[Ffmpeg-devel-irc] ffmpeg.log.20160203

burek burek021 at gmail.com
Thu Feb 4 02:05:01 CET 2016


[00:10:35 CET] <jkqxz> shadow42085:  You are in the right channel.  Linux native packages are not relevant to a cross compile.  Have you looked at <https://trac.ffmpeg.org/wiki/CompilationGuide/CrossCompilingForWindows>?
[00:15:02 CET] <shadow42085> jkqxz i have been there and they dont tell me which packages only scripts that can build the environment needed but i want to incorporate the Cross-Compilers into my system's native build path
[00:18:56 CET] <jkqxz> Reading one of those scripts will tell you what they build (there is a lot).
[00:20:39 CET] <jkqxz> What are you actually wanting at the end of it, when you say it should be incorporated into your system's native build path?
[00:22:24 CET] <shadow42085> i dont want extra directories i have to call like what it say's in the ubuntu/debian compilation guide
[00:29:06 CET] <shadow42085> I have already installed mingw32 libs
[00:29:11 CET] <jkqxz> Can you be a bit clearer about what extra directories you are referring to?
[00:29:44 CET] <shadow42085> ffmpeg_bulid bins
[00:30:39 CET] <shadow42085> where as they can be in my systems native /bins diectory or other native directories
[00:31:13 CET] <shadow42085> hope I am  not to confusing i am trying to be direct as possible
[00:33:21 CET] <jkqxz> Nothing which has been cross-built should be in your system directories, because they are not usable there.  Cross-compiling, you will make a separate sysroot containing all of the cross-built libraries and headers which go into ffmpeg, then copy the final pieces to Windows to actually use.
[00:35:56 CET] <jkqxz> Some of the build tools (the compiler, etc.) will be usable on the host, and they could go in system directories with suitable naming to distinguish them from native tools (hence i686-w64-mingw32-gcc, etc.).
[00:36:52 CET] <shadow42085> ok
[00:37:57 CET] <shadow42085> so i just need to cross-compile the libs in the guide?
[00:39:40 CET] <jkqxz> Pretty much, yes.  (And the scripts to do it for you exist because this takes a long time and is highly finicky.)
[00:40:55 CET] <shadow42085> ok
[00:44:06 CET] <shadow42085> being finicky is normal since we are going from a native linux system to windows and windows itself is finicky as well
[00:53:09 CET] <shadow42085> well thanks anyways
[01:43:48 CET] <aolko> hi guys
[01:44:22 CET] <aolko> any ideas how to _only overlay image_ on a hls live stream and return it back?
[04:30:53 CET] <barry_> hello. can anybody out there help me with compiling ffmpeg on a RPi 2....I've tried every resource on the Net and I'm stuck with a few problems
[04:31:38 CET] <waressearcher2> barry_: hallo und herzlich willkommen
[04:31:47 CET] <barry_> gates
[04:32:45 CET] <barry_> and danke
[04:37:18 CET] <klaxa> barry_: what problems did you run into?
[04:38:59 CET] <barry_> hi klaxa, I can compile using the instructions for debian and then I send it off to youtube...but the straem has a glitch about every 5 seconds. it starts out strong and then degrades
[04:39:16 CET] <klaxa> what
[04:39:30 CET] <klaxa> so you got it to compile?
[04:39:41 CET] <barry_> yes I did
[04:40:12 CET] <klaxa> and by "send it off to youtube" you mean stream some video?
[04:40:19 CET] <barry_> but I don't know if I am using the correct flags
[04:40:22 CET] <klaxa> pastebin your command with output on pastebin or something
[04:40:23 CET] <barry_> yep
[04:40:49 CET] <barry_> hang on an I'll get it
[04:42:08 CET] <barry_> raspivid -o - -t 0 -fps 30 -b 5000000 | ffmpeg -re -ar 44100 -ac 2 -acodec pcm_s16le -f s16le -ac 2 -i /dev/zero -f h264 -i - -vcodec copy -acodec aac -ab 128k -g 50 -strict experimental -f flv rtmp://a.rtmp.youtube.com/live2/##mykey##
[04:42:39 CET] <barry_> this is the input side
[04:43:58 CET] <klaxa> are those the settings recommended by youtube?
