sample of your input?
[16:22] <Stefff> I dont think its a ffmpeg bug. Chances are better that I need the Bugreport ;)
[16:25] <Stefff> I use this cli http://pastebin.com/gaTg7y8x for capturing from a tvcard. Are those decent setting for mpeg2?
[17:17] <ibsk8in31> burek:  It's also still chopping all the silence when converting to .mp3
[17:17] <burek> ibsk8in31, can you submit a bug report regarding your issue please?
[17:18] <ibsk8in31> I'm doing a lot of searching, do you think it's my microphone settings on the flash client side?
[17:20] <burek> I don't know, but I wouldn't encode audio/video while doing a capture
[17:21] <burek> I would probably do it like this: http://ffmpeg.org/trac/ffmpeg/wiki/How%20to%20grab%20the%20desktop%20(screen)%20with%20FFmpeg
[17:24] <ibsk8in31> Oh I'm recording a conversation (Webcam and Microphone) through Flash Media Server, not screen capturing.
[17:24] <burek> http://ffmpeg.org/trac/ffmpeg/wiki/How%20to%20capture%20a%20webcam%20input
[17:32] <ibsk8in31> Thanks for the links, but I need this to be on a webpage like a flash object is.  Webuser A and Webuser B -> Record the conversation & Archive
[17:33] <JoeyJoeJo> I'm getting a message that the VBV buffer isn't set. How can I set it? I'm already using -bufsize and that didn't help
[17:40] <relaxed> JoeyJoeJo: pastebin
[18:19] <knoob> Hello. I created 15 thumbnails from a large video (over 4B) using this command : ffmpeg -i /tmp/media.ts -ss 00:08:33 -r 1/513 -s qvga -vframes 15 tn/%02d.png. But it took 18 minutes and a lot of memory and cpu, is there any way to optimize the process ?
[18:21] <relaxed> try moving the seek before the input
[18:24] <JoeyJoeJo> relaxed: Here is my command and its output - http://pastebin.com/Zv07qi4D
[18:28] <knoob> thanks relaxed, i'll try
[18:28] <JEEB> JoeyJoeJo, you need both bufsize and maxrate for vbv
[18:31] <creep> knoob<< yes, get a better computer, and it will be faster
[18:32] <creep> ( you can't just seek in a video file to any place that has keyframes )
[18:33] <knoob> of course it will be faster but it's a core i5
[18:34] <creep> 18 minutes using an i5 sounds much
[18:35] <Nils`> Hi ppl, i got a small question/pblm
[18:35] <knoob> yeah and it consumed like 4GB of RAM and 3GB of swap
[18:35] <creep> is it a bluray disc?
[18:36] <knoob> no just a recorded stream on a regular HDD
[18:36] <Nils`> how come that if I use this command  line "ffmpeg -i src.avi -vcodec copy -acodec copy dest.avi"
[18:36] <Nils`> i get a differrent bpp in src.avi and dest.avi
[18:37] <Nils`> 12 in src.avi, 24 in dest.avi, reported by mplayer
[18:38] <creep> knoob<< here are 2 modes seek and skip mode, skip mode plays the file and skips until specified time
[18:38] <Nils`> (i use ffmpeg version 0.8.3-4:0.8.3-0ubuntu0.12.04.1, it may not help)
[18:39] <knoob> mh ok
[18:39] <knoob> ss stands for seek set right ?
[18:39] <knoob> does it actually skip?
[18:40] <creep> i remember you have to put commands before or after -i to set skip or seek mode
[18:41] <ubitux> Nils`: this is not ffmpeg
[18:41] <ubitux> it's an old version of a forked project
[18:41] <knoob> ok, what relaxed suggested should speed up the process
[18:42] <ubitux> Nils`: please try with ffmpeg, we can't support that fork here
[18:42] <Nils`> ubitux: so i'll better compile the latest source and retry ?
[18:42] <Nils`> ok
[18:42] <knoob> I'll give it a try
[18:42] <ubitux> Nils`: at least we might be able to help you if it doesn't work
[18:42] <creep> knoob<< in seeking mode i guess ffmpeg extracts the next keyframe
[18:44] <Nils`> ubitux: thanks for the help. in 'real' ffmpeg, is there any way to control bpp apart from using pix_fmt ?
[18:45] <ubitux> doesn't it do what you want?
[18:46] <knoob> creep: ok but what ffmpeg does when I say "-b 1/513" does it skip or seek ?
[18:48] <Nils`> i'll give it a try with a fresh version once compiled. but in my ubuntu-packaged version, for example, according to the doc, yuv420p is meant to have 12bpp color depth, and in the re-encoded version, mplayer says it's 24bpp
[18:49] <Nils`> the original file, when played in mplayer, has 12bpp
[18:49] <ubitux> i'll have a look to your cmd line output
[18:49] <ubitux> when you'll upgrade :)
[18:49] <Nils`> sure. thx
[18:53] <JoeyJoeJo> JEEB: I tried adding bufsize and maxrate and I got a new error "MPEG1/2 does not support 15/1 fps" - http://pastebin.com/DMQ3UrB1
[19:00] <JEEB> well, I think that pretty much says what it means on the tin :)
[19:07] <JoeyJoeJo> I'm getting another error from libavcodec. This error is in my syslog and is generated from zoneminder - ERR [Unable to read packet from stream 0: error -541478725]
[19:07] <JoeyJoeJo> I'm trying to view the RTSP video stream from an IP camera
[19:28] <Nils`> ubitux: same result with the latest ffmpeg
[19:28] <Nils`> compiled from the git source
[19:29] <Nils`> though, i cannot exclude a pebkac in my ffmpeg command line ;)
[19:30] <ubitux> :)
[19:30] <gix> has anyone tried building ffmpeg with msvc and the c99 wrapper? what magic is needed so that configure (especially the cc/ld checks) completes?
