[Ffmpeg-devel-irc] ffmpeg.log.20140116

burek burek021 at gmail.com
Fri Jan 17 02:05:01 CET 2014


[00:00] <l_r> is there a way to mux h264 raw video from a device with a clock  in a mpegts with timestamps?
[00:00] <l_r> i hope what i said makes sense
[00:01] <l_r> basically i have raw h264 without timestamps
[00:09] <llogan> jedir0x: make sure the output is yuv420p
[00:11] <jedir0x> llogan, it is - and i figured it out with help of the mailing list.
[00:11] <jedir0x> thanks though!
[00:23] <caribou|> hello
[00:24] <caribou|> is there something wrong with that command 'ffmpeg -r 20 -i "img (%d).png" -q:v 1 -b:v 1500k timelapse.mp4' ?
[00:24] <caribou|> "Ôãimg: No such file or directory"
[00:25] <caribou|> it doesn't seem to like my spaces even though i'm surrounding the file name with quotes
[00:28] <caribou|> used another command and it works so i guess the reported error doesn't help much :-O
[00:34] <llogan> worksforme
[00:42] <thebombzen> also caribou| I don't know why you're using -q:v and -b:v, becaue -q:v sets VBR quality and -b:v sets ABR bitrate, which don't work together AFAIK
[00:43] <caribou|> thanks ! I naively copy pasted
[01:00] <mastereve> Hello all o/
[01:02] <jedir0x> Hola compadre
[01:06] <mastereve> Hey did anyone make the demuxing example code work as the CLI counterpart?
[01:07] <mastereve> I get some wierd results from that code
[01:56] <dtb> Hello. A little confused about 'filter_complex'. Trying a few examples from here: http://www.ffmpeg.org/ffmpeg-filters.html but getting "Unrecognized option 'filter_complex'" - I even tried downloading a recent static build linked from ffmpeg.org and that still didn't have it. Is it still available?
[01:56] <Hello71> -c:a vs -acodec?
[01:57] <Hello71> dtb: full command?
[01:59] <dtb> Hi Hello71 - http://pastie.org/8637502 contains the output.
[02:00] <Hello71> maybe you need 2.x?
[02:01] <Hello71> ah, -acodec is an alias for -codec:a
[02:01] <Hello71> so is -c:a
[02:02] <dtb> Hello71, doh - I thought all these would be latest: http://ffmpeg.gusari.org/static/64bit/ - grabbing 2.0.1 from http://johnvansickle.com/ffmpeg/ now.
[02:07] <dtb> Hello71, thanks for the help! ffmpeg is doing its thing now :)
[02:38] <dtb> hello again, I'm wondering if it's possibly to change the example from https://trac.ffmpeg.org/wiki/How%20to%20speed%20up%20/%20slow%20down%20a%20video so that it will speed up the video AND audio by a factor of 4. I tried (incorrectly?) to string multiple atempo's together in filter_complex but that only gave a 2x audio speedup:
[02:38] <dtb> ffmpeg -i input.mkv -filter_complex "[0:v]setpts=0.5*PTS[v];[0:a]atempo=2.0[a];[0:a]atempo=2.0[a]" -map "[v]" -map "[a]" -map "[a]" -map "[a]"
[02:39] <dtb> > correction:  ffmpeg -i input.mkv -filter_complex "[0:v]setpts=0.25*PTS[v];[0:a]atempo=2.0[a];[0:a]atempo=2.0[a]" -map "[v]" -map "[a]" -map "[a]" -map "[a]"
[04:21] <thebombzen> dtb: did you try this: ffmpeg -i input.mkv -filter_complex "[0:v]setpts=0.25*PTS[v];[0:a]atempo=2.0,atempo=2.0[a]" -map '[v]' -map '[a]'
[04:22] <thebombzen> also, remember to multiply the framerate of the output so you don't drop frames
[04:22] <llogan> he's gone, jim
[04:22] <thebombzen> "{
[04:22] <thebombzen> :P *
[04:23] <llogan> but you do have a correct answer
[04:23] <thebombzen> though I don't know why you would use complex filtergraphs if you can avoid it. using simple filtergraphs whenever possibly is more intuitive and easier to understand the command-line
[04:24] <thebombzen> I'd much prefer -filter:v "setpts=0.25*PTS" and -filter:a "atempo=2.0,atempo=2.0"
[04:24] <thebombzen> the [0:v] and [a] stuff makes the command line harder to parse (for humans)
[06:47] <basilgohar> Is it possible, if I have 4 videos, to combine them into one with each video playing in a corner, taking-up 1/4 the total screen?  If so, what filters or other arguments would I use to achieve that?
