[Ffmpeg-devel-irc] ffmpeg.log.20150603

Thu Jun 4 02:05:01 CEST 2015

[00:16:20 CEST] <ruukasu> is there any way to remove a subtitle track from an mkv without re-encoding anything? `-acodec copy -vcodec copy` also copies the subtitle track which I no longer need
[00:18:04 CEST] <grepper> ruukasu: did you try -sn ?
[00:18:21 CEST] <llogan> or explicitly map what you want, or use a negative mapping
[00:18:36 CEST] <llogan> -c copy -map -0:s
[00:18:42 CEST] <llogan> several options to do what you wanty
[00:18:46 CEST] <llogan> *want
[00:19:19 CEST] <ruukasu> ah
[00:19:21 CEST] <ruukasu> -sn worked
[00:19:23 CEST] <ruukasu> thanks
[00:51:29 CEST] <JodaZ> why remove *_*
[01:33:13 CEST] <Canar> hey guys, i'm trying to unpack the b-frames on an old xvid file
[01:33:52 CEST] <Canar> so i've got the ffmpeg-git build that has the feature, as well as the deb-multimedia.org version installed
[01:34:23 CEST] <Canar> issue is that the former won't display the message that the video has such a b-frame while the latter lacks the features to remove them
[01:34:47 CEST] <Canar> i'm not sure where the problem lies
[01:36:44 CEST] <llogan> Canar: you mean this? http://ffmpeg.org/ffmpeg-bitstream-filters.html#mpeg4_005funpack_005fbframes
[01:37:29 CEST] <Canar> precisely, yes
[01:38:29 CEST] <llogan> for both versions
[01:38:37 CEST] <Canar> yeah
[01:38:54 CEST] <Canar> wondered if there was something obvious
[01:38:58 CEST] <Canar> guess not
[01:38:59 CEST] <Canar> sec
[01:41:19 CEST] <Canar> http://pastebin.com/cAqLk4Ry
[01:41:34 CEST] <Canar> therels me executing the git build containing the bsf
[01:42:00 CEST] <Canar> http://pastebin.com/3rdYXyCq
[01:42:16 CEST] <Canar> there's ffprobe from deb-multimedia saying there's an issue
[01:43:06 CEST] <Canar> that particular build was me trying to mimick the deb-multimedia.org build flags
[01:43:25 CEST] <Canar> problem does not resolve using less explicit flags
[01:43:40 CEST] <llogan> you don't need to copy those flags...only what you need
[01:43:51 CEST] <Canar> well
[01:44:01 CEST] <Canar> i tried the less verbose approach
[01:44:03 CEST] <Canar> >,>
[01:44:13 CEST] <Canar> doesn't change anything
[01:44:20 CEST] <Canar> the bsf doesn't appear to do anything
[01:44:31 CEST] <llogan> does the new ffprobe squawk the same message?
[01:44:51 CEST] <Canar> no
[01:44:59 CEST] <Canar> as i said at the start
[01:45:20 CEST] <llogan> you can test to see if the bsf does anything: ffmpeg -i ... -f md5 -
[01:45:58 CEST] <Canar> yes, the output and input files differ
[01:46:10 CEST] <Canar> but ffprobe gives the same error regardless
[01:47:59 CEST] <llogan> i meant to, on the new ffmpeg, run the command with and without the bsf to see if the md5 changes
[01:48:21 CEST] <llogan> i don't know why the message doesn't show up anymore
[01:48:36 CEST] <Canar> again, yes, the md5 changes
[01:49:03 CEST] <Canar> as does the filesize, etc
[01:54:13 CEST] <llogan> Canar: you can perform a git bisect to see when the message quit being displayed
[01:55:04 CEST] <Canar> yeah... i was just hoping that someone had dealt with something like this
[01:55:27 CEST] <Canar> i have no desire to dump hours more time into one little error message when it'd be more cost-effective to go pirate an error-free source
[01:56:14 CEST] <Canar> just a bizarre circumstance that i had no capacity to comprehend
[01:57:08 CEST] <llogan> the fact that you're here makes me conclude you have no girlfriend, and therefore have several hours to volunteer
[01:57:28 CEST] <Canar> lol
[01:57:40 CEST] Action: llogan was joking of course
[01:58:15 CEST] <Canar> ...maybe i'll try the git bisect approach later though
[01:58:29 CEST] <Canar> used to suggest avidemux
[01:58:38 CEST] <Canar> which performed abominably at helping the circumstance also
[01:58:51 CEST] <Canar> i have to wonder if there isn't something more deeply messed up about the file
[01:59:06 CEST] <Canar> turns out the new avidemux has deprecated most of the bframe-fixing features
[01:59:54 CEST] <Canar> WHY DIVX WHY
[02:15:06 CEST] <llogan> Canar: probably http://git.videolan.org/?p=ffmpeg.git;a=commit;h=1f47c2399da2e814228f2526b09f8ab94acc7989
[02:16:56 CEST] <llogan> then the message was later changed to mention using the bsf instead of avidemux
[02:24:59 CEST] <cehoyos_> Canar: Could you provide a sample?
