[Ffmpeg-devel-irc] ffmpeg.log.20161002

burek burek021 at gmail.com
Mon Oct 3 03:05:01 EEST 2016


[00:00:49 CEST] <iive> but they cover different things.
[00:01:27 CEST] <iive> one covers expression, the other covers ideas.
[00:25:10 CEST] <iive> hum, can't find good article on when mpeg2 patents expire. mpeg2 has been published at 1996, so 20 years have passed.
[00:26:44 CEST] <iive> however, things are never so simple. e.g. mp3 was published in 1995 and it's last patent expires Decemper 2017, because it's been granted a few years after been filled in (and publication)
[00:27:17 CEST] <iive> mpeg-la does publish list of expired patents, but i couldn't find list of when the rest would expire...
[00:27:28 CEST] <iive> didn't look too hard.
[02:59:49 CEST] <woodbeatstz> Can anyone suggest how to livestream from Linux??
[03:00:30 CEST] <woodbeatstz> I have ffmpeg installed on my remote server
[03:00:52 CEST] <krs_> woodbeatstz: google ffmpeg streaming, you'll find some examples for twitch and others
[03:01:50 CEST] <woodbeatstz> I have tried that and what I need is someone to set  it up
[03:02:17 CEST] <woodbeatstz> I am not a Linux guy  I apoligize
[03:04:37 CEST] <woodbeatstz> What would help is real samples of a Linux server running FFmpeg and doing a live stream
[03:05:40 CEST] <woodbeatstz> I mean like a You Tube thing or LiveStream  is that do able on a linux server?
[03:09:21 CEST] <woodbeatstz> well thanks anyway
[04:18:23 CEST] <fypz> hi, I need some help, I always followed the guides available on trac.ffmpeg.org and easily did it, but this new guide to integrate x265 in it is breaking my balls
[04:19:57 CEST] <fypz> I follow all the steps, almost nothing to compile, but the command to compile ffmpeg says there has to be a configure file here : /home/me/ffmpeg_build/lib/pkgconfig
[04:20:03 CEST] <fypz> well, there isn't
[04:20:33 CEST] <fypz> so I can't do "PATH="$HOME/bin:$PATH" PKG_CONFIG_PATH="$HOME/ffmpeg_build/lib/pkgconfig" ./configure \"
[04:21:34 CEST] <furq> which guide are you following
[04:21:45 CEST] <fypz> https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu
[04:21:47 CEST] <fypz> this
[04:21:53 CEST] <fypz> i'm on Linux Mint 17.3
[04:22:00 CEST] <fypz> so it should work
[04:24:22 CEST] <fypz> only thing that was sketch was the compiling of libvpx, not sure if that was made right, some errors in make and make install, where it installed 6-7 things then said nothing to install
[04:24:34 CEST] <fypz> after doing that : PATH="$HOME/bin:$PATH" make make install
[04:24:42 CEST] <furq> do you have anything in $PREFIX/lib/pkgconfig
[04:24:42 CEST] <fypz> for libvpx
[04:25:49 CEST] <fypz> well, before it's in the lib directory, it's in ffmpeg_build which has "bin" "include" and "lib"
[04:27:11 CEST] <furq> by $PREFIX i mean $HOME/ffmpeg_build
[04:27:18 CEST] <furq> or whatever prefix you're installing to
[04:28:13 CEST] <fypz> the ffmpeg I have running right now did not need those prefix and PATH things, I'm savvy enough to compile other programs generally, I made my own avidemux2.6.10-qt for example
[04:28:29 CEST] <fypz> it's home/me/ffmpeg_build
[04:31:01 CEST] <fypz> but whenever those show up, I didn't grasp what this kind of stuff ever meant, like I compiled the svn of kvirc, version 4.3.1, and I was copy pasting the cmake command because it was a long cmake command which I didn't understand, other than adding in some parameters to it, which is all in the txt for that one
[04:32:48 CEST] <fypz> so excuse my ignorance, thankfully the x264 supporting ffmpeg guide did not have those PREFIX$ and PATH things, I got a vague idea of what they mean, especially PATH
[04:33:50 CEST] <furq> i've never understood why debian/ubuntu guides complicate themselves with all that
[04:33:59 CEST] <furq> the default prefix is fine there
[04:34:02 CEST] <fypz> I modified VLC with a hg command I found somewhere to be able to play x265 files, but I need a x265-able ffmpeg for avidemux and other video programs
[04:35:07 CEST] <furq> but yeah do you have [lib]x265.pc in $HOME/ffmpeg_build/lib/pkgconfig
[04:35:56 CEST] <fypz> there's x265.pc there yes
[04:36:16 CEST] <furq> pastebin your ffmpeg config.log then
[04:36:17 CEST] <fypz> with fdk-aac.pc and vpx.pc
[04:36:21 CEST] <furq> preferably the last 100 lines or so
[04:37:30 CEST] <fypz> uhm, I don't where that is, and I don't think the ffmpeg I compiled myself logged anything like operations in a folder where it's own update is attempted to be compiled
[04:38:10 CEST] <fypz> i mean, why would it
[04:39:07 CEST] <furq> there should be a config.log in $HOME/ffmpeg_sources
[04:39:23 CEST] <fypz> right, I'll open that
[04:39:49 CEST] <furq> if the error message is "x265 not found using pkg-config" then it normally doesn't actually mean that
[04:41:47 CEST] <fypz> no logs in ffmpeg_sources, just a x254 folder a libvpx-1.5.0 folder and the ffmpeg folder itself
[04:42:04 CEST] <fypz> x265*
[04:42:25 CEST] <furq> in ffmpeg/ then
[04:42:43 CEST] <fypz> config.log ?
