[Ffmpeg-devel-irc] ffmpeg.log.20150114

burek burek021 at gmail.com
Thu Jan 15 02:05:01 CET 2015


[00:29] <fahadash> Hello
[00:29] <fahadash> Is there any way to figure out the flv file's bit-rate ?
[00:29] <fahadash> Its a sound file
[01:40] <compstomp> Hi guys, I see there is now a static build of ffmpeg for Mac OSX (its been a while).  What sort of trickery was needed to build this / would it be extremely difficult to replicate on a typical developer-oriented macbook pro? Thanks!
[01:47] <gustavo_> I'm trying to rotate a video with ffmpeg
[01:47] <gustavo_> but its not working ...
[01:48] <gustavo_> can someone help me ?
[01:48] <compstomp> They will immediately prompt you to do a pastebin (or otherwise) of the commands you input (exactly as you have) and the output the console gives you.
[01:49] <compstomp> So if you do that now it will save you and them some time :)
[01:53] <gustavo_> compstomp: http://pastebin.geany.org/rVCdO/
[01:58] <klaxa> gustavo_: try adding: -c:a copy
[01:59] <gustavo_> klaxa: after "-i" ?
[01:59] <gustavo_> ops
[01:59] <klaxa> after the mp4 input file
[01:59] <gustavo_> I meant, after the file
[01:59] <gustavo_> ok
[02:01] <gustavo_> klaxa: http://pastebin.geany.org/9uRTC/
[02:02] <klaxa> before the -vf
[02:03] <gustavo_> very strange
[02:03] <gustavo_> http://pastebin.geany.org/yDqqv/
[02:03] <gustavo_> I've got the same output
[02:06] <klaxa> ffmpeg -i VID-20150109-WA0027.mp4 -c:a copy -vf transpose=1 out.mp4
[02:07] <gustavo_> wow
[02:07] <gustavo_> worked !
[02:07] <gustavo_> haha
[02:07] <gustavo_> thanks man
[02:09] <llogan> compstomp: you can contact the volunteers who supply the binaries to see if they will provide you their script or instrucitons or whatever. the project itself does not provide any builds.
[02:10] <llogan> and their configure line will provide some insight
[02:28] <compstomp> How might I be able to contact them? And just running a quick otool -L  (same as ldd on mac) on their binary I downloaded, I see it is in fact linked to several dylibs (although not the codecs).  Is this strictly static? Or is the fact that the codecs are baked in enough to label it as static? Thanks!
[04:18] <AlexRussia> hey
[04:18] <AlexRussia> does it possible make from mp4 video hls(http live streaming) ?
[08:07] <Guest69557> c_14: so I finaly ran k9copy from terminal, to see what was actualy going on, and , it is remastering every vob, because the only difference between the original and the compressed version is the size all files have the same name, and same number of files
[09:30] <jongq> how can i download a youtube video with FFmpeg ? tx
[09:32] <ramiro> jongq: http://rg3.github.io/youtube-dl/
[09:35] <jongq> i mean by command line
[09:36] <relaxed> youtube-dl is a command line
[09:36] <relaxed> *program
[09:36] <jongq> ok tx
[09:43] <jongq> when playing a video with ffplay how to not show the consol ? tx
[11:01] <anddam> hello
[11:08] <anddam> I'm trying to cut an m4a with 00:06:28.03 duration (source ffprobe) using     ffmpeg -ss 00:47:00 -i orig.m4a -acodec copy cut.m4a
[11:10] <anddam> I got an empty file (with correct container tho'), what am I doing wrong?
[11:54] <anddam> even using    -ffmpeg -ss 00:47:00 -i orig.m4a  -to 00:06:28 cut.m4a   produces an empty file
[11:54] <anddam> actually the file is 0kB if not using  -acodec copy   and 1kB if using it
[11:55] <anddam> I'd like to tell ffmpeg "read starting from -ss and go on till the end of file"
[11:57] <anddam> ok, I see the point
[11:57] <anddam> starting at 00:47:00 on a 00:06:28 long file produces zero output
[12:03] <leandrosansilva> Hello to all. I'm trying to concatenate some jpeg images which are accessible from a http server, with the format http://server/frame.jpg?frame={n}, where n is a number, and generate a video from them. I'm using concat format (ffmpeg -f concat -i -) and receiving the input list from stdin. the problem is that the ffmpeg http parser is able to guess each input format when it finishes with .jpg, but not when there's some GET params. For a single image
[12:03] <leandrosansilva> , I can use -f mjpeg, but I can't to it when using multiple files, because it conflicts with -f concat. My ffmpeg version is 2.2 (compiled from source). Is that a way to force ffmpeg to use mjpeg as the format for each input when using concat?