[04:44:08 CET] <barry_> oh darn it...I deleted my output
[04:44:09 CET] <klaxa> ah wait
[04:44:26 CET] <klaxa> it's definitely not an issue with ffmpeg
[04:44:31 CET] <klaxa> since it's not re-encoding the video
[04:44:34 CET] <klaxa> i think?
[04:44:48 CET] <klaxa> did you try to play whatever raspivid outputs locally? over the network or something?
[04:45:05 CET] <barry_> I think that is true as I have read tat it is just copying it?
[04:45:09 CET] <klaxa> like: raspivid [...] | nc -l 1234
[04:45:27 CET] <klaxa> and on the client: nc <ip of raspi> 1234 | mpv -
[04:45:30 CET] <klaxa> or vlc or whatever
[04:45:59 CET] <klaxa> yes, with the settings you provided, ffmpeg will copy the video stream and add a silent audio stream
[04:46:13 CET] <barry_> I'm just getting the hang of it all, but I did follow some instructions for setting up a local webserver and the delay was about 5 seconds
[04:46:23 CET] <klaxa> although i'm not sure you can do -g 50
[04:46:41 CET] <klaxa> -g 50 means "make it an I-Frame every 50 frames"
[04:46:48 CET] <klaxa> which you cannot influence if you are not encoding
[04:48:42 CET] <barry_> there's a guy who posted the input and he used a docker container for the binaries....and his youtube stream is incredible, but I haven't been able to reproduce his work
[04:49:21 CET] <barry_> I'll have to stop by and talk to you folks tomorrow night after I get all of my ducks in a row
[04:49:52 CET] <klaxa> ok
[04:49:58 CET] <barry_> can I post a URL here for you to look at?
[04:50:43 CET] <barry_> etiquette wise?
[07:11:35 CET] <Marcurling> Can't ffmpeg output.mkv ?
[07:12:49 CET] <Marcurling> got Unable to find a suitable output format for 'pipe:' pipe:: Ivalid argument trying to.
[07:14:20 CET] <relaxed> that's not a complete ffmpeg command
[07:15:30 CET] <Marcurling> ffmpeg -vcodec copy -acodec copy -i input.mp4 output.mkv
[07:16:32 CET] <relaxed> ffmpeg -i input.mp4 -vcodec copy -acodec copy output.mkv
[07:16:54 CET] <Marcurling> Gives [NULL @ 000000000261f680] Unable to find a suitable output format for 'pipe:'
[07:16:54 CET] <Marcurling> pipe:: Invalid argument
[07:25:41 CET] <Marcurling> >ffmpeg -vcodec copy - acodec copy -i "Doctor.Who.(2005).9x07.Vérité.Ou.Conséquences.(Part.1).FR.LD.
[07:25:41 CET] <Marcurling> WEBRip.x264-LiBERTY.[tvu.org.ru].mp4" output.mkv
[07:25:41 CET] <Marcurling> ffmpeg version N-78263-g5fc310f Copyright (c) 2000-2016 the FFmpeg developers
[07:25:41 CET] <Marcurling>   built with gcc 5.3.0 (GCC)
[07:25:41 CET] <Marcurling>   configuration: --enable-gpl --enable-version3 --disable-w32threads --enable-avisynth --enable-bzli
[07:25:42 CET] <Marcurling> b --enable-fontconfig --enable-frei0r --enable-gnutls --enable-iconv --enable-libass --enable-libblu
[07:25:44 CET] <Marcurling> ray --enable-libbs2b --enable-libcaca --enable-libdcadec --enable-libfreetype --enable-libgme --enab
[07:25:46 CET] <Marcurling> le-libgsm --enable-libilbc --enable-libmodplug --enable-libmp3lame --enable-libopencore-amrnb --enab
[07:25:48 CET] <Marcurling> le-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-librtmp --enable-libschroedinger
[07:25:50 CET] <Marcurling>  --enable-libsoxr --enable-libspeex --enable-libtheora --enable-libtwolame --enable-libvidstab --ena
[07:25:52 CET] <Marcurling> ble-libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-