[02:29:23 CEST] <cehoyos_> Canar: You can use ffplay to see the warning
[03:17:02 CEST] <JustAPerson> -/close
[03:32:42 CEST] <Canar> llogan: that might actually be it
[03:33:12 CEST] <Canar> cehoyos_: going to see if the change llogan linked describes my problem. the "problematic" probe is from a build that predates said commit
[04:18:24 CEST] <llogan> Canar: a sample is, of course, always welcome.
[04:40:14 CEST] <Canar> llogan: i plead the fifth =3
[04:44:11 CEST] <Canar> http://canar.moe/head.avi
[04:44:23 CEST] <Canar> head -c 4MiB
[04:44:47 CEST] <Canar> enough to trigger the complaint
[04:45:18 CEST] <Canar> i'd laugh were the complaint spurious
[04:55:05 CEST] <Canar> llogan / cehoyos_ see above
[05:39:20 CEST] <llogan> Canar: thanks
[06:58:57 CEST] <ruukasu> I have an mkv with a sub track and I want to add another track. I'm trying `ffmpeg -i in.mkv -i in.ass -acodec copy -vcodec copy -scodec copy out.mkv`, but it's only outputting the subtitles from the original mkv, without the additional ones in the .ass. am I doing something wrong?
[07:05:18 CEST] <relaxed> ruukasu: try, ffmpeg -i in.mkv -i in.ass -map 0 -map 1 -c copy out.mkv
[07:12:27 CEST] <ruukasu> relaxed: perfect, thanks
[07:52:37 CEST] <chocolate-elvis> I have a sRGB 4:4:4 DPX sequence I want to convert to a intermediate file for later x264 /-pix_fmt yuv420p encoding. what's a good approach? encode to a 0 CRF x264 file?
[08:42:23 CEST] <brontosaurusrex> chocolate-elvis: there is also lossless x264
[08:46:00 CEST] <brontosaurusrex> i guess crf 0 should be lossless if it is 8bpc
[08:51:19 CEST] <gonz58> MAX_RESYNC_SIZE in  libavformat/mpegts.c, does it determine the amout of bytes FFmpeg is investigating in order to find a resync byte (0x47)?
[09:57:07 CEST] <SleepyPikachu> Anyone got a one liner to combine a single image with an audio file to make a video file?
[10:18:53 CEST] <muber> I want to use a mp4 as input and then forward it to a tcp socket, but the process is too quick, I want it to feed slowly as same as fps that the video is played. how should I do ? I use the command below:
[10:19:02 CEST] <muber>  /opt/ffmpeg -i test.mp4 -f flv -c:a copy -c:v copy tcp://127.0.0.1:6666
[10:19:05 CEST] <BtbN> -re input option
[10:19:37 CEST] <muber> BtnN: OK, I will try -re, Thanks.
[11:07:43 CEST] <chama> Hi, I have a issue with looping an audio in ffmpeg. Following is the command I'm using,
[11:07:53 CEST] <chama> ffmpeg -i 1.mp3 -filter_complex "[0:a][0:a][0:a]concat=n=3:v=0:a=1[a]" -map [a] out.mp3
[11:08:10 CEST] <chama> is there anything wrong with the command I'm using?
[11:09:42 CEST] <chama> it need to loop through the audio 3 times. But first loop is ok and 2,3 loops are not fully available in the out put
[11:09:52 CEST] <chama> something like
[11:10:01 CEST] <chama> original duration = 10
[11:10:10 CEST] <chama> output duration = 10+7+7
[11:10:44 CEST] <chama> what am i doing wrong or is there any other way to do this?