[04:42:48 CEST] <furq> yes
[04:42:52 CEST] <fypz> alright
[04:45:11 CEST] <fypz> https://www.pastecry.pt/zCkzUV#Az5Ym4Faq0Nes7Zem2Nea0Geg9Pev
[04:45:23 CEST] <fypz> the last 110 lines or so
[04:45:57 CEST] <furq> there's no error there
[04:47:32 CEST] <fypz> okay, but you've seen that the configure file that should be in home/me/ffmpeg_build/lib/pkconfig isn't there
[04:47:41 CEST] <fypz> which prevents compilation from even happening
[04:48:08 CEST] <furq> you just said it was there
[04:48:12 CEST] <fypz> /ffmpegsources $ PATH="$HOME/bin:$PATH" PKG_CONFIG_PATH="$HOME/ffmpeg_build/lib/pkgconfig" ./configure \
[04:48:14 CEST] <fypz> > ^C
[04:48:16 CEST] <fypz> fypz at fypz-desktop ~/ffmpegsources $ PATH="$HOME/bin:$PATH" PKG_CONFIG_PATH="$HOME/ffmpeg_build/lib/pkgconfig" ./configure --prefix="$HOME/ffmpeg_build" --pkg-config-flags="--static" -extra-cflags="-I$HOME/ffmpeg_build/include" --extra-ldflags="-L$HOME/ffmpeg_build/lib"  --bindir="$HOME/bin"  --enable-gpl --enable-libass --enable-libfdk-aac --enable-libfreetype--enable-libmp3lame --enable-libopu --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx2
[04:48:16 CEST] <fypz> 64 --enable-libx265 --enable-nonfree
[04:48:18 CEST] <fypz> bash: ./configure: No such file or directory
[04:48:20 CEST] <fypz> fypz at fypz-desktop ~/ffmpegsources $
[04:48:35 CEST] <fypz> no, I said it was missing lol
[04:48:38 CEST] <furq> yeah configure is in ~/ffmpegsources/ffmpeg
[04:48:49 CEST] <furq> i assume the guide thought you'd extract it directly in ~/ffmpegsources
[04:48:55 CEST] <fypz> I just follow the guide
[04:49:35 CEST] <fypz> the things you compile that you can't download from your distro all go into ffmpeg_build
[04:50:37 CEST] <fypz> so you're saying I should ignore the whole PATH=@HOOME/bin:$PATH thing and just compile straight in the ffmpeg folder?
[04:50:46 CEST] <fypz> with the parameters I want
[04:50:51 CEST] <furq> no
[04:51:00 CEST] <furq> cd ffmpeg
[04:51:02 CEST] <furq> then run that command again
[04:53:41 CEST] <fypz> that worked...I think I forgot to go back to the ffmpeg folder
[04:53:43 CEST] <fypz> lol
[04:53:53 CEST] <fypz> well, thanks a lot
[04:54:24 CEST] <fypz> i'm on sick leave with real bad bronchitis, the hydrocodone cough syrup is making feel woozy
[04:55:10 CEST] <furq> codeine is bad stuff
[04:55:12 CEST] <fypz> thanks again for your time
[04:55:16 CEST] <furq> or fun stuff, depending on which one of my friends you ask
[04:55:28 CEST] <fypz> hydrocodone is different than codeine
[04:56:09 CEST] <fypz> but they are both weak opiates, i just get a small 4oz bottle from my doc whenever I get those recurring fall bronchitis
[04:56:20 CEST] <fypz> not enough to get hooked to
[04:57:05 CEST] <fypz> cocktail of antibiotic/anti-tussive/antihistamine
[04:58:00 CEST] <fypz> thankfully this time, I went to get rid of it fast, a few years ago it turned into pneumonia, had to take prednisolone on top of that to help breathing
[04:58:11 CEST] <furq> nice
[04:59:14 CEST] <fypz> but yeah, coughing so hard your rib bones hurt will likely get you a codeine or hydrocodone script and at least in Canada they are pure, in the US they add in things that are unneeded only to punish someone who would take more teaspoons than needed
[04:59:24 CEST] <fypz> mostly that stuff is the real danger
[05:00:09 CEST] <fypz> like that promethazine/codeine syrup..they get anticholinergic delerium from the promethazine if they try to misuse it
[05:00:52 CEST] <fypz> but yeah, still glad it's illegal in Canada to add compounds to a medication only to punish or even kill an abuser
[05:01:52 CEST] <fypz> alright, thanks again for your time
[05:02:25 CEST] <fypz> that deserves a little money for your project, any way to support you
[05:04:07 CEST] <fypz> oh last question, I've made a .deb from the ffmpeg I have in use right now using checkinstall, is that discouraged? I remember older guides using checkinstall
[05:04:26 CEST] <fypz> a personal .deb for your own machine is always practical
[07:07:41 CEST] <mccc> Hi, I'm using ffmpeg to constantly record segments (-f segment) to a directory I'm watching with systemd (DirectoryNotEmpty) so I can transfer them with rsync.  My problem is I need to not transfer files until they are complete.
[07:07:49 CEST] <mccc> Is there a way to get ffmpeg to write segments to a temporary file, and move the temporary file when complete?  I'm not an expert in ffmpeg but looking through the docs, could that be implemented as a muxer?
[07:41:37 CEST] <EugenA> hi, I'm trying to do streaming to rtmp server. I'm reading file, which is growing. ffmpeg cannot continuously read the file, but it stops. Can I avoid it somehow?
