[Ffmpeg-devel-irc] ffmpeg.log.20130913

[burek]

[00:00] <fazias> it was included in the pastebin, for how I found out the point where to split, well...
[00:01] <fazias> that involves a script that I copied from a stackoverflow answer
[00:39] <willwh> hi guys - I am curious, I've been testing with ffprobe: http progressive delivered content I can get bitrate data from FLVs
[00:39] <willwh> no dice with mp4 files though
[00:40] <willwh> and I've tried quite a few
[00:40] <willwh> runnin' 'em through mp4box they are valid
[00:40] <willwh> is that a codec thing, or, where would I be able to get some more info?
[00:40] <willwh> I'm hunting - if anyone has any experience - give me a yell :) (I'll idle for a few days)
[00:42] <willwh> ah............ scratch that :)
[00:42] Action: willwh slaps himself
[01:17] <chrisballinger> hey everybody
[01:18] <chrisballinger> other than reading the source, are there any examples on how to do the equivalent of "ffmpeg -i input.mp4 -f mpegts -vcodec copy -acodec copy -vbsf h264_mp4toannexb output.ts" with libavformat?
[01:19] <chrisballinger> ive been studying the muxing.c and demuxing.c examples in the docs but it's a little hard to wrap my head around those examples
[01:27] <vl4kn0> Hi, does format of audio frame after decoding it depend on codec used? I want to detect silent frame in video and I guess I just count all values that are close to zero and then compute percentage of the values and if the percentage is big enough (>98%) then the frame could be considered silent. I'm just not sure whether this algorithm can be applied to all codes?
[01:38] <llogan> vl4kn0: http://ffmpeg.org/ffmpeg-filters.html#silencedetect
[01:40] <vl4kn0> llogan: aha! I tried to find this but I was looking at the libav documentation
[01:40] <llogan> JoeyJoeJo: Junior Shabadoo
[01:41] <llogan> to me, libav means the FFmpeg libraries.
[01:43] <vl4kn0> except, there is no silecedetect filter in libavfilter
[05:56] <praveenmarkandu> hi. is there a way to transcode apple prores to h264
[05:56] <praveenmarkandu> but directly to main profile
[05:57] <praveenmarkandu> when doing prores > x264 it only allows me high422
[06:03] <relaxed> praveenmarkandu: use -profile main
[06:04] <relaxed> but keep in mind that the profile is also determined by the frame size/rate and other factors.
[06:05] <relaxed> http://en.wikipedia.org/wiki/H.264/MPEG-4_AVC#Profiles
[06:07] <relaxed> er, I'm thinking of levels, but the profile does take color space and bit depth
[06:09] <relaxed> praveenmarkandu: pastebin you're command and output, then I'll be able to see what's going on
[06:09] <relaxed> your*
[06:18] <praveenmarkandu> ok
[06:20] <praveenmarkandu> relaxed: http://pastebin.com/XHfQRNdj
[06:25] <relaxed> praveenmarkandu: add -pix_fmt yuv420p
[06:26] <praveenmarkandu> oh. is the colour space not letting me encode at main profile?
[06:26] <relaxed> color space and bit depth
[06:27] <relaxed> your source is 10bit and yuv422p
[06:28] <relaxed> main only supports 8bit yuv420p
[06:29] <praveenmarkandu> ok. gotcha
[06:29] <praveenmarkandu> thanks
[06:34] <praveenmarkandu> relaxed: btw, for an h264 10mbps source file ,if i do a -vcodec copy, and also use -b:v 2000k will it transcode it to 2mbps?
[06:36] <relaxed> -vcodec copy is a stream copy
[06:37] <relaxed> it's one or the other
[06:38] <praveenmarkandu> oh okay. so no
[06:52] <chrisballinger> heya are there any good examples of doing bitstream copies (and with h264_mp4toannexb) with just libavformat?
[08:54] <praveenmarkandu> hi guys
[08:54] <praveenmarkandu> is there a way to fix the bitrate on a prores file?
[08:54] <praveenmarkandu> or do i just rely on -profile:v and let that do the work?