[11:19:17 CEST] <cehoyos_> chama: Works fine here with a random sample, please provide console output (pastebin)
[11:22:24 CEST] <Guest40634> Hi everyone! I have a longstanding OpenEXR conversion problem, concerning linear to srgb conversion. I was hoping that the 2.5 or 2.6 versions solve it but did not find anything in the release notes.
[11:22:30 CEST] <chama> @cehoyos_ : pastebin.com/pScrVuTn
[11:24:36 CEST] <Guest40634> I haven't found a proper colorspace conversion, so to approximate I use a 2.2 gamma correction using -vf "lutrgb=r=0.454545:g=0.454545:b=0.454545"
[11:25:04 CEST] <cehoyos_> chama: It works fine here with a random sample, please upload your sample.
[11:25:26 CEST] <cehoyos_> Guest40634: Where did you report your issue? (Before now I mean)
[11:25:28 CEST] <Guest40634> And it sort of works but it looks that it does the gamma correction after the float/half to 8-bit conversion and that creates banding artifacts.
[11:26:16 CEST] <Guest40634> cehoyos: I did not report it, because I found it mentioned years ago in old mailing list threads.
[11:26:32 CEST] <cehoyos_> That is usually bad (if you want the issue fixed).
[11:26:39 CEST] <Guest40634> cehoyos: What I did not found is whether it was fixed in the meantime.
[11:26:53 CEST] <cehoyos_> How / why should it?
[11:27:06 CEST] <cehoyos_> I thought there is a patch that fixes your issue but cannot find it atm.
[11:27:11 CEST] <cehoyos_> I may mix things up, sorry.
[11:27:29 CEST] <cehoyos_> (But I thought I had just read the "float gamma" issue a few days ago.)
[11:27:35 CEST] <Guest40634> cehoyos: Well, if it was reported in 2009, 2010, 2012 and even patches were sent than I suspected that it was noticed. :)
[11:27:54 CEST] <cehoyos_> So allow me to repeat: Where was it reported?
[11:28:14 CEST] <cehoyos_> Did you test the patches? Did they work for you?
[11:28:57 CEST] <Guest40634> For example in this thread: https://ffmpeg.org/pipermail/ffmpeg-devel/2014-April/157139.html
[11:30:13 CEST] <cehoyos_> This particular patch was added to FFmpeg over one year ago.
[11:30:20 CEST] <Guest40634> cehoyos: I did not have the skills to apply the patch and compile ffmpeg on windows, I hoped having a working solution.
[11:30:29 CEST] <cehoyos_> So I assume that either the issue is fixed or the patch does not fix it...
[11:30:56 CEST] <cehoyos_> What command line did you test?
[11:30:57 CEST] <Guest40634> hmmm, if it was applied, than probably I should somehow specify the gamma value for the exr loading.
[11:31:23 CEST] <Guest40634> Can you tell me how to do that?
[11:31:34 CEST] <cehoyos_> "-gamma 1.1" ?
[11:31:38 CEST] <cehoyos_> (I didn't test)
[11:31:58 CEST] <Guest40634> I will. If it does than I'm officially an idiot. :)
[11:32:01 CEST] <chama> @cehoyos_ : this is my sample www.filedropper.com/1_80
[11:32:16 CEST] <cehoyos_> chama: I am away now for an hour or so, will test later.
[11:32:33 CEST] <chama> ok thanks :)
[11:42:30 CEST] <Guest40634> cehoyos: It works indeed! Perfect, you've made my day! If you know a linear->srgb conversion too, instead of using gamma 2.2, than you also make my whole week. :)
[12:04:12 CEST] <cehoyos_> Guest40634: My problem is that I (still) don't know what "gamma" and "linear conversion" are;-(
[12:04:27 CEST] <cehoyos_> But I suspect the 3dlut filter can do what you need.
[12:11:23 CEST] <baadf00d> Hello, i'm trying to get QMediaPlayer, a Qt class to play a video stream. I'm using ffmpeg to send a video to ffserver and connect to it from the application I have made. My application can load local files encoded with ffmpeg using libx264 encoder. I can play the stream from ffserver via mplayer VLC and ffplay. But what fails is when I try to connect the QMediaPlayer instance to ffserver. On ffserver i can see a line containing [GE
[12:11:29 CEST] <baadf00d> When i try to connect to the stream
[12:12:10 CEST] <cehoyos_> chama: Sorry, I can reproduce your issue with every input...
[12:12:36 CEST] <cehoyos_> Workaround is: ffmpeg -i 1.mp3 -i 1.mp3 -i 1.mp3 -filter_complex [0:a][1:a][2:0]concat...