[08:52:26 CEST] <mccc> I'm looking through segment.c and it seems to support this for the list file, but not for the segments.  Unless I'm wrong I don't think this is supported.  I'd love to hear confirmation either way if anyone knows.  Thanks :)
[09:04:31 CEST] <orge> http://pastebin.com/raw/g3xgcCsg if you look at the first two commands, you'll see that by advancing the seek point from 10.513 to 10.515 decreases the frame count from 374 to 125. since this is copying rather than transcoding, you would think this is casued by a key frame between those two seek points, right? but the third command shows that the only nearby key frame isn't between those time points
[15:46:43 CEST] <Mumee> Hey :) I have a little question. I want to convert a mkv file to a mp4 file with the video and one audio stream. But I want the it have always the same language, now to my problem: if the source file dont have this language, how can I tell ffmpeg that it should use the first one?
[16:08:42 CEST] <furq> Mumee: i don't think you can do that with a single command
[16:08:48 CEST] <furq> you'd probably have to check ffprobe first
[16:13:50 CEST] <Mumee> Ok thanks but is there a way to make a batch script that can select it automatically with ffprobe?
[16:14:59 CEST] <furq> http://vpaste.net/1YNsR
[16:15:02 CEST] <furq> that's how you'd do it in shell
[16:15:05 CEST] <furq> no idea about batch
[16:17:34 CEST] <Mumee> Sorry i think i ment shell (the cli in linux?), but i will try to understand what you write
[16:19:07 CEST] <furq> http://vpaste.net/ga6yk
[16:19:22 CEST] <furq> if you're just remuxing then that should work
[16:22:06 CEST] <Mumee> also now I am using this command: http://pastebin.com/raw/P0e3cWRa I also want to convert the video
[16:22:27 CEST] <Mumee> and I dont really understand your code...
[16:24:40 CEST] <furq> http://vpaste.net/H3JZe
[16:25:01 CEST] <furq> save that as convert.sh, run `chmod +x convert.sh`, then run `./convert.sh input.mkv output.mp4`
[16:26:16 CEST] <oerg866> Hello there, I have a question about the -ss and -to parameters. How do they work when I have multiple input files and -map parameters?
[16:27:18 CEST] <oerg866> say my command was this:
[16:27:18 CEST] <oerg866> ffmpeg -i "ofc_monnem.avi" -i "ofc_monnem.wav" -map 0:0 -map 1:0 -vcodec libx264 -acodec aac -pix_fmt yuv420p -preset medium -b:v 5000k -b:a 256k -vf yadif=1,setdar=4:3 C:\temp\ofc_monnem.mp4
[16:27:43 CEST] <oerg866> where would i have to place the -ss and -to commands and do i have to do them multiple times?
[16:27:58 CEST] <furq> it'll seek both files if you use it as an output option
[16:28:19 CEST] <oerg866> But wouldn't that be very slow?
[16:28:44 CEST] <oerg866> I read about that on the ffmpeg wiki and it says
[16:28:45 CEST] <c_14> very is relative, but it'll have to decode everything up to the seek point in all input files
[16:28:46 CEST] <oerg866> ""Here, the input will be decoded (and discarded) until it reaches the position given by -ss. This will be done very slowly, frame by frame. "
[16:29:04 CEST] <furq> it'll take as long as it takes to decode that much of the files
[16:29:21 CEST] <oerg866> It'll take a long time then, since the .avi files are lossless
[16:29:35 CEST] <furq> you can specify it once for each input if you prefer
[16:29:46 CEST] <furq> s/for/before/
[16:29:54 CEST] <oerg866> Ahh so that's how it works!
[16:30:02 CEST] <oerg866> many thanks, that just spared me a lot of headaches :)
[16:30:57 CEST] <furq> i can't help but notice that ffprobe selects subtitle streams with m:language:eng
[16:31:07 CEST] <furq> is there a way to get that stream specifier for just audio streams
[16:33:06 CEST] <c_14> I think I had something like 0:m:language:jpn:a at one point
[16:33:10 CEST] <c_14> Not sure if it worked though
[16:33:38 CEST] <furq> that doesn't seem to work
[16:33:42 CEST] <Mumee> I tested the code you send me.. but could it be that you forget the map in my conversion command?
[16:33:52 CEST] <furq> oh
[16:33:53 CEST] <furq> yeah i did
[16:34:19 CEST] <Mumee> and i dont know if it helps but what about this for your problem: -sn
[16:35:13 CEST] <furq> that doesn't work either
[16:36:48 CEST] <furq> i guess it's back to a distressingly long ffprobe command
[16:37:51 CEST] <Mumee> hmm :/ but can you tell me how i add the mapping thing in my command? i am relay bad at scripting..
[16:39:03 CEST] <c_14> furq: -sn should work (as long as it's after the positive mapping)
[16:39:11 CEST] <c_14> At least -map -0:s seems to disable the subtitle streams
[16:39:34 CEST] <furq> it works but you'll then just get no video
[16:39:40 CEST] <furq> i was hoping for an error
[16:39:57 CEST] <furq> Mumee: http://vpaste.net/IzPQe
[16:40:37 CEST] <furq> speaking of ffprobe, that "stream_disposition=0" should probably be documented somewhere
[16:41:17 CEST] <furq> 15:39:34 ( furq) it works but you'll then just get no video
[16:41:19 CEST] <furq> s/video/audio/
[16:54:18 CEST] <Mumee> furq i tested the code... and it works :3 thank you very much
[17:11:28 CEST] <s0126h> does ffmpeg support h265 encoding
[17:13:25 CEST] <c_14> If built with --enable-libx265
[17:13:58 CEST] <s0126h> does binary in ffmpeg website  use   enable libx265
[17:14:32 CEST] <c_14> This one does https://www.johnvansickle.com/ffmpeg/
[17:14:47 CEST] <c_14> As does the windows one from zeranoe
[17:14:57 CEST] <fritsch> just run: ffmpeg -encoders
[17:14:58 CEST] <fritsch> and see
[17:27:03 CEST] <TwinTailed> [libav], when transcoding to h264, i am getting a missing audio frame (last frame)
[17:34:27 CEST] <Mumee> I am trieng to understand your code.. but i dont understand this part "[ "$type" != "audio" ] && map="0:a:0"" I understand the part in the square brackets, but what does the rest mean?