[09:33] <cbsrobot_> praveenmarkandu: in prores_ks you can set bits_per_mb and/or mbs_per_slice
[09:33] <praveenmarkandu> oh. thanks
[09:34] <praveenmarkandu> i assume the difference between prores and prores_ks is minimal
[09:34] <praveenmarkandu> in terms of quality
[09:35] <cbsrobot_> honestly I don't know exactly, but prores_ks supports 4444
[09:36] <praveenmarkandu> oh okay
[09:36] <praveenmarkandu> thanks
[09:36] <praveenmarkandu> got questions answered in #ffmpeg. today was a good day
[09:39] <cbsrobot_> praveenmarkandu: well then just do ma a favor and help someone else today too ;-p
[09:41] <praveenmarkandu> my knowledge of ffmpeg is very rudimentary
[10:44] <rooty> i know this is likely not the right channel but i was wondering if anyone know of a method to take an input stream (icecast) convert it to lower quality stream rates and rebroadcast ?
[11:12] <Jenser> hello everyone. is there a way to write the scale=xxx:yyy command only once when merging serveral files together with -filter_complex?
[11:14] <durandal_1707> depends what filters you use
[11:21] <Jenser> i am trying to combine several videos (already made to the same format), for not losing to much quality the first step was to convert them to a common intermediate format (19020x1080 at 10MBit), now i want to combine them to one long clip. concat: ...  worked fine with -vf yadif (... ),scale=xxx:yyy as output, but with -filter_complex this seams not to work. for me it looks like i have to define the size every video input file, although it's all 
[11:26] <kein> Hi there
[11:27] <durandal_1707> Jenser: it get cut of, could you pastebin both commands?
[11:28] <kein> I got some troubles with audio filers "channelsplit" end "amerge". When processing 4channels -> 4 streams it's fine. When 4 channels -> 2 streams. the second stream is mixed ? First stream is fine ???
[11:30] <BoR0> does anyone have any experience with DirectSound? basically I need a queue management with audio buffer rather than setting notifications on position and filling buffer accordingly. is it possible to do this with plain DirectSound API or does there exist any nice wrapper for this?
[11:31] <BoR0> I have a C++ library that wraps around libffplay, and instead of doing while (position != y...) fetch audio from ffmpeg I need to fill packets to directx as soon as ffplay has data for me
[11:32] <kein> ffloger > there you go http://pastebin.com/HmWySybD
[11:34] <kein> I got some troubles with audio filers "channelsplit" end "amerge". When processing 4channels -> 4 streams it's fine. When 4 channels -> 2 streams. the second stream is mixed ? First stream is fine ??? Find details here http://pastebin.com/HmWySybD
[11:45] <kein> I got some troubles with audio filters "channelsplit" end "amerge". When processing 4channels -> 4 streams it's fine. When 4 channels -> 2 streams. the second stream is mixed ? First stream is fine ??? Find details here http://pastebin.com/HmWySybD
[11:45] <durandal_1707> that is not COMPLETE output
[11:53] <kein> durandal_1707 http://pastebin.com/J6WaCFMQ
[11:54] <kein> I got some troubles with audio filters "channelsplit" end "amerge". When processing 4channels -> 4 streams it's fine. When 4 channels -> 2 streams. the second stream is mixed ? First stream is fine ??? Find details here http://pastebin.com/J6WaCFMQ
[11:54] <Jenser> @durandal http://pastebin.com/Es2pSFMF
[11:54] <Jenser> this command ignores the scale @ -vf filter
[11:55] <Jenser> thatswhy i want to move this command to the complex_filter for defining the output video size
[11:56] <kein> Jenser : put all the command "yadif, scale" in the filter complex. It should do the trick
[11:57] <kein> Jenser : -filter_complex "concat, yadif=... , scale=..."
[11:57] <Jenser> to which position? don't want to define this for each video input file
[11:57] <Jenser> the documentation for complex filter isn't the very best, i think...
[11:58] <kein> Jenser : first you "concat" so the scale and yadif are placed after, so it will process the output of concat
[11:59] <kein> Jenser : and i agree with the documentations, not enough examples :s
[11:59] <kein> [AFK] dinner time
[11:59] <Jenser> mahlzeit
[12:06] <durandal_1707> kein: i do not see how amerge could mix channels
[12:37] <kein> [back]
[12:38] <kein> Durandal_1707 : I do not see too, but in audition channels 3&4 are mixed. Don't no if bug or if I missed something
[12:39] <kein> Hi guys/ I got some troubles with audio filters "channelsplit" end "amerge". When processing 4channels -> 4 streams it's fine. When 4 channels -> 2 streams. the second stream is mixed ? First stream is fine ??? Find details here http://pastebin.com/J6WaCFMQ
[12:57] Last message repeated 1 time(s).