[12:09] <relaxed> leandrosansilva: wget them first
[12:10] <relaxed> or pipe them through it
[12:11] <leandrosansilva> relaxed, the problem here is that as there's a lot of images, I was looking for a way to not need to download them first, but creat the video "on the fly"
[12:11] <leandrosansilva> how can I use the second alternative, the piped one?
[12:12] <leandrosansilva> the problem, IMO, is the http parser, which is not able to guess the input format when it does not finish with .jpeg
[12:12] <leandrosansilva> or .jpg
[12:12] <leandrosansilva> even when the Content-Type is image/jpeg
[12:13] <anddam> bye
[12:14] <relaxed> you're going to download them images whether you use ffmpeg or wget
[12:14] <leandrosansilva> yes, but I guess that with ffmpeg I do not need to download all them first in order to encode the video, right?
[12:15] <leandrosansilva> I suppose each frame is download when it's necessary, and then discarded after it's usage
[12:16] <relaxed> wget can grab them in order and pipe the output to ffmpeg
[12:17] <leandrosansilva> oh, now I see it...
[12:18] <relaxed> something like, wget 'http://server/frame.jpg?frame='{001..300} -O - 2>/dev/null|ffmpeg -f image2pipe -i - ...
[12:18] <leandrosansilva> ok, I'll try that, thank you :-)
[12:19] <relaxed> where is {001..300} the full range from 001 through 300 (if you're using bash)
[12:26] <leandrosansilva> relaxed, I got the same "non ormat guessing" behavior in ffmpeg
[12:26] <leandrosansilva> error "pipe:: could not find codec parameters"
[12:33] <relaxed> leandrosansilva: are you using the right name for the framenumber?
[12:34] <leandrosansilva> Using -vcodec mjpeg on input it seems to work
[12:34] <relaxed> ok, my test didn't require it
[12:35] <relaxed> which version are you using?
[12:35] <leandrosansilva> 2.2
[12:35] <leandrosansilva> yes, it is old :-)
[12:38] <leandrosansilva> relaxed, thanks for the help :-)
[12:42] <relaxed> you're welcome
[14:21] <pron> ell except tatoos @ vaginas
[15:54] <norinder99> Hi, while re-encoding a video, does the audio bitrate mean the total bitrate of all audio channels or just one audio channel?
[15:56] <klaxa|work> norinder99: it should be the bitrate of (at least) one audio stream
[15:56] <klaxa|work> you could define different bitrates for different streams
[15:56] <klaxa|work> a stream can consist of multiple channels
[15:57] <norinder99> klaxa|work, ok, thanks!
[15:58] <dAnjou> hi, i'd like to create a 10s video with no audio and a black screen. i have this command which creates a black screen video with the length of an audio input file: ffmpeg -loop 1 -i black.png -i audio_src.flv -acodec copy -shortest dst.flv
[15:59] <dAnjou> i don't know what to look for in the docs
[16:00] <dAnjou> got it
[16:00] <dAnjou> ffmpeg -loop 1 -i black.png -t 10 dst.flv
[16:01] Action: dAnjou googled inefficiently
[16:50] <mhernandez1> Hey all, working on a problem where i am trying to decode an unknown pcm stream. I think i need to iterate over all the pcm formats, but every time i get this error
[16:50] <mhernandez1> https://gist.github.com/anonymous/367700422f1fc433c21e
[16:51] <mhernandez1> audacity, seems to be able to at least import the raw data, but it comes out garbled, and the ui is just too difficult to iterate over all possible combinations of pcm/bitdepth/(be|le)
[16:52] <saste> mhernandez1, probably corrupted data
[16:52] <saste> you will get meaningful data only by chance
[16:53] <saste> garbage in -> garbage out
[16:53] <mhernandez1> Thats interesting because its actually a programming challenge for an interview, and they provide tons of packets with checksums, and I had to filter out the bad packets, and join all the good packets together to create this file
[16:54] <mhernandez1> i think the problem is not knowing 1. the bit depth, 2. the endianness, 3. not being proficient enough with the ffmpeg api
[16:54] <klaxa|work> you are not setting the input format for decoding, but the output format
[16:54] <klaxa|work> specify -c:a pcm_u32be or whatever before -i
[16:54] <mhernandez1> oh, thanks
[16:54] <klaxa|work> also maybe add -f rawaudio before the codec
[16:55] <klaxa|work> someone else also came around with a similar problem, have you looked at the file in a hex editor?