[17:35:04 CEST] <SchrodingersScat> assigning 0:a:0 to $map
[17:36:03 CEST] <SchrodingersScat> if it's successful then it continues
[17:37:42 CEST] <Mumee> you mean if the code in the brackets is successful it ignores the assigning?
[17:38:03 CEST] <SchrodingersScat> Mumee: no, && is 'and' so both have to be evaluated as true
[17:38:50 CEST] <lexileo> it should be map==, not map=
[17:39:08 CEST] <SchrodingersScat> who's that?
[17:39:11 CEST] <SchrodingersScat> why
[17:39:22 CEST] <Mumee> ok.. then i dont understand why that code works..
[17:39:27 CEST] <lexileo> because == compares, = assigns
[17:39:29 CEST] <SchrodingersScat> is it being evaluated or assigned?
[17:39:42 CEST] <SchrodingersScat> if it's being evaluated then that looks wrong in other ways too
[17:40:14 CEST] <c_14> It's being assigned
[17:40:27 CEST] <c_14> If $type is not audio, then map is 0:a:0
[17:40:29 CEST] <Mumee> also the complete code ist from furq... http://vpaste.net/IzPQe
[17:41:11 CEST] <SchrodingersScat> basically that looks like shorthand for an if, because if $type != audio then the second bit is ran
[17:41:22 CEST] <c_14> It is shorthand for an if
[17:41:23 CEST] <SchrodingersScat> so saves declaring it later, i think
[17:41:43 CEST] <SchrodingersScat> okie dokie then
[17:42:09 CEST] <Mumee> also the && is a replacement for then?
[17:43:05 CEST] <SchrodingersScat> Mumee: for a &&, bash should stop if the first test is negative, is what's happening.
[17:44:20 CEST] <Mumee> But it doesn´t stop... it sets map to map="0:a:0"
[17:44:24 CEST] <SchrodingersScat> I guess I don't see []'s being used outside of if/while/for/etc much.  && is pretty standard.
[17:44:36 CEST] <SchrodingersScat> Mumee: was type != audio?
[17:45:10 CEST] <Mumee> i posted the code: http://vpaste.net/IzPQe
[17:47:12 CEST] <SchrodingersScat> Mumee: I'm saying that if it didn't stop, then $type != audio
[17:48:45 CEST] <Mumee> Sorry.. my english is not the best... :/ But you are just talking from that line? the code after that will be run anyway?
[17:49:05 CEST] <SchrodingersScat> Mumee: if the condition evaluates as true
[17:49:30 CEST] <SchrodingersScat> Mumee: basically if the exit code is 0 for 'successful'
[17:50:44 CEST] <Mumee> Thanks.. i hope i have understand it now :)
[17:51:15 CEST] <TwinTailed> Does anyone know why I am getting a missing last frame from audio stream when transcoding with libav?
[17:52:47 CEST] <furq> Mumee: it's exactly the same as if [ "$type" != "audio" ]; then map="0:a:0"; fi
[17:54:26 CEST] <Mumee> Ok then i understood it
[17:54:47 CEST] <furq> i could have also done [ "$type" == "audio" ] || map="0:a:0"
[17:56:50 CEST] <Mumee> Is this anywhere documented? because till today i never saw this..
[17:58:52 CEST] <furq> http://pubs.opengroup.org/onlinepubs/009604499/utilities/xcu_chap02.html#tag_02_09_03
[17:59:01 CEST] <SchrodingersScat> Mumee: it's bash, man bash
[17:59:35 CEST] <SchrodingersScat> Mumee: oh, or sh
[17:59:50 CEST] <furq> the same text is in bash's manpage
[18:00:54 CEST] <Mumee> Well.. i said i am not good in this..
[18:14:51 CEST] <TwinTailed> Does anyone know why I am getting a missing last frame from audio stream when transcoding with libav?
[18:17:51 CEST] <classy_cameltoe> what are the pros and cons of utvideo vs huffyuv ?
[18:19:03 CEST] <classy_cameltoe> reason I am asking is utvideo seems to crash vspipe a lot.
[18:19:15 CEST] <furq> ut video should be faster and give better compression but it's less widely supported
[18:19:24 CEST] <furq> i'd pick ffv1 over either of them
[18:20:39 CEST] <classy_cameltoe> gonna take a look at ffv1. Do you know if it is fully supported by VapourSynth?
[18:20:57 CEST] <furq> as an input?
[18:21:03 CEST] <classy_cameltoe> output
[18:21:35 CEST] <furq> the output codec shouldn't make any difference
[18:22:43 CEST] <classy_cameltoe> Just had a look at my convoluted way of doing this. Going to be piping the video out of VS with VSpipe into ffmpeg to mux in the sound. Guess the best way to try is to test it.
[18:23:19 CEST] <classy_cameltoe> but yeah, ut video crashed way to much. huffyuv is not crashing...
[18:23:41 CEST] <classy_cameltoe> C:\VideoStuff\VS\VSPipe.exe -y C:\VideoStuff\VS\PART_2.vpy - | C:\VideoStuff\ffmpeg.exe -i - -i D:\Videos\tape_1_dalerna_LA_dahab.avi -c:v utvideo -c:a copy -map 0:0 -map 1:1 -vf setdar=4:3 D:\Videos\ALL_finished.avi -y
[18:24:46 CEST] <classy_cameltoe> ohhhh spiffy. ffv1 works very nicely.
[18:28:00 CEST] <classy_cameltoe> Is there any consensus on a archival format for the preservation of video?