[12:58] <durandal11707> kein: i do not see how it could be mixed
[12:58] <durandal11707> by mixed, what you mean?
[13:00] <kein> durandal11707 : when listening with headphones and selecting audio "french", left and right sounds are different (test file). when selecting "english" left and right are mixed. Original file is not
[13:00] <kein> durandal11707 I got the files and a good upload if you want
[13:02] <kein> durandal11707 : by mixed I mean in left speaker I can ear both sounds (3 & 4) same with the right speaker
[13:03] <durandal11707> that looks like bug
[13:04] <kein> durandal11707 : i'm not the kind of guy to scream bug quickly. maybe you can test on your hardware ? ^^
[13:22] <kein> durandal11707 : when amerge with source (4 streams - mono) I don't get the bug...
[13:28] <durandal11707> perhaps you could use asplit and pan filter?
[13:28] <durandal11707> this could also be faster
[13:30] <kein> durandal11707 : asplit : copy input to multiple output... it's not the goal here. (if I get it)
[13:32] <durandal11707> no you split it to 2 outputs and use pan on all of them
[13:33] <durandal11707> you just do downmix from 4 to 2(stereo)
[13:33] <kein> durandal11707 : i'm not familier with this filter; I'll try it right now ;)
[13:37] <kein> durandal11707 : can we mp ?
[13:46] <Jenser> @kein can you propably tell me the correct positioning of the scale command? i didn't get it running yet. command is so far ... -filter_complex '[x:y] (...) concat=n=5:v=1:a=1 [v] [a],scale=xxx:yyy' -map
[13:46] <Jenser> this obviously isn't the right position for scale
[13:48] <kein> Jenser : -filter_complex '[0:0] [0:1] [1:0] [1:1] [2:0] [2:1] [3:0] [3:1] [4:0] [4:1] concat=n=5:v=1:a=1 [v] [a], yadif=1:-1:1,scale=720:576'
[13:50] <kein> oups : -filter_complex '[0:0] [0:1] [1:0] [1:1] [2:0] [2:1] [3:0] [3:1] [4:0] [4:1] concat=n=5:v=1:a=1 [v] [a], [V] yadif=1:-1:1,scale=720:576'
[13:53] <Jenser> output with label [v] does not exist in any defined filter graph or was already used elsewhere
[13:53] <Jenser> do i need to put this command behind the "-map" command?
[13:56] <kein> Jenser : mp
[13:58] <Jenser> urban dictionary tells me: mp = military police
[13:58] <Jenser> don't think you wanted to tell me this ;)
[14:01] <kein> privite message :p
[14:01] <kein> *private
[14:01] <kein> ffmpeg -i $FILES [audio-options] [video-options] -filter_complex '[0:0] [0:1] [1:0] [1:1] [2:0] [2:1] [3:0] [3:1] [4:0] [4:1] concat=n=5:v=1:a=1; [0:v] yadif=1:-1:1,scale=720:576' -y test.avi
[15:59] <vl4kn0> Hi, is it guaranteed that video and audio frame arrive in the same packet?
[16:52] <mcnulty_> hi
[16:54] <mcnulty_> I would like to change /tmp folder for recording/screencast...
[19:27] <chrisballinger> hi, are there any good examples on how to use libavformat to change the container of media without re-encoding?
[20:41] <ZaB|SHC|> hi there. i've got a problem applying two audio filters. i want to record from three sources, mix them into one stream and then increase the volume. my command at the moment looks like: ffmpeg -f pulse -i app1.monitor -f pulse -i app2.monitor -f pulse -i default -filter_complex 'amix=inputs=3[mixed]; [mixed]volume=volume=2[out]' test.mp3. this leads to: Output pad "default" with type audio of the filter instance "Parsed_volume_1" of
[20:41] <ZaB|SHC|>  volume not connected to any destination
[20:53] <Barbariandude> Hi guys! I'm trying to stream to twitch.tv with both my computer's audio output and my microphone's input. All I'm getting if I put "-i default" is my microphone in really scratchy shitty quality. If I put "-i hw:0,0" I get alsa errors saying "Device in use"
[20:53] <Barbariandude> Anyone got any ideas?