[06:47] <basilgohar> I *think* it's possible using overlay, but I think that'll be overly complex if there's another filter that can explicitly position and scale the videos in question...
[09:19] <GT1> Hello guys, I have a quick question. ffmpeg can output gif images but the size seems just too damn big. Is there a way to reduce the size of it within ffmpeg?
[09:21] <DeadSix27> i think gif in ffmpeg there was no further way to decrease the size
[09:21] <DeadSix27> for*
[09:21] <DeadSix27> i always just ran it through an optimizer after
[09:23] <ubitux> GT1: try -pix_fmt rgb8
[09:23] <DeadSix27> ye and that
[09:23] <DeadSix27> right, saw that on stackoverflow once.
[09:25] <GT1> k, i'll try
[09:26] <GT1> that halfed the size
[09:26] <GT1> thx :)
[09:26] <GT1> amI right that a gif can only have 256 color palate? cuz the image seems to be pixelated every time
[09:27] <sspiff> anyone present that has a bit of experience with MPEG transport stream handling?
[09:27] <ubitux> GT1: yes but right now it's the default palette
[09:27] <ubitux> GT1: so it's only relying on the dithering to get a better output
[09:28] <ubitux> it would be nice if the palette was better because the pal8 pix fmt is the default&
[09:28] <sspiff> I've got a stream, which contains only one PID, no PSI's or anything, just one subtitle stream and I'm wondering if there's an easy way to have avformat handle the TS packet parsing for me
[09:28] <GT1> so there is no way making it better?
[09:31] <GT1> I mean is there a way to choose a custom color palate? to make the output better?
[09:32] <ubitux> no that i know
[09:32] <ubitux> not*
[09:32] <GT1> k, thx
[09:32] <ubitux> patches welcome :)
[09:32] <ubitux> but maybe i'm wrong :p
[09:33] <GT1> :) it would be nice, but I don't have nor the knowledge nor the time for that :( maybe some day I hope.
[09:35] <GT1> The problem is that I'm costumed to higher level of programming ( generally using some api, sdk or something like that), but with low level c++ programming, not to mention image processing is out of my league :) But if you need a game a website or an app to be done you can count on me :)
[09:36] <GT1> and by low level, not by skill i mean but technical meaning :P low level is lot harder :D
[09:59] <Amnesia> is anyone over here familiar with "ismv" and "isma" streams?
[10:00] <Amnesia> I'd like to rip one, but I don't know how
[10:00] <Amnesia> the stream consists of mpeg4 fragments
[10:11] <bencc1> when concatenating several movies with the same codec and parameters, does it decode/reencode or just stich them together?
[10:14] <juanmabc> tsup!
[10:15] <juanmabc> can i apply ladspa filters through ffmpeg?
[10:15] <juanmabc> and extra mile, programaticall through libav?
[10:16] <juanmabc> game changer here ;)
[10:36] <allengreen> how to calculate the elapsed time in millisecond between two frames from AVPackets?
[10:37] <lilalinux> how do I recode a mkv from hi10p(10bit) to ordinary (8bit?)
[10:39] <juanmabc> allengreen: you are going to need to study pts's
[10:39] <Mavrik-> allengreen, that's a wierd question, can you elaborate more?
[10:39] <juanmabc> 'presentation time stamps'
[10:39] <allengreen> I read, but I am kind of stupid.
[10:39] <juanmabc> is programatically not weird Mavrik-
[10:40] <allengreen> so if I have two AVPackets,  how to get the elapsed time in milliseconds?
[10:40] <Mavrik> ah
[10:40] <Mavrik> allengreen, PTS and DTS are timestamps of your packets
[10:41] <Mavrik> they're just in different units - you need to multiply them with time base of your stream to get seconds
[10:41] <Mavrik> if your packet is missing PTS take DTS
[10:42] <Mavrik> so... calculate difference between PTS/DTS and multiply by streams timebase
[10:42] <allengreen> I'll try again, I tried millions of times.