[18:28:43 CEST] <furq> ffv1 seems like a good choice
[18:29:11 CEST] <furq> good compression, open-source, widely supported and will exist for at least as long as ffmpeg does
[18:31:41 CEST] <furq> https://en.wikipedia.org/wiki/FFV1
[18:31:46 CEST] <furq> apparently a few institutions are already using it
[18:31:47 CEST] <classy_cameltoe> yeah reading that now
[18:32:06 CEST] <classy_cameltoe> As of 2015, standardization of FFV1 through the Internet Engineering Task Force (IETF) is work in progress as part of the European PREFORMA Project
[18:32:38 CEST] <JEEB> the only issue is that FFV1 is not as well supported in video editors (commercial), the only thing that supported it was the VFW component in FFDShow-tryouts, and that thing is unfortunately a mess and I don't think it has been developed for ages now
[18:32:40 CEST] <furq> there are other lossless codecs with better compression but they're mostly/all patent-encumbered or closed-source
[18:32:50 CEST] <JEEB> and it lacks a QT component methinks
[18:32:56 CEST] <furq> the only competitor i know of as far as compatibility is huffyuv
[18:32:57 CEST] <JEEB> (for mac editors)
[18:33:38 CEST] <JEEB> basically if you're OK with transcoding into ut video or something for editing, then ffv1 archival should be OK
[18:35:06 CEST] <furq> it's part of lav filters isn't it
[18:35:14 CEST] <JEEB> that's DShow
[18:35:19 CEST] <JEEB> DShow is generally not used by video editors
[18:35:25 CEST] <furq> apparently premiere uses dshow
[18:35:32 CEST] <classy_cameltoe> Prefer to stick with something that will be around in say 20 years. Seems ffv1 might be a good one to use.
[18:35:33 CEST] <JEEB> ok, color me surprised
[18:35:38 CEST] <furq> according to that page anyway
[18:35:47 CEST] <JEEB> I'd be really surprised if it does, though :D
[18:36:05 CEST] <JEEB> since all of those editors have worked really hard to only have QT/VfW/custom internal stuff available for input/output
[18:36:18 CEST] <JEEB> and output with DShow doesn't really exist although it definitely is possible
[18:36:21 CEST] <furq> does ut video work with quicktime
[18:36:25 CEST] <JEEB> yes
[18:36:28 CEST] <JEEB> it has a QT component
[18:36:36 CEST] <furq> how annoying
[18:36:40 CEST] <classy_cameltoe> lol
[18:36:43 CEST] <JEEB> the umezawa guy made a thing for VfW, DShow, MF, QT
[18:37:15 CEST] <furq> it's not an issue if you're not using a commercial editor anyway
[18:37:28 CEST] <JEEB> true
[18:37:45 CEST] <TwinTailed> JEEB: Have you got any idea why I am having a missing last frame from the audio stream when transcoding using libav?
[18:38:07 CEST] <JEEB> in like 2013 or so ffv1 was still much less usable, I remember encoding some stuff with it and then the decoding was *slow*
[18:38:11 CEST] <JEEB> that seems to have been fixed since
[18:38:17 CEST] <JEEB> but left a really bad aftertaste
[18:38:29 CEST] <furq> decoding is pretty fast in my experience
[18:38:33 CEST] <furq> i've not compared it to ut though
[18:38:58 CEST] <JEEB> ut has no real optimizations so if whatever was happening with my encodes isn't happening any more it should be quite fast
[18:39:05 CEST] <JEEB> although ut video is pretty darn simple :D
[18:39:09 CEST] <classy_cameltoe> I guess long term standards are not practical. I have to dick around with Ada all day at work since they wont let it go...
[18:39:17 CEST] <furq> i vaguely remember decoding speed being a selling point of ut
[18:39:23 CEST] <JEEB> (I was able to implement as my babby's first encoder)
[18:39:30 CEST] <JEEB> yes, the official components are optimized very well
[18:39:45 CEST] <JEEB> the libavcodec encoder isn't really :P
[18:40:02 CEST] <JEEB> (and I couldn't just grab the optimizations from original as it's GPL)
[18:40:12 CEST] <JEEB> (although I could have just asked if it was OK or something)
[18:40:17 CEST] <furq> thanks richard
[18:42:33 CEST] <classy_cameltoe> Get this every now and then...
[18:42:37 CEST] <classy_cameltoe> [yuv4mpegpipe @ 0000000000dd6860] Thread message queue blocking; consider raising the thread_queue_size option (current value: 8)
[18:42:50 CEST] <classy_cameltoe> what it mean?
[18:45:35 CEST] <classy_cameltoe> somebody on the Internet (gotta be correct) says to use this -thread_queue_size 512. Any truth to this?
[18:55:24 CEST] <furq> it doesn't make any difference in my experience
[18:57:25 CEST] <furq> there's no harm in using it though
[18:59:17 CEST] <classy_cameltoe> thx furq. Kinda wondering what is meant by queue blocking of threads. I got a hella lot cores on this machine.
[18:59:49 CEST] <classy_cameltoe> A duel Xeon 2630v3 setup.
[19:00:13 CEST] <TwinTailed> when do i use avcodec_flush_buffers after EOF?
[19:09:24 CEST] <flux> twintailed, ..did you its documentation? you don't: http://ffmpeg.org/doxygen/trunk/group__lavc__misc.html#gaf60b0e076f822abcb2700eb601d352a6
[19:09:38 CEST] <TwinTailed> did
[19:09:59 CEST] <TwinTailed> flux: I wanted to know whether after sending NULL or when getting EOF from decoder
[19:11:01 CEST] <flux> well, none of the doc/examples/*.c use it
[19:11:24 CEST] <flux> ffmpeg.c itself uses it once, when seeking
[19:12:06 CEST] <flux> apparently ffplay.c uses it once, but the conditions it uses it in are a bit unclear to me..