[19:31] <ubitux> gix: check doc/plateform.texi
[19:35] <gix> that's what i did. either you guys use a different version of the c99 wrapper or something is missing. it states to make sure that msvc link.exe is used, but the wrapper calls it with -o instead of -out. hardcoding LD_O in the makefile changes that, but then link.exe cannot cope with /tmp-paths
[19:39] <Nils`> ubitux: http://pastebin.com/KC1KQxRD
[19:42] <smellynoser> Howdy - is it possible to output to a fifo?
[19:42] <smellynoser> Perhaps using stdout and redirecting to a fifo?
[19:45] <JEEB> named pipe? on *nix I /think/ you should be able to just output to the file
[19:47] <ubitux> Nils`: looks strange to me
[19:47] <ubitux> are you sure it's not a bug in mplayer?
[19:48] <Nils`> this idea came to my mind. actually i'm not sure
[19:48] <Nils`> is there a way to check this with ffmpeg ?
[19:49] <ubitux> you can ffprobe the file
[19:49] <ubitux> you'll see the pixfmt matches
[19:52] <ubitux> that information could be added
[19:52] <ubitux> give me a few minutes
[19:53] <Nils`> here you go : http://pastebin.com/nU9DWKx7
[19:54] <Nils`> not sure if it's related, but the bitrate seems almost doubled in the destination video
[19:54] <Nils`> compared to the source
[19:56] <Nils`> crap, it was the system's ffprobe
[19:56] <Nils`> with latest ffprobe http://pastebin.com/4pdtXvD0
[20:03] <ubitux> bits_per_pixel=12
[20:03] <ubitux> seems to be ok to me
[20:04] <ubitux> i have no idea why mplayer displays this
[20:04] <ubitux> but i tried this: http://pastie.org/4903834
[20:04] <ubitux> and ./ffprobe -show_streams raises 12
[20:05] <ubitux> which is normal, since the bits_per_pixel is computed based on the pix fmt
[20:07] <Nils`> i'll patch & test
[20:09] <Nils`> bits_per_pixel=12
[20:10] <Nils`> but it's retrieved from the pix fmt
[20:11] <ubitux> i don't think it would be even possible for ffmpeg to decode that file if that bpp was different
[20:12] <Nils`> so it makes sense to get the same result as the ffmpeg -pix_fmts
[20:12] <ubitux> so i'd really go for a printing problem with mplayer
[20:12] <Nils`> ok
[20:12] <Nils`> so the fact that the bitrate is doubled
[20:13] <Nils`> is unrelated to this ?
[20:18] <t4nk730> hello
[20:18] <t4nk730> i have a .mkv HD quality
[20:18] <Nils`> ubitux: i have to go, thx for your help
[20:19] <t4nk730> i need to convert it to .avi while keeping quailty
[20:19] <Nils`> and you've got some greetings from manfred
[20:19] <t4nk730> im on ubuntu btw
[20:20] <relaxed> t4nk730: ffmpeg -i input -c:v mpeg4 -vtag "xvid" -q:v 3 -c:a libmp3lame -ac 2 -b:a 192k output.avi
[20:21] <ubitux> Nils`: haha ok ;)
[20:21] <t4nk730> thanks relaxed going to give it a try now
[20:23] <t4nk730> what about ffmpeg -i output.avi -vcodec mpeg4 -b 4000k -acodec mp2 -ab 320k converted.avi ?
[20:36] <relaxed> why would you encode the avi?
[20:36] <t4nk730> im just doing it so i can play it back on my ps3
[20:42] <relaxed> The ps3 can decode h264 level 4.1 so you may just need to remux the streams into the mp4 container.
[20:43] <t4nk730> so that command wont work for ps3?
[20:44] <relaxed> pastebin the output of ffmpeg -i input.mkv
[20:46] <t4nk730> http://pastebin.com/r2TZK0cj
[20:47] <relaxed> do you require the subs?
[20:48] <t4nk730> no
[20:50] <relaxed> ffmpeg -i input.mkv -c:v copy -c:a libfdk_aac -ac 2 -b:a 256k output.mp4
[20:51] <relaxed> it may need to be,  ffmpeg -i input.mkv -map 0:v -map 0:a -c:v copy -c:a libfdk_aac -ac 2 -b:a 256k output.mp4
[20:51] <t4nk730> ok gonna try that one
[20:53] <t4nk730> will this keep the HD quality?
[20:54] <relaxed> yes, you're copying the video and encoding the audio to aac
[20:55] <t4nk730> thanks alot gonna wait untill this finish gonna take a min
[20:58] <t4nk730> think im gonna do this for now own instead of the shitty mediaserver
[21:00] <t4nk730> is there a way i can make subtitles only on non english parts?
[21:00] <t4nk730> or ill have to do it for the whole movie?
[21:04] <bitpurity> With the 1.0 release of ffmpeg, a new file named "time.h" had been added to the libavutil folder. This is conflicting with the ANSI time.h when trying to build ffmpeg. Has anybody found a work-around for this other than changing the ffmpeg code?
[21:06] <microchip_> bitpurity: this is known and has only one comment thus far https://ffmpeg.org/trac/ffmpeg/ticket/1783#comment:1
[21:09] <bitpurity> microchip_: thanks. I was using the wrong search option on the ffmpeg site
[23:32] <ibsk8in31> In Avisynth, Is there any way to be selective about what gets displayed for .Info()?
[00:00] --- Thu Oct  4 2012


More information about the Ffmpeg-devel-irc mailing list