[07:25:54 CET] <Marcurling> libx264 --enable-libx265 --enable-libxavs --enable-libxvid --enable-libzimg --enable-lzma --enable-d
[07:25:56 CET] <Marcurling> ecklink --enable-zlib
[07:25:58 CET] <Marcurling>   libavutil      55. 17.100 / 55. 17.100
[07:26:00 CET] <Marcurling>   libavcodec     57. 24.101 / 57. 24.101
[07:26:02 CET] <Marcurling>   libavformat    57. 24.100 / 57. 24.100
[07:26:04 CET] <Marcurling>   libavdevice    57.  0.101 / 57.  0.101
[07:26:06 CET] <Marcurling>   libavfilter     6. 28.100 /  6. 28.100
[07:26:10 CET] <Marcurling>   libswscale      4.  0.100 /  4.  0.100
[07:26:12 CET] <Marcurling>   libswresample   2.  0.101 /  2.  0.101
[07:26:14 CET] <Marcurling>   libpostproc    54.  0.100 / 54.  0.100
[07:26:16 CET] <Marcurling> Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'Doctor.Who.(2005).9x07.V®rit®.Ou.Cons®quences.(Part.1).F
[07:26:19 CET] <Marcurling> R.LD.WEBRip.x264-LiBERTY.[tvu.org.ru].mp4':
[07:26:21 CET] <Marcurling>   Metadata:
[07:26:23 CET] <Marcurling>     major_brand     : isom
[07:26:25 CET] <Marcurling>     minor_version   : 1
[07:26:27 CET] <Marcurling>     compatible_brands: isom
[07:26:29 CET] <Marcurling>     creation_time   : 2016-01-23 17:33:45
[07:26:31 CET] <Marcurling>   Duration: 00:45:27.68, start: 0.000000, bitrate: 1131 kb/s
[07:26:33 CET] <Marcurling>     Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709/unknown/unknown), 72
[07:26:35 CET] <Marcurling> 0x404 [SAR 1:1 DAR 180:101], 969 kb/s, 25 fps, 25 tbr, 100 tbn, 50 tbc (default)
[07:26:37 CET] <Marcurling>     Metadata:
[07:26:41 CET] <Marcurling>       creation_time   : 2016-01-23 08:03:31
[07:26:43 CET] <Marcurling>       handler_name    : Doctor.Who.2005.S09E07.FRENCH.LD.WEB-DL.x264-LiBERTY
[07:26:45 CET] <Marcurling>     Stream #0:1(fre): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 157 kb/s (default
[07:26:47 CET] <Marcurling> )
[07:26:49 CET] <Marcurling>     Metadata:
[07:26:51 CET] <Marcurling>       creation_time   : 2016-01-23 17:33:48
[07:26:53 CET] <Marcurling>       handler_name    : GPAC ISO Audio Handler
[07:26:55 CET] <Marcurling> [NULL @ 000000000241f6a0] Unable to find a suitable output format for 'pipe:'
[07:26:57 CET] <Marcurling> pipe:: Invalid argument
[07:26:59 CET] <Marcurling> ooops
[07:27:56 CET] <relaxed> oops indeed, don't do that again
[07:28:28 CET] <Marcurling> never used pastedbin, sorry
[07:29:29 CET] <relaxed> it's a website where you paste your output there and then give us the URL
[07:31:09 CET] <Marcurling> I did but... which function should I use then ?
[07:35:14 CET] <Marcurling> Should be http://pastebin.com/7eK4VvgB
[07:36:16 CET] <relaxed> ffmpeg -i input  <other options here>  output
[07:37:28 CET] <relaxed> hmm, if that doesn't work rename the input to Doctor.Who.9x07.mp4 and try again
[07:39:19 CET] <Marcurling> ffmpeg -i input  <other options here>  output is working fine, thanks !
[09:36:22 CET] <Artix> Hello, anyone have a command line for SD PAL v210 to d10 please ? i have tried  http://pastebin.com/Hst5pbcx but i'm not happy with it.
[10:32:38 CET] <tfilk> I want to dump only audio track from video: "ffmpeg -i input.avi -ac 2 -ar 44100 -ab 192k -y -f wav 2.wav", but there are 3 tracks, how to choose the track ? what option ?
[10:33:26 CET] <tfilk> is it "-map 0:a:0" ?