[19:14:22 CEST] <TwinTailed> flux: no idea what causes missing last audio frame?
[19:15:04 CEST] <flux> twintailed, as I said, you should just keep internal counters and determine how many frames of audio data you receive and output
[19:15:21 CEST] <flux> twintailed, and then find if and why there is a discrepancy
[19:15:23 CEST] <TwinTailed> flux: how can i get which frame i am writing?
[19:15:47 CEST] <TwinTailed> flux: pkt->pos;?
[19:15:50 CEST] <flux> you don't need to keep track of frame numbers. just if (frame type == audio) { ++number_of_audio_frames; }
[19:16:33 CEST] <flux> also other suggestions: try to use input material that ONLY has a video track, and input material that ONLY has an audio track
[19:16:46 CEST] <flux> I sort of expect that it's always the last frame you drop, not just the last audio frame
[19:17:00 CEST] <flux> next you could generate test material that ie. has only 1 frame
[19:17:04 CEST] <flux> or two frames
[19:17:18 CEST] <flux> and just step through your code with it and get better understanding of the problem..
[19:17:28 CEST] <TwinTailed> flux: ffprobe showed that all video frames are there
[19:18:38 CEST] <flux> welll, let's say you have two audio tracks
[19:18:47 CEST] <flux> do you expect that both of the tracks will miss the last frame?
[19:19:35 CEST] <TwinTailed> flux: i don't know tbh
[19:20:26 CEST] <flux> well, find out ;)
[19:22:30 CEST] <khdsfbads> how to extract audio frommp4 file and save it as webm audio ? and how to convert m4a file to webm audio ?
[19:24:51 CEST] <khdsfbads> *from mp4*
[19:53:13 CEST] <TwinTailed> flux: the has_b_frames is 0 for the original video file, but is 2 for the generated
[19:53:19 CEST] <TwinTailed> flux: would it cause anything?
[20:07:01 CEST] <Ponder1787> I have cleaned up a bunch of temporal noise in some video8 that I had. Looks good but I now have what looks like a non moving screen superimposed. Any idea what that can be? Seen this before when using temporal "cleaning"
[20:07:46 CEST] <Ponder1787> Or better yet. How would I clean that up. Using a mix of VapourSynth and ffmpeg
[20:14:44 CEST] <TwinTailed> flux: number of successfully decoded audio frames is 5205, number of frames passeing avcodec_send_frame is also 5205, but the number of frames passing avcodec_receive_packet is 5204
[20:15:57 CEST] <TwinTailed> by passing i mean the function returns success (0)
[20:27:46 CEST] <TwinTailed> flux: the function failing in that frame is returning AVERROR(EAGAIN)
[20:33:33 CEST] <mizux> hi
[20:34:26 CEST] <mizux> due my home video player limitation -_- do you know how to change the compatible_brands ?
[20:34:32 CEST] <mizux> from isomiso2avc1mp41 to isommp42
[20:34:46 CEST] <mizux> it seems the first one is not recognize
[20:35:08 CEST] <mizux> also minor version is 512 for the first and 0 for the later
[20:55:33 CEST] <TwinTailed> mizux: ffmpeg doesn't support mp42
[20:58:32 CEST] <mizux> TwinTailed, do you mean i can convert isom/isomiso2avc1mp41 to mp42/isommp42 ?
[20:58:37 CEST] <mizux> can't*
[21:07:39 CEST] <TwinTailed> mizux: I doubt, but why would you anyways?
[21:10:04 CEST] <mizux> the first one is not recongize by my home video player -_-
[21:10:17 CEST] <mizux> but the second yes
[21:14:26 CEST] <brrz> dont know if this is the right channel, but I am trying to get VAAPI with intelHD to work. vainfo seems to work but ffmpeg exits with No VA display found for device: /dev/dri/card0
[21:17:25 CEST] <brrz> whole ffmpeg command: http://pastebin.com/8NGDFJXz
[21:21:13 CEST] <mizux> MP4Box -brand mp42 file.mp4
[21:21:26 CEST] <mizux> (package gpac) Thanks all
[21:53:18 CEST] <nutopic> hello
[21:53:54 CEST] <nutopic> trying to encode bink video to x264
[21:54:09 CEST] <nutopic> ffprobe and ffmpeg don't like it
[21:54:16 CEST] <nutopic> "a1_p1_paul_presentation.bk2: Invalid data found when processing input"
[21:54:21 CEST] <nutopic> i thought it was supported?
[21:59:19 CEST] <TwinTailed> nutopic: x264 is supported by ffmpeg
[21:59:31 CEST] <TwinTailed> using libx264
[22:18:57 CEST] <orge> http://pastebin.com/raw/g3xgcCsg the first two commands show that by advancing the seek point from 10.513 to 10.515 the frame count decreases from 374 to 125. since this is copying rather than transcoding, you would think this is caused by a key frame between those two seek points, right? but the third command shows that the only nearby key frame isn't between those time points. can someone explain? am i misunderstanding something?
[22:27:39 CEST] <DHE> seek position is open to a bit of interpretation. audio and video frames often aren't exactly time synchronized, and some formats have a third pure timestamp set of metadata.
[22:28:26 CEST] <DHE> that is, if you run "ffprobe -show_frames" the pts and dts won't be strictly monotonic when not filtering between audio and video
[22:38:43 CEST] <nutopic> yeah i know twintailed my issue was with the input
[22:42:54 CEST] <orge> DHE, i don't entirely understand what you're saying, but this is just video, no audio
[22:47:55 CEST] <DHE> no audio in the stream?
[22:49:57 CEST] <orge> DHE, no. the only stream 0:0 is video. although it's from a mkv with video audio subtitles, etc
[22:53:33 CEST] <oerg866> If I split videos using -ss and -t, what do I have to do to make sure that i can join them together again?