[12:12:55 CEST] <baadf00d> But the stream won't play. I'v looked all over for an answer to this, but no success. But finally I got an email from Qt, saying that on windows they use Windows Media Foundation as a backend. And pointed me to this table of compatible things.
[12:12:56 CEST] <baadf00d> https://msdn.microsoft.com/en-us/library/windows/desktop/dd757927(v=vs.85).aspx
[12:13:36 CEST] <baadf00d> But I coudln't make it work. Can someone please tell me what configurations should work with this?
[12:14:39 CEST] <chama> @cehoyos_ : that way it worked....
[12:14:53 CEST] <chama> is there any other alternative to loop an audio?
[12:16:13 CEST] <baadf00d> loop input?
[12:16:45 CEST] <baadf00d> how about: ffmpeg -f concat -i loop.txt ...
[12:18:19 CEST] <baadf00d> and in the file a bunch of lines consisting of : file '1.mp3'
[12:22:21 CEST] <cehoyos_> chama: I created ticket 4590, using the concat demuxer is of course also a workaround.
[12:22:35 CEST] <cehoyos_> For some files, using the concat protocol also works.
[12:24:34 CEST] <chama> thanks cehoyos_ :)
[13:57:14 CEST] <Guest40634> cehoyos: By linear to srgb conversion I was referring to using the srgb colorspace transformation that resembles gamma 2.2 but is slightly different at the "darker" regions. This is a valid source for the algorithm: http://prolost.com/blog/2005/11/16/srgb.html
[13:58:29 CEST] <Guest40634> cehoyos: By using gamma 2.2 for the openexr loader I got rid of the ugly banding artifacts, but there is still some color and value shift at low values.
[14:02:16 CEST] <cehoyos_> Guest40634: Can you reproduce this with a sample consisting of one solid colour?
[14:04:49 CEST] <Guest40634> cehoyos: With a single color I can only show the math difference between srgb conversion and gamma 2.2.
[14:06:36 CEST] <cehoyos_> So how can the colour shift be reproduced.
[14:06:38 CEST] <cehoyos_> ?
[14:08:12 CEST] <Guest40634> cehoyos: Even a single color changes hue differently with the two methods, because of the separate RGB channel if/else calculation of srgb.
[14:08:58 CEST] <cehoyos_> So it is correct that the methods produce different outputs?
[14:09:27 CEST] <Guest40634> cehoyos: If the image or the single color has different channel values and at least one of them is below the if/else "border".
[14:10:12 CEST] <Guest40634> cehoyos: Yes. If I do a gamma and an srgb conversion in a composite software for an image than the two is not identical, the difference shows lots of changes in the dark regions.
[14:12:29 CEST] <baadf00d> I just got an update from the guys at Qt. "At least RTSP should work, although it also has limitations as WMF supports only asf and mp3 over RTSP."
[14:12:56 CEST] <baadf00d> Now i'm confused, how do you configure that?
[14:38:51 CEST] <animax> hi all. https://trac.ffmpeg.org/wiki/Encode/H.264#AdditionalInformationTips does anyone know if this http://www.pasteall.org/pic/88976 is a mistake in writing? does the author mean cbr (green marked) level?
[14:39:42 CEST] <animax> because there is no crf parameter in the commands ...
[16:07:55 CEST] <anoop_r> hi
[16:08:44 CEST] <baadf00d> I'm starting to feel like i'm going insane, is this a normal user experience with ffmpeg?
[16:09:18 CEST] <anoop_r> i had tried to compile ffmpeg in mingw-w64 win64 shell
[16:09:54 CEST] <anoop_r> License: LGPL version 2.1 or later
[16:09:54 CEST] <anoop_r> Creating config.mak, config.h, and doc/config.texi...
[16:09:54 CEST] <anoop_r> ./configure: line 1311: cmp: command not found
[16:10:00 CEST] Last message repeated 2 time(s).