[16:55] <mhernandez1> klaxa|work: so ffmpeg -f rawaudio -c:a pcm_u32be -i file f.wav
[16:55] <klaxa|work> yes, for example
[16:55] <mhernandez1> i have not examined it in a hex editor, but i used nodejs to build it, and examined the hex in their
[17:08] <mhernandez1> holy shit i freakin got it
[17:09] <mhernandez1> at pcm_u8 the recording is finally slow enough to distiguish linguistic patterns, well i dont know who else is in this channel so i wont give anything away
[17:26] <bencc1> can I fill a drawbox rectangle?
[17:26] <bencc1> or is there another way to draw a white filled rectangle on a video?
[17:27] <bencc1> maybe t=max. trying
[17:30] <bencc1> t=max gives me an error
[17:39] <bencc1> just using a large t=1000 works
[17:39] <bencc1> not sure why t=max doesn't work as in the example in the docs
[19:01] <netw1z> is there a simple command like to stream a mp4 from shell to youtube live?
[19:09] <phr3ak> hello if u have a little time could you please help me to save an rtmp stream with rtmp dump? I extracted an url from rtmp stream but I cannot save it with rtmpdump. Is it because a protection?
[19:09] <phr3ak> http://www.kubinyitamas.hu/
[19:10] <phr3ak> I tried with this: ./rtmpdump -r "rtmp://91.82.85.71:1935/livesource571/_definst_/almoskiralytv.sdp"
[19:20] <phr3ak> I solve it
[20:17] <Jan-> is it normal for xavc-i to prores transcodes of 3840x2160 material to go at 2 fps?
[20:18] <klaxa> maybe? that really depends on settings and hardware
[20:18] <Jan-> hardware is a 3.06ghz quad core i7
[20:21] <Jan-> D:\FS7\Clip>c:\ffmpeg -i 489_0022.MXF -c:a copy -c:v prores_ks -profile:v 3 -pro
[20:21] <Jan-> besize 50000000 test.mov
[20:42] <Sirisian|Work> Why doesn't this work: 'ffmpeg -i input.mp4 -vcodec libvpx-vp9 output.webm'? I get the following error: "Failed to initialize encoder: Memory allocation error" this is like the latest ffmpeg.
[20:43] <klaxa> sounds like you don't have enough memory
[20:44] <Sirisian|Work> Indeed. This is a low end server. hmm.
[20:45] <klaxa> you could add swap space
[20:45] <klaxa> if it's a low end server the encode will take very long in any case
[20:48] <llogan> if it's a high end server the VP9 encode will take very long in  any case
[21:03] <voip_> hello guys
[21:09] <voip_> I have problem when i taking live stream and sending to oter server. After 30 min as result I have audio shift
[21:09] <voip_> ffmpeg -re -i http://source/index.m3u8 -vcodec copy -preset fast -acodec aac -strict experimental -ar 44100 -ac 2 -b:a 96k -vcodec libx264  -b:v 1300k -maxrate 1500k -bufsize 3000k -g 60 -f flv rtmp://output/live/stream
[21:10] <llogan> -re should not be used with live input streams
[21:11] <voip_> llogan, thank you
[21:11] <llogan> also you forgot to include the complete console output, as per tradition
[21:12] <llogan> (the tradition being your forgetting to do so)
[21:12] <Sirisian|Work> klaxa, swap space worked. Good idea. :)
[21:12] <voip_> also one question , actualy i dont want process video, i need just copy. Id it correct if i using in my command  -bufsize 3000k -g 60, or i shuld delete it ?
[21:13] <llogan> "-vcodec libx264 -b:v 1300k -maxrate 1500k -bufsize 3000k -g 60 -preset fast" will be ignored
[21:13] <llogan> if you stream copy
[21:13] <voip_> thank you !