[22:53:43 CEST] <oerg866> (I'm editing commercials out of some old TV recordings)
[22:54:52 CEST] <stah0121> hello. just wanted to say I think ffmpeg is an amazing project and I'm thankful for everyone who has contributed. I plan to send a donation soon
[22:55:06 CEST] <ericwooley> I'm getting an ffmpeg error: ': Input/output error' when I try to list the devices using: `ffmpeg -f avfoundation -list_devices true -i ""` ===> full output http://pastebin.com/rD1ZWHup
[22:55:15 CEST] <ericwooley> anyone have any idea what might be going on
[22:57:50 CEST] <rkern> ericwooley: I can reproduce it. Let me see if it's something easy.
[22:58:10 CEST] <ericwooley> @rkern Thank you sir
[22:58:41 CEST] <ericwooley> @oerg866 I'm no expert, but I don't think you need to do anything special to be able to rejoin them
[23:02:22 CEST] <stah0121> is there some kind of standardized documentation for all the different AV codecs ?
[23:04:56 CEST] <ericwooley> @stah0121 Someone can correct me if I am wrong, but my understanding is that codecs follow a pretty string api, so ffmpeg can interact with them the same way no matter what codec it is. So the standard doc would be just the ffmpeg docs. Anything special for a specific codec would be, by definition, non-standardizable
[23:05:09 CEST] <ericwooley> and by string i mean strick
[23:05:11 CEST] <ericwooley> strict
[23:05:16 CEST] <stah0121> if I wanted to create an application that handles encoding or decoding of 1 kind of codec, where would I find info on how to implement it?
[23:05:16 CEST] <ericwooley> oh man, bad typing day for me
[23:05:44 CEST] <DHE> stah0121: there's a doc/examples directory with a handful of narrow use case but still useful applications
[23:06:26 CEST] <ericwooley> what do you mean by encoding/decoding?
[23:07:32 CEST] <DHE> are you planning to write a codec or just write a frontend using libav?
[23:08:45 CEST] <stah0121> I'm partially trying to get a better understanding of how everything fits together, because for example there are lots of apps that can handle things like transcoding wav to mp3 and spitting out a file.
[23:09:30 CEST] <stah0121> I am trying to do some research on creating my own home media type server, and I'd like to take advantage of some of the ffmpeg libraries but I am trying to get a feel for ffmpeg itself, and the things its implemented to do
[23:10:01 CEST] <DHE> most common parameters are available in the AVCodecContext structure. bitrate, quality caps, keyframe intervals, image resolution, etc.
[23:10:22 CEST] <DHE> that's a vast oversimplification...
[23:10:27 CEST] <stah0121> ha, yeah I hear ya
[23:10:36 CEST] <stah0121> I just cloned the main git repo and have been browsing the code
[23:11:28 CEST] <stah0121> my background is in network protocols and following RFC's and things, and for audio/video stuff it sort of seems like codecs are sort of like protocol definitions ? or am I wrong
[23:11:42 CEST] <DHE> failing that, there is an AVDictionary which allow passing arbitrary key=value pairs for codec-specific or even generic options
[23:14:23 CEST] <DHE> a decoder codec is responsible for taking an AVPacket (variable size payload from disk) and producing an AVFrame (raw video frame, fixed size, width * height * bytes_per_pixel)
[23:14:27 CEST] <DHE> and the encoder does the reverse
[23:15:14 CEST] <DHE> while codecs are specified by standards and there are reference encoders/decoders, there tends to be a lot of work in making a good codec that performs well and produces good image quality
[23:15:19 CEST] <ericwooley> rkern: find anything? It's not so bad, except it's causing output to go through stderr, not stdout. So I can listen to both streams, but it's just adding complication. Any idea if it's something I need to fix on my system or if it's an error with ffmpeg?
[23:16:25 CEST] <ericwooley> If you think it would be a relatively easy thing to do, I can try to put together my first PR that fixes it
[23:16:41 CEST] <rkern> ericwooley: you can ignore it - an error is returned in all cases when you list devices
[23:17:20 CEST] <ericwooley> K, for now I will just list on stderr and stdout then. Thanks for checking into it
[23:18:08 CEST] <ericwooley> I'll create a github issue for it
[23:18:13 CEST] <rkern> The issue is in avfoundation.m in avf_read_header. When it lists devices, it does a "goto fail;" to clean up instead of initializing.
[23:18:36 CEST] <ericwooley> err, never mind, there doesn't seem to be issues on ffmpeg
[23:18:48 CEST] <ericwooley> interesting
[23:19:34 CEST] <ericwooley> is there an ffmpeg bug tracker? and rkern: would you like me to open an issue, it seems like you have tracked it down already
[23:19:51 CEST] <DHE> part of the wiki, see the topic
[23:20:49 CEST] <ericwooley> ah yes, always read the header. My bad.
[23:21:39 CEST] <rkern> ericwooley: not sure if you've contributed before, but if you'd like to, this is something easy to start with.
[23:22:25 CEST] <ericwooley> ffmpeg is such an awesome project, I would love to contribute. I am currently trying to figure out what the whole process is
[23:22:48 CEST] <ericwooley> I have done little more than install a binary, so this might take me a while haha
[23:23:54 CEST] <Ponder1787> lol, cool the linux environment in Windows 10 actually works. Try something like watch -n1 'grep MHz /proc/cpuinfo |sort -k4rn' and it works. raw sockets too... finally!
[23:24:14 CEST] <ericwooley> I don't know if I have time until next weekend though rkern. If you want it to be fixed sooner rather than later, it might not be a good idea to wait on me
[23:25:04 CEST] <rkern> Not a problem. There's a ton of good info at https://ffmpeg.org/developer.html#Contributing and new devs are always welcome.