[20:53] <llogan> ZaB|SHC|: add -map "[out]"
[20:54] <llogan> or remove [out]
[20:54] <llogan> but being explicit is probably better
[20:55] <Barbariandude> llogan: In the ffmpeg command itself, after "-f alsa"?
[20:55] <Barbariandude> llogan: Oh, responding to a diff guy
[20:55] <Barbariandude> llogan: nvm!
[20:55] <ZaB|SHC|> llogan: thx, removing [out] helped
[20:56] <llogan> ZaB|SHC|: you can shorten it: 'amix=inputs=3,volume=volume=2'
[20:57] <llogan> but again, it's usually a good idea to use named links
[20:58] <Barbariandude> fflogger: http://pastebin.com/hPqdMb8s
[20:58] <ZaB|SHC|> llogan: thx, i'll stick to the named links
[20:58] <llogan> fflogger is a bot
[20:58] <Barbariandude> llogan: Oops :P
[20:58] <llogan> the complete console output is missing
[20:58] <Barbariandude> llogan: Yup doing that now
[20:59] <Barbariandude> llogan: Ummm... this is a new error... http://pastebin.com/Ba2YXfdb
[21:00] <llogan> there is no context for the error without the actual, unscripted command
[21:01] <Barbariandude> llogan: I had an additional space
[21:01] <Barbariandude> llogan: Fixed that error
[21:02] <Barbariandude> llogan: http://www.twitch.tv/barbariandude
[21:02] <Barbariandude> I am both speaking into my mic and have music playing
[21:02] <Barbariandude> No alsa input is going through
[21:02] <Barbariandude> What's wrong with my command?
[21:03] <llogan> i don't know. i'm waiting for the console output.
[21:06] <Barbaria2dude> llogan: http://pastebin.com/BvTL1dzx
[21:08] <llogan> also show output of: arecord -l
[21:10] <Barbaria2dude> llogan: Ok, arecord -l showed card 0 device 0, so I changed "-i default" to "-i hw:0,0" and I got microphone output, but not actual audio. 1 second to put aplay -l and arecord -l into a pastebin
[21:10] <Barbaria2dude> http://bpaste.net/show/132196/
[21:11] <Barbaria2dude>  http://bpaste.net/show/132197/
[21:12] <llogan> does it sound shitty if you use arecord? then you know it's not ffmpeg.
[21:13] <llogan> other than that...i don't know. burek knows more about alsa than i do. also see: https://trac.ffmpeg.org/wiki/Capturing%20audio%20with%20FFmpeg%20and%20ALSA
[21:13] <chrisballinger> hi, are there any good examples on how to use libavformat to change the container of media without re-encoding?
[21:15] <llogan> did you see docs/examples/demuxing.c and muxing.c?
[21:15] <llogan> i assume they may be useful, but I am only a cli tool user
[21:15] <Barbaria2dude> llogan: No, mic sounds fine in arecord. Mic also sounds fine in ffmpeg now that I put hw:0,0 instead of letting ffmpeg autoselect. Only problem is it's not recording the computer's audio output as well. aplay -l shows card 0 device 0 for output, which is the same as arecord -l which confuses me
[21:16] <chrisballinger> yeah i saw the examples but it is unclear how to do the equivalent of -vcodec copy
[21:16] <Barbaria2dude> llogan: So I'm not sure how to mix both into a stream
[21:17] <llogan> chrisballinger: try libav-user mailing list if you don't get an answer here or by looking at the code
[21:17] <llogan> Barbaria2dude: http://ffmpeg.org/ffmpeg-filters.html#amerge
[21:17] <chrisballinger> yeah ive been studying the code for the last few days
[21:18] <llogan> Barbaria2dude: https://trac.ffmpeg.org/wiki/AudioChannelManipulation
[21:20] <Barbaria2dude> llogan: Thank you for your help in narrowing this down. I'll ask #alsa why both my microphone and my audio output are both card 0 device 0.
[21:20] <llogan> good luck.
[21:58] <Barbaria2dude> llogan: Ok, apparently hw:0,0 is both my mic and my headphone output, it's just my headphone output is incredibly quiet. If both my mic and my headphone output are on the same subdevice, is there any way to alter the sound levels in the command?