[16:10:18 CEST] <BtbN> your shell is lacking basic commands
[16:10:36 CEST] <BtbN> namely, cmp
[16:11:30 CEST] <anoop_r> did i have to set any path
[16:11:34 CEST] <anoop_r> i had used this ./configure --target-os=mingw32
[16:11:57 CEST] <Canar> no, it's as BtbN says
[16:12:15 CEST] <anoop_r> when only ./configure used error got Unknown OS
[16:12:24 CEST] <Canar> your mingw lacks basic programs needed for the compile like cmp
[16:12:37 CEST] <anoop_r> how to install cmp
[16:13:01 CEST] <Canar> consult mingw documentation
[16:13:51 CEST] <Canar> given how debian's support for mingw and cygwin is lately, you're probably even better off installing linux and cross-compiling for windows
[16:16:26 CEST] <baadf00d> I just made ffserver throw out a segmentation fault by connecting to it's rtsp stream from WMP! I think i'm done for today.. I hope my ffmpeg.org forum account gets its activation email soon so I can ask my questions there, they might be too hard for this channel
[16:23:51 CEST] <durandal_1707> Report ffserver bugs to tracker
[16:37:24 CEST] <t4nk103> Hello
[16:38:13 CEST] <t4nk103> i got a compilation probleme, hope someone help me.
[16:38:20 CEST] <t4nk103> (sorry for my bad english)
[16:38:46 CEST] <t4nk103> i try to build "motion" with ffmpeg support
[16:39:47 CEST] <t4nk103> i'm on xubuntu 14.04 (x64)
[16:40:15 CEST] <t4nk103> i build ffmpeg from source with compilation guide (https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu)
[16:40:25 CEST] <t4nk103> it work well
[16:41:02 CEST] <t4nk103> now i try to build motion with ffmpeg support like :
[16:41:04 CEST] <t4nk103> ./configure --with-ffmpeg=/home/romain/ffmpeg_build
[16:41:20 CEST] <t4nk103> and i got an error message
[16:41:43 CEST] <t4nk103>  /usr/bin/ld: /home/romain/ffmpeg_build/lib/libavcodec.a(libvorbisenc.o): référence au symbole non défini «vorbis_encode_setup_vbr»
[16:42:02 CEST] <t4nk103>  //usr/lib/x86_64-linux-gnu/libvorbisenc.so.2: error adding symbols: DSO missing from command line
[16:42:55 CEST] <t4nk103> so i tink it's a ffmpeg probleme (libavcodec.a need to know 'vorbis_encode_setup_vbr')
[16:43:29 CEST] <t4nk103> i dont know how to workarround this error, can someone help me please ?
[16:54:10 CEST] <c_14> You built that ffmpeg on the computer you're on right now?
[17:02:57 CEST] <t4nk103> yes
[17:03:24 CEST] <t4nk103> hello c_14, sorry for reply time, yes i'm on the conputer
[17:05:26 CEST] <c_14> does `ffmpeg -f lavfi -i aevalsrc=0 -c:a libvorbis -f ogg -y /dev/null' execute successfully or does it throw the same error?
[17:06:55 CEST] <t4nk103> i try...
[17:07:45 CEST] <c_14> Oh, and make sure you execute the right binary. If you followed that guide, it should be in ~/bin
[17:08:15 CEST] <c_14> I want to see if the problem is reproducible using only ffmpeg
[17:09:38 CEST] <t4nk103> No probleme with you command, this seem to work
[17:10:03 CEST] <t4nk103> yes is the good executable, my ~/bin is at the head of my path
[17:10:11 CEST] <t4nk103> ffmpeg -h say
[17:10:22 CEST] <t4nk103> ....configuration: --prefix=/home/romain/ffmpeg_build....
[17:11:05 CEST] <t4nk103> ffmpeg version 2.6 Copyright (c) 2000-2015 the FFmpeg developers   built with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1)   configuration: --prefix=/home/romain/ffmpeg_build....
[17:15:52 CEST] <t4nk103> 'ffmpeg -f lavfi -i aevalsrc=0 -c:a libvorbis -f ogg -y /dev/null' work successfully
[17:16:05 CEST] <diroots> hi all. as tcpsyphon use a tcp stream based on jpeg 2000, is there any way we can stream from a linux box using ffmpeg to a osx box using tcpsyphon?
[17:16:44 CEST] <diroots> Any one of you did hear about that?
[17:18:53 CEST] <c_14> t4nk103: I think you need to add -lvorbis to the linker arguments
[17:21:17 CEST] <t4nk103> ok i'm going to hack the motion makefile
[17:21:27 CEST] <t4nk103> i try now
[17:21:50 CEST] <t4nk103> many thanks for give me a way c_14
[17:22:00 CEST] <anoop_r> i had compiled ffmpeg using msys2 64bit shell please see http://pastebin.com/