[23:25:33 CEST] <rkern> If I'm around I'd be happy to help you out.
[23:26:19 CEST] <stah0121> Thanks for the info DHE.
[23:26:54 CEST] <stah0121> If someone were to start contributing, what area would you say needs the most help ?
[23:26:55 CEST] <ericwooley> Awesome thanks! This will be my new project over the week (if I get a free minute) and into next weekend
[23:27:25 CEST] <DHE> dunno, I am but a user. a power user maybe, but a user all the same
[23:28:37 CEST] <Dresk|Dev> So ffmpeg is pretty darn slow on my Pixel C tablet (Tegra X1 SoC) and my NVIDIA Shield Tablet (Tegra K1 SoC), we're using 2.8.6, we've got NEON enabled and I believe as much GL stuff as we can do, obviously we're not hardware decoding, but have the recent versions of ffmpeg improved Android performance at all?
[23:30:15 CEST] <Ponder1787> Talk about The Pixel. The Pixel phone is coming out in two days right?
[23:30:35 CEST] <Ponder1787> Google Pixel that is.
[23:31:01 CEST] <Dresk|Dev> Yeah they are announcing them, gonna be using Snapdragon 820 SoCs, not the best things out there, the Tegra X1 is definitely more powerful, but it's at 2560x1800
[23:31:12 CEST] <Ponder1787> Not the 821?
[23:31:19 CEST] <Dresk|Dev> Doesn't look like it
[23:31:48 CEST] <Ponder1787> Hope they keep a sane price point on them. Been using Nexus for years.
[23:32:18 CEST] <Dresk|Dev> The prices actually are said to start at $650
[23:32:24 CEST] <Ponder1787> ouch
[23:32:47 CEST] <Ponder1787> ... reluctantly reaches into wallet.
[23:33:31 CEST] <Dresk|Dev> So, any idea on my question?
[23:34:04 CEST] <Ponder1787> You sure NEON is working?
[23:34:35 CEST] <stah0121> anyone got any books they recommend on ffmpeg and/or low level audio/video development ?
[23:34:52 CEST] <ossifrage> I'm having troubles with my 'drawtext' jumping around by a few pixels, even tough I'm using a fixed width font...
[23:35:12 CEST] <JEEB> Dresk|Dev: feel free to check if this works any better :) https://github.com/mpv-android/mpv-android/releases/tag/mpv-android-2016-09-18
[23:35:28 CEST] <JEEB> (there's now basic mediacodec integration in FFmpeg which that is utilizing)
[23:36:32 CEST] <ossifrage> I'm using 'x=w-tw-4:y=4' for the text position. Any ideas? (the font is Roboto-Bold.ttf)
[23:37:01 CEST] <ericwooley> stah0121 Well, i know of one easy bug haha
[23:37:52 CEST] <stah0121> haha
[23:38:08 CEST] <Dresk|Dev> Ponder1787: For Neon, we have #define HAVE_NEON 1 , #define HAVE_NEON_EXTERNAL 1 , #define HAVE_NEON_INLINE 0 , #define HAVE_INTRINSICS_NEON 1 , #define CONFIG_NEON_CLOBBER_TEST 0
[23:38:20 CEST] <stah0121> what sort of development environment is common for ffmpeg stuff ?
[23:38:36 CEST] <stah0121> I'm used to vim, ctags, gcc, etc
[23:38:43 CEST] <JEEB> your usual gcc/shell/yasm (last for IA32/x86_64)
[23:38:45 CEST] <Ponder1787> then you will be fine
[23:39:00 CEST] <JEEB> as for editor everyone uses what they're comfy with
[23:39:09 CEST] <JEEB> I use vim and a more graphical editor usually
[23:39:24 CEST] <Ponder1787> I use everything from emacs to nopepad
[23:39:31 CEST] <Ponder1787> notepad
[23:39:43 CEST] <stah0121> I was doing a bit of reading on yasm, I'm not familiar. Seems powerful though. Do you sort of use it in place of doing like a gcc compile ?
[23:39:59 CEST] <stah0121> JEEB what sort of graphical editor ?
[23:40:12 CEST] <JEEB> no, it's for stuff where people go and write highly optimized versions of a part of a decoder/encoder/filter
[23:40:18 CEST] <DHE> anything. eclipse, gedit...
[23:40:52 CEST] <JEEB> sublime text is currently what I use since it kind of got me, although for non-FFmpeg stuff I use stuff from intellij
[23:41:20 CEST] <stah0121> gotcha
[23:42:57 CEST] <stah0121> I sometimes wonder if I should switch to emacs haha, but vim has done well for me and overall seems fairly straight forward for how complex you can get with it
[23:43:58 CEST] <Ponder1787> once use start using emacs you will prolly stick to it. I love it. Use a nice solorized theme.
[23:44:29 CEST] <ossifrage> The problem is tw isn't a constant value for a fixed length string and a fixed with font...
[23:45:00 CEST] <ossifrage> I just gave up and used 'x=w-max_glyph_w*12' for my 12 character string
[23:48:15 CEST] <stah0121> does ffprobe rely on any graphical component ?
[23:48:34 CEST] <BtbN> depends on what you link it against.
[23:49:10 CEST] <stah0121> when I was reading about ffprobe I kept thinking about the spectrum analyzer in DAW programs like ableton
[23:49:33 CEST] <stah0121> if I hooked it into some graphic lib, would I be able to get a similar sort of realtime graphical window ?
[23:50:10 CEST] <BtbN> no
[23:52:35 CEST] <stah0121> ok
[00:00:00 CEST] --- Mon Oct  3 2016



More information about the Ffmpeg-devel-irc mailing list