[10:43] <Mavrik> remember that timebase is stored like a fraction
[10:43] <Mavrik> e.g. a TS stream will have timebase of 1/90000
[10:44] <juanmabc> decoding time stamps are not going to make it sometimes
[10:44] <juanmabc> the last resort when not present, use an user clock from fps
[10:45] <Mavrik> mhm
[10:45] <juanmabc> so basically you need a logical chain, if pts else if dts else assume fps
[10:45] <Mavrik> I think there's a very handy av_guess_timestamp method
[10:45] <Mavrik> that does exactly that
[10:45] <juanmabc> i'd love ffmpeg integrated itself some pts calculations already
[10:46] <juanmabc> av_guess_timestamp, note taken, i'll try that too
[10:46] <Mavrik> ah, you need to decode video first:
[10:46] <Mavrik> http://ffmpeg.org/doxygen/trunk/frame_8h.html#acb9d9082c10e625e51a68caceabd4458
[10:47] <Mavrik> av_frame_get_best_effort_timestamp
[10:47] <juanmabc> ah, that i do use ;)
[10:49] <juanmabc> i'd love someone to give me that when i satrted to do video, allengreen https://code.google.com/p/openmedialibrary/
[10:52] <juanmabc> it was done at ffmpeg-0.11 or so, hence it lacks packed/aligned audio consideration so the sound is gonna be bad, that is done and gonna be merged in another place: https://code.google.com/p/openxl/source/browse/src/FL/flSound.c
[10:53] <juanmabc> mixed audio & video on open*l, but since both use a get in memory then pass it, can help a lot of projects
[10:54] <juanmabc> i'm currently at it, so it's super messy ;)
[10:54] <juanmabc> have fun, /me moves on
[10:54] <juanmabc> ladspa anyone?
[10:55] <juanmabc> you are going to make me research, he
[10:55] <juanmabc> see you
[11:17] <sledges> hello
[11:18] <sledges> I got an mp4 file, when I split it, and some of the splits i speed up, then combine all back, I get jumps (to keyframe?) and delays
[11:18] <sledges> how to convert it to some raw format, do the editing, and convert it back?
[11:23] <sledges> found: http://stackoverflow.com/questions/15778774/using-ffmpeg-to-losslessly-convert-yuv-to-another-format-for-editing-in-adobe-pr
[12:07] <sspiff> when I create an AVIOContext, will ffmpeg seek forward/backward a lot? Do I have to support seeking? How much should I buffer?
[13:25] <__raven> how to get a boolean answer out of edge detevtion? i need some technical filters to dezect blury and black video files. any ideas?
[13:35] <pisto> hello. is there a table that explains all the available pixel formats?
[13:35] <pisto> not just list them
[15:04] <bencc> I'm trying to concatenate video files
[15:04] <bencc> this is the mylist.txt file http://dpaste.com/1554633/
[15:04] <bencc> I'm getting an error: Line 1: unknown keyword '  f'
[15:06] <bencc> this is the full console (windows) http://dpaste.com/1554640/
[15:33] <bencc> it was a bad encoding
[15:38] <__raven> how to get a boolean answer out of edge detevtion? i need some technical filters to dezect blury and black video files. any ideas?
[16:38] <sspiff> the buffer in AVIOContext, is that for use in my read/write functions or is that used by ffmpeg to buffer the output of the read function?
[17:03] <k0bi> ok i am extracting mp3 with ffmpeg and i want to do it in cuts like each min file
[17:03] <k0bi> C:\Users\Administrator>"E:\wwwroot\base64Player\ffmpeg"  -i "ip live=1" -vn -ar 44100 -ac 2 -ab 128k  "C
[17:03] <k0bi> :\ffmpeg\xrxxddddxptLtx-%03d.mp3"
[17:04] <k0bi> i was thinking to add -itsoffset 60 but no luck with that
[17:09] <thebombzen> k0bi: use -ss to skip and -t to cut the duration (both in seconds, or in hh:mm:ss form). I would write a script to do it repeatedly. That is, to skip 3 minutes in from the beginning and only encode 1 minute, use -ss 3:00 -t 1:00
[17:25] <Wma> hello all
[17:26] <Wma> how can one tell if an audio file was encoded to AAC with Nero or FAAC?
[17:26] <Wma> Does FAAC write a special header?