[10:33:38 CET] <relaxed> tfilk: that's the first audio stream
[10:34:13 CET] <Mavrik> Yep, "0" - 1st input file, "a" - audio, "0" - 1st stream
[10:36:25 CET] <tfilk> so I want to dump all 3 audio streams in different files, so should I go one by one "-map 0:s:0, -map 0:s:1, -map 0:s:2" or there is one command that do all stuff at once /
[10:36:28 CET] <tfilk> ?
[10:36:44 CET] <relaxed> https://trac.ffmpeg.org/wiki/Map
[10:36:59 CET] <relaxed> :s: is subs
[10:39:45 CET] <tfilk> also ffprobe shows streams as: "Stream #0:1(en): Audio, Stream #0:2(en): Audio, Stream #0:3(en): Audio, ", so it starts from "1" so there is no "-map 0:a:0" stream ? so all streams will be numbered as "-map 0:a:1, -map 0:a:2, -map 0:a:3" or "-map 0:a:0, -map 0:a:1, -map 0:a:2" ?
[10:39:46 CET] <relaxed> you can use one instance or three
[10:40:45 CET] <relaxed> no, the count starts at 0. You can use -map 0:1 if that's easier for you
[10:41:25 CET] <tfilk> thanks
[10:41:42 CET] <relaxed> so "-map 0:a:0" is the same as "-map 0:1" in your example
[17:02:55 CET] <MINIMAN10000> Out of curiosity when using h264 is it possible to improve black gradients at a constant bitrate?
[17:49:35 CET] <kbarry> I'm experiencing an audible phenomenon when listening to an HLS stream i'm generating (HLS generated by Wowza)
[17:49:42 CET] <kbarry> with loglevel set to 56, I am seeing an error in the logs (Header missing)
[17:50:19 CET] <kbarry> followed shortly by a clicki/micro-silence, or a short "vrrrp" (like a tiny amount of sped-up audio)
[17:51:40 CET] <kbarry> coincidentally, when I use VLC (With verbose logging on), I see a error at the same time, (different error)
[19:19:02 CET] <RazWelles> Hey, I'm trying to point ffmpeg2theora to a custom build of ffmpeg?
[19:19:23 CET] <RazWelles> I set the pkg_config_path variable to the ffmpeg git directory but no luck
[19:20:45 CET] <J_Darnley> "no luck" is not an error message
[19:20:57 CET] <RazWelles> sec
[19:22:30 CET] <RazWelles> http://hastebin.com/bicinojoco.vbs
[19:23:03 CET] <RazWelles> So I have ffmpeg compiled in another directory which I've set in PKG_CONFIG_PATH
[19:23:26 CET] <J_Darnley> That must be the worst paste site I have ever seen
[19:23:56 CET] <J_Darnley> Is there not a better, more detailed log?
[19:25:20 CET] <RazWelles> Not really, it just lists the libraries individually, and I can paste that- where it says (no) I've actually had it say yes last night by manually adding each ffmpeg subdirectory to the PKG_CONFIG_LIBPATH but it still choked and gave the same PKG_CONFIG_PATH message
[19:25:50 CET] <J_Darnley> It doesn't make a config.log file?  One that actually says what it is doing?
[19:26:15 CET] <RazWelles> Where would I look for that? I can do a find for a log sec
[19:26:29 CET] <J_Darnley> In the directory you ran configure.
[19:26:52 CET] <RazWelles> Found it, not much better, I"ll paste
[19:27:23 CET] <RazWelles> http://hastebin.com/xulesaharo.coffee
[19:27:31 CET] <RazWelles> The libraries its showing as not found are part of ffmpeg
[19:27:32 CET] <relaxed> what can ffmpeg2theora do that ffmpeg with libvorbis and libtheora can not?
[19:28:11 CET] <RazWelles> I just found it easier to edit, I suppose I could go through and edit ffmpeg instead but its a much bigger project to wade through
[19:28:13 CET] <J_Darnley> Useless log is useless!
[19:28:43 CET] <RazWelles> I need to pass more data from commandline into theora
[19:29:22 CET] <J_Darnley> I guess that the software is too old and that causes whatever test it is doing to fail when you try a new ffmpeg.