[21:58] <Barbaria2dude> llogan: Or is it just a hard reality that I can't mix the audio levels?
[21:59] <Barbaria2dude> I've found numerous guides for mixing audio levels of different subdevices, nothing for the same subdevice
[22:16] <klaxa> Barbaria2dude: you will most likely need a mixer-board or two independent sources
[22:17] <klaxa> which would be two different subdevices then probably
[22:18] <Barbaria2dude> klaxa: I've managed to kludge the solution by just changing volume levels (increasing output to near-deafening levels and reducing the volume via a hardware thing on the headphones), now I'm facing a different problem
[22:18] <Barbaria2dude> klaxa: Video is fine on the stream, audio is cutting out every 20 seconds or so in time with "alsa buffer xrun"
[22:19] <klaxa> i never got those to stop :x
[22:19] <klaxa> maybe wrong sampling rate?
[22:19] <klaxa> that's the only thing that comes to mind
[22:20] <Barbaria2dude> klaxa: What sample rate would you recommend?
[22:20] <klaxa> the one alsa produces
[22:21] <klaxa> or maybe exactly half of it
[22:21] <klaxa> or something
[22:21] <klaxa> streaming 96khz is somewhat unncessary, don't you agree? :)
[22:22] <Barbaria2dude> klaxa: Absolutely, but where do I check alsa's sample rate?
[22:22] <Barbaria2dude> The one it produces, I mean
[22:22] <klaxa> ffmpeg will list it as an input
[22:23] <klaxa> but now that i think about it
[22:23] <klaxa> that shouldn't even matter
[22:23] <klaxa> alsa buffer xruns are mysterious to me
[22:24] <Barbaria2dude> klaxa: http://pastebin.com/egtP1cWe What should my sample rate be from that?
[22:25] <klaxa> 48khz right?
[22:26] <Barbaria2dude> klaxa: Giving me an incorrect parameter error on that
[22:27] <klaxa> pastebin the command line + complete output
[22:29] <Barbaria2dude> klaxa: http://pastebin.com/BnnUXq9B
[22:30] <klaxa> ah yes, good ol' flv container
[22:30] <klaxa> you will need to resample to 44.1khz
[22:30] <klaxa> no matter what
[22:30] <klaxa> because flv is weird like that
[22:31] <Barbaria2dude> klaxa: http://www.twitch.tv/barbariandude
[22:32] <Barbaria2dude> klaxa: Talking continuously, cutting out all the time
[22:33] <klaxa> new beer is always good
[22:33] <Barbaria2dude> Indeed.
[22:34] <Barbaria2dude> I need alcohol to prevent myself from tearing my computer into pieces
[22:35] <klaxa> i personally prefer to use pulseaudio for recording those things
[22:35] <Barbaria2dude> *sigh* I was hoping you wouldn't say that
[22:35] <Barbaria2dude> I'll use it if I have to
[22:35] <Barbaria2dude> But I hate it
[22:35] <klaxa> it's just not deterministic
[22:35] <klaxa> that is what i hate most about pulseaudio
[22:35] <klaxa> other than that, it's pretty nice
[22:35] <klaxa> leaving out the part where it eats CPU like a whore
[22:36] <Barbaria2dude> What I hate most is that it tries to be too clever and mutes random shit and I need to go to pavucontrol to unmute them
[22:36] <Barbaria2dude> No pulseaudio, I WANT both audio and teamspeak playing audio at the same time!
[22:36] <Barbaria2dude> *music and teamspeak
[22:37] <klaxa> there *should* be a setting for that
[22:38] <klaxa> i don't know what causes it either though
[22:44] <Barbaria2dude> klaxa: All this trouble to stream some ownage meepo dota 2 gameplay on gentoo :P
[22:45] <Barbaria2dude> Hopefully it'll be worth it
[22:48] <klaxa> now imagine all the "trouble" writing the code you are using :P
[22:55] <willwh> klaxa: whores eat cpu?
[22:55] <klaxa> maybe?
[22:55] <willwh> :]
[22:55] <willwh> hahaha
[22:56] <klaxa> i don't know any whores so i wouldn't know :V
[00:00] --- Sat Sep 14 2013