[17:32] <jnvsor> Wma: in ffmpeg or in a lib?
[17:33] <Wma> well, I just want to see if it's possible to check a file to find out which AAC encoder was used
[17:34] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/aacdec.c;h=365df2117d17a48a901c52849c3b585b042f3906;hb=HEAD#l2230
[17:34] <JEEB> see this function
[17:34] <JEEB> the sscanf check for the string libfaac with a version
[17:34] <Wma> wow that's just gibberish for me I'm afraid :s
[17:34] <JEEB> so yes, it seems like libfaac does write some header information
[17:34] <JEEB> into the file
[17:34] <Wma> can I check it with a HEX editor?
[17:35] <JEEB> you can try, but not exactly sure where that is written. Also the lack of that doesn't mean anything, either
[17:36] <JEEB> just that if that branch in the code gets executed (as in, that string is there), then it most probably was encoded by faac
[17:36] <Wma> any hint how that header starts?
[17:37] <jnvsor> Wma: ffmpeg -i file
[17:38] <jnvsor> that will tell you the file codecs etc, should be fine for normal use
[17:38] <JEEB> it seems like he specifically wants to know the encoder used
[17:38] <JEEB> not just the format
[17:39] <Wma> yup :)
[17:40] <Wma> i'm just curious because some website is selling AAC audio files
[17:40] <Wma> i'm just curious about the encoder used etc
[17:48] <pisto> Wma, try ffprobe ./file
[17:49] <pisto> if you're lucky it outputs the info you need
[17:49] <pisto> or strings ./file
[18:09] <bgilbert> Hi, I was wondering if there was a way to transcode rtp packets in ffmpeg using only stdin/stdout?
[18:36] <jnvsor> bgilbert: Yes, are you outputting to another rtmp stream?
[18:38] <bgilbert> jnvsor: yea, well rtsp currently.....I was hoping to stream packets in through standard in and get them streamed back through standard out. Then send them out to another stream
[18:57] <jnvsor> bgilbert: You can by using pipe:0, pipe:1 input and output, but ffmpeg supports streams in/out without needing to pipe to other programs
[19:16] <bgilbert> jnvsor: thanks, I'll go check that out
[19:41] <Guest18902> trying to convert a video for 3ds.. using command. "mpeg -y -i "left.avi" -i "right.avi" -vcodec copy -acodec adpcm_ima_wav -ac 2 -vcodec copy -map 0:0 -map 0:1 -map 1:0 "VID_0001.AVI"
[19:42] <Guest18902> getting an error 'number of stream maps must match number of output streams.   any ideas?
[20:09] <Guest18902> messed stuff up reconverting so i can paste to u guys =/
[21:07] <llogan> Guest18902: any luck on that paste?
[23:09] <Guest18902> finally
[23:10] <Guest18902> http://pastebin.com/ifJkewwf   line 534 is where it starts.
[23:11] <Guest18902> I convert .mkv to a .avi and crop it (3d movie) then so same for right (but quit it cuz i already did it prior) then error trying to combine them
[23:11] <Guest18902> someone please help thanks
[23:11] <Guest18902> 400x240 is ideal screen size..
[23:14] <llogan> Guest18902: that's the fake ffmpeg. it's not from FFmpeg. so understandably we can't support it here
[23:15] <llogan> you can download a build of ffmpeg
[23:15] <llogan> or compile ffmpeg. https://trac.ffmpeg.org/wiki/UbuntuCompilationGuide
[23:15] <llogan> or get help from the fork somehow
[23:26] <Zeranoe> Does anyone know what kind of material I'm dealing with here: http://paste.debian.net/76683/ I added the line breaks at logical spots, but the patterns might be longer. I can resort it if it helps. I want to know how to properly deinterlace it
[23:32] <Guest18902> dam
[23:33] <Guest18902> ubuntu ships with fake ffmpeg?
[23:33] <Guest18902> echo #ubuntu >> u are a bunch of slackers.
[23:34] <Guest18902> llogan: once i compile a proper version -- are all the commands I'm running right?
[23:43] <llogan> Guest18902: which command. your paste contained 12 of them
[23:57] <Zeranoe> Did I not supply the right information?
[00:00] --- Fri Jan 17 2014


More information about the Ffmpeg-devel-irc mailing list