[19:29:57 CET] <RazWelles> Let me check the git log on ffmpeg2theora
[19:30:17 CET] <RazWelles> Strange, last edit was Jan 10 of this year
[19:30:30 CET] <RazWelles> But it was only edited once last year..
[19:30:50 CET] <J_Darnley> Can you tell me where I can find it?
[19:31:13 CET] <RazWelles> J_Darnley: http://v2v.cc/~j/ffmpeg2theora/
[19:31:15 CET] <RazWelles> There you go
[19:31:34 CET] <RazWelles> I'm basically trying to point the builds to custom directories so I'm not running make install or scons install
[19:32:13 CET] <J_Darnley> "2016-01-01 - update to current FFmpeg api"
[19:35:34 CET] <RazWelles> I guess that means it should be working?
[20:52:40 CET] <cousin_luigi> Greetings.
[20:57:07 CET] <cousin_luigi> What's new with the new AAC encoder in trunk, licence-wise?
[20:57:25 CET] <c_14> nothing?
[20:59:38 CET] <cousin_luigi> "In both cases, you will enjoy an audible quality improvement and as well as fewer licensing headaches. "
[20:59:52 CET] <cousin_luigi> https://ffmpeg.org/index.html#news
[21:00:13 CET] <c_14> The licence of the builtin AAC encoder never changed.
[21:00:20 CET] <c_14> It's just better now so you don't have to use the other ones.
[21:00:22 CET] <furq> it means you don't have to use https://android.googlesource.com/platform/external/aac/+/master/NOTICE
[21:00:26 CET] <c_14> (And 2 of the old ones got removed)
[21:00:29 CET] <JEEB> hh
[21:00:35 CET] <JEEB> the license didn\t change
[21:00:44 CET] <JEEB> it was always LGPL
[21:00:51 CET] <cousin_luigi> ok
[21:01:07 CET] <JEEB> but yes, the internal aac encoder got a whole lot better in master's HEAD :P
[21:01:17 CET] <furq> is it any good at vbr yet
[21:01:19 CET] <cousin_luigi> so it can be used liberally whenever there are no patent concerns, right?
[21:01:45 CET] <furq> also have the changes made it into 2.8 releases yet
[21:01:49 CET] <JEEB> follow the software license (LGPL methinks, check the header to be sure) and other
[21:01:53 CET] <JEEB> furq: they will not
[21:01:58 CET] <JEEB> I mean, it's new features :P
[21:02:03 CET] <JEEB> not just bugfixes
[21:02:10 CET] <furq> are they held for 2.9
[21:02:13 CET] <cousin_luigi> furq: I don't think so
[21:02:26 CET] <JEEB> they will come up in any next release, whatever the version for that will be
[21:02:26 CET] <cousin_luigi> 2.9 also for dca
[21:02:42 CET] <furq> oh
[21:02:43 CET] <JEEB> also fdk-aac is still the thing you want to use for HE-AAC
[21:02:58 CET] <JEEB> what the updated lavc encoder excels at is LC
[21:03:03 CET] <cousin_luigi> Speaking of dropping libdca, is there a dependency between it and mdts?
[21:03:06 CET] <furq> people have been talking about this since before 2.8.5, but i guess that was specifically a bugfix release
[21:03:07 CET] <cousin_luigi> JEEB: hmm
[21:03:35 CET] <JEEB> furq: yes, but that's a point release off an earlier branch-off :P
[21:04:16 CET] <JEEB> 2.8 was a branch off and 2.8.5 is a patched version of that with additional bugfixes cherry-picked from master
[21:04:27 CET] <furq> oh hey 2.8.6 is out now
[21:04:30 CET] <cousin_luigi> yeah
[21:05:17 CET] <JEEB> the dca decoder was replaced in master with one that is pretty much libdcadec
[21:05:25 CET] <JEEB> by the author of libdcadec
[21:05:32 CET] <cousin_luigi> JEEB: yes, I understand as much. foo86 right?
[21:05:35 CET] <JEEB> yes
[21:06:03 CET] <cousin_luigi> but I saw a reference to that other decoder in the relevant commit and wondered if I had to enable it or something
[21:06:24 CET] <JEEB> what
[21:06:36 CET] <JEEB> it was a replacement so no multiple decoders :P
[21:06:45 CET] <cousin_luigi> ...bitte warten...