[21:13] <llogan> but the input may not be suitable for your network and/or clients
[21:13] <klaxa> Sirisian|Work: out of curiosity, is it anywhere near acceptable encoding speeds?
[21:14] <Sirisian|Work> klaxa, a frame every 5 seconds!
[21:14] <Sirisian|Work> shits cash
[21:14] <Sirisian|Work> >_>
[21:14] <llogan> see you in 2018
[21:14] <Sirisian|Work> I'm on frame 64. :(
[21:14] <llogan> the good news is that when you're done VP12 will be available
[21:15] <c_14> And it'll be just as slow?
[21:15] <llogan> of course!
[21:16] <Sirisian|Work> Customers are going complain. We have people that want better 4K support. I'll need to add decimals to the progress bar :|
[21:16] <c_14> Just as slow on vastly improved hardware that is, on identical hardware it'll be even slower.
[21:16] <llogan> google doesn't seem to give a shit about end users encoding with it so it will always be slow
[21:21] <JEEB> they surprisingly seem to have added some kind of threading to the vp9 encoder lately
[21:21] <JEEB> haven't tested if it works, though
[21:26] Action: llogan flabbergasted
[21:55] <IsoDesu> Hi, I'm trying to hardsub an styled ASS subtitle to a MKV video, the subtitle was extracted from a MKV anime show with fonts attached. The console outputs many "fontconfig: cannot find font X, falling back to Y" while working. I gotta discover how to make ffmpeg use those fonts attached to the original video for the output MKV to show exactly like the original video.
[21:58] <klaxa> you can skip the extracting part of the subtitles as you can just specify an mkv file for the subtitles/ass videofilter
[21:58] <klaxa> regarding the fonts, i don't think there is a proper solution that tells ffmpeg/fontconfig to look through the fonts that are attached to the matroska file
[21:59] <klaxa> IsoDesu: ^
[22:01] <IsoDesu> if I try "ffmpeg -i video.mkv -vf "ass=subtitle.mkv" out.mkv" the output is [Parsed_ass_0 @ 0x2b92ee0] Could not create a libass track when reading file 'video.mkv'. [AVFilterGraph @ 0x2b87c20] Error initializing filter 'ass' with args 'video.mkv'
[22:01] <IsoDesu> Error opening filters!
[22:01] <IsoDesu> "ass=video.mkv"***
[22:01] <IsoDesu> I'm using ffmpeg 2.5.3 official static build.
[22:02] <klaxa> huh?
[22:02] <klaxa> can you pastebin your complete command with output?
[22:02] <IsoDesu> I followed the wiki page
[22:02] <klaxa> use pastebin.com or something similar
[22:04] <IsoDesu> just a moment
[22:07] <IsoDesu> http://pastebin.com/9KFReGyi
[22:07] <IsoDesu> Is what happens with your suggestion
[22:09] <klaxa> hmm that's really weird
[22:10] <klaxa> can you run the same command with -loglevel debug?
[22:10] <IsoDesu> where do I put it in the command?
[22:10] <klaxa> you can put it before everything, so ffmpeg -loglevel debug -i [...]
[22:12] <klaxa> ah
[22:13] <klaxa> can you replace ass with subtitles?
[22:13] <klaxa> in the videofilter i mean
[22:16] <klaxa> hmm i think -loglevel debug was not what i actually meant
[22:16] <IsoDesu> http://pastebin.com/jdFCzyfq
[22:17] <IsoDesu> Would I get the desired result with subtitles filter instead of ass filter? I'll try that
[22:17] <IsoDesu> I followed this wiki: http://trac.ffmpeg.org/wiki/HowToBurnSubtitlesIntoVideo
[22:19] <klaxa> well it also says so in that wiki entry
[22:20] <IsoDesu> subtitles filter didn't solve it, still can't find the fonts
[22:20] <klaxa> oh well regarding the fonts, <klaxa> regarding the fonts, i don't think there is a proper solution that tells ffmpeg/fontconfig to look through the fonts that are attached to the matroska file
[22:21] <JEEB> you can use mpv to re-encode such files
[22:21] <JEEB> with the subtitles properly rendered
[22:21] <klaxa> oh right
[22:21] <IsoDesu> fontconfig is looking elsewhere for the fonts, how do I put them there?
[22:21] <JEEB> IsoDesu, I recommend you just use mpv for those files; simpler, no need for extracting the fonts or the subs :P
[22:22] <JEEB> http://mpv.io/
[22:22] <IsoDesu> mpv is new to me, i'll have a look.
[22:22] <JEEB> it's the least retarded mplayer fork
[22:23] <JEEB> it has a rewritten encoding component in it (compared to mencoder, which has an ancient thing) that utilizes ffmpeg's libraries for it
[22:23] <JEEB> the main benefit is that mpv can demux segment linked and virtual timelined matroska files
[22:24] <IsoDesu> hm
[22:24] <JEEB> and that it can load fonts from the file + render the subtitles
[22:24] <IsoDesu> is it better than SMPlayer?
[22:24] <JEEB> smplayer can now use mpv as a backend, IIRC. although the support is pretty retarded
[22:24] <JEEB> smplayer is a frontend
[22:24] <JEEB> mplayer/mplayer2/mpv are the things it supports right now
[22:25] <IsoDesu> hm
[22:25] <IsoDesu> so you think mpv's engine is better than mplayer's?
[22:26] <JEEB> at this point I see no reason to use mplayer :P
[22:26] <JEEB> if you have a relatively sane opengl driver, mpv also has the best opengl video renderer in OSS
[22:27] <IsoDesu> Oh you suggested me to play this video with mpv, instead of having to burn or merge the subtitle to the video. But getting to know how to make it hardsubbed is my goal.
[22:28] <JEEB> no
[22:28] <JEEB> I suggested you mpv because of the encoding component
[22:28] <JEEB> <JEEB> you can use mpv to re-encode such files
[22:28] <JEEB> <JEEB> with the subtitles properly rendered
[22:28] <JEEB> mplayer had mencoder, which was separate and for the most part completely awful
[22:29] <JEEB> so it was removed, and a new encoding feature added
[22:29] <IsoDesu> So I can use mpv as encoder to hardsub subtitles and output MKVs?
[22:29] <JEEB> yes
[22:29] <IsoDesu> My bad I thought it was just a player.
[22:29] <JEEB> I wouldn't say "you can use mpv to re-encode such files"
[22:29] <JEEB> if I didn't mean that
[22:29] <JEEB> also "it has a rewritten encoding component in it"
[22:30] <JEEB> anyways, try grabbing as new mpv as possible and looking through the encoding documentation
[22:30] <JEEB> there should be plenty of documentation, and there is an IRC channel for it, #mpv-player
[22:31] <IsoDesu> So to finish, do you recommend me to use mpv as backend in SMPlayer instead of mplayer?
[22:33] <IsoDesu> Thanks guys.
[22:42] <iive> IsoDesu: ffmpeg subtitle filter should work just fine. however if you are under windows, you are going to miss fontconfig config file...
[22:42] <IsoDesu> I'm under Fedora 21
[22:43] <IsoDesu> I tested with my distribution's build and the official static build, the same result.
[22:49] <iive> IsoDesu: you extract the .ass file, try to give it as parameter to the ass filter.
[22:50] <IsoDesu> That result's the fontconfig: cannot find font X, falling back to Y
[22:51] <c_14> IsoDesu: can you pastebin the commandline and output when you use the subtitles filter?
[22:52] <IsoDesu> Which argument I use, the video.mkv or the extracted subtitle.ass?
[22:53] <c_14> the mkv
[22:54] <IsoDesu> just a moment
[22:55] <IsoDesu> The conversion takes time, i'll take a previously made one.
[22:59] <IsoDesu> All the finished jobs I have were made using ass filter, would you wait 20 minutes to start a new one?
[23:03] <IsoDesu> I gotta go now, if mpv don't solve it maybe I back later
[23:03] <IsoDesu> Thanks
[23:22] <AlexRussia> what mean error: [aac @ 0x7ffb052d8680] The encoder 'aac' is experimental but experimental codecs are not enabled, add '-strict -2' if you want to use it.
[23:22] <AlexRussia> ?
[23:23] <AlexRussia> I did add this things, but it's still tell me it and crash
[23:23] <AlexRussia> or not even crash, just finish
[23:25] <AlexRussia> ok, I guess, i've got it
[00:00] --- Thu Jan 15 2015


More information about the Ffmpeg-devel-irc mailing list