[21:06:49 CET] <JEEB> there were some mistakes done regarding keeping multiple things
[21:06:55 CET] <JEEB> but dcadec thankfully isn ot one of those
[21:07:42 CET] <cousin_luigi> https://github.com/FFmpeg/FFmpeg/commit/ae5b2c52501d5009fe712334428138a9b758849b <- see mdct
[21:07:59 CET] <cousin_luigi> is it related or just in the same commit?
[21:08:29 CET] <J_Darnley> "mdct" is a component that is required by the dca decoder.
[21:08:33 CET] <JEEB> ^
[21:08:34 CET] <JEEB> this
[21:08:47 CET] <cousin_luigi> exactly as I suspected
[21:08:54 CET] <cousin_luigi> that was my question btw
[21:09:20 CET] <cousin_luigi> do I have to enable it along with dca?
[21:09:28 CET] <J_Darnley> ?
[21:09:43 CET] <JEEB> pretty sure if you are enabling the decoder it will select that thing it depends on
[21:09:48 CET] <JEEB> since it's an internal thing
[21:09:50 CET] <cousin_luigi> k
[21:09:59 CET] <JEEB> and isn't limited by licensing
[21:10:15 CET] <JEEB> just like those other decoders or encoders depend on features in lavc :P
[21:10:33 CET] <JEEB> J_Darnley: I guess this person is doing --disable-everything --enable-decoders=dca or something
[21:10:42 CET] <JEEB> and he asked if it will enable mdct as well :P
[21:11:00 CET] <cousin_luigi> JEEB: exactly, needed to avoid enabling patent-encumbered stuff
[21:11:02 CET] <cousin_luigi> by mistake
[21:11:06 CET] <J_Darnley> Ah.  If that ever fails then we definitely want to know about it.
[21:11:18 CET] <J_Darnley> That is everything then.
[21:11:21 CET] <drv> i suspect the only non-patent-encumbered build of ffmpeg is --disable-everything
[21:11:37 CET] <J_Darnley> Some patent somewhere will cover something according to some troll.
[21:12:01 CET] <cousin_luigi> drv: yes, but then we want also to build a non-US version
[21:12:22 CET] <cousin_luigi> so I need to enable it explicitly
[21:12:31 CET] <cousin_luigi> by the way, is 2.9 long to come?
[22:01:29 CET] <cortexman> i'm trying to get either 30 or 60 images per second to show in a video: ffmpeg -r .5 -pattern_type glob -i '*.jpg' fps=60 -c:v libx264 output.avi
[22:02:07 CET] <c_14> ffmpeg -framerate 60 -pattern_type glob -i '*.jpg' -vf fps=60 -c:v libx264 out.avi
[22:04:16 CET] <cortexman> yeah, that works
[22:04:31 CET] <c_14> actually
[22:04:37 CET] <c_14> get red of the -vf fps=60
[22:04:43 CET] <c_14> there's no point, the input is already 60fps
[22:08:19 CET] <cortexman> can you help set the output to a percentage (such as 25%) of the input as well?
[22:08:29 CET] <c_14> What?
[22:08:56 CET] <cortexman> i tried to set -vf scale= using division operators but the output video was very blurry
[22:09:05 CET] <cortexman> i want the output video to be like 25% of the present resolution
[22:09:07 CET] <c_14> Ah, you want to scale down?
[22:09:26 CET] <cortexman> yeah
[22:10:01 CET] <c_14> -vf scale=iw*.25:ih*.25 <- doesn't work?
[22:10:58 CET] <cortexman> yep that one works. division didn't work well for some reason
[22:10:59 CET] <cortexman> thanks
[22:20:12 CET] <cortexman> c_14, this video is weighing in a several hundred megs. any ideas for making it reasonable-ish?
[22:20:50 CET] <c_14> change the crf? use a slower preset? set a bitrate?
[22:21:55 CET] <c_14> https://trac.ffmpeg.org/wiki/Encode/H.264
[22:22:52 CET] <cortexman> i set the crf, it helped although there is a bit of aliasing at 18
[00:00:00 CET] --- Thu Feb  4 2016


More information about the Ffmpeg-devel-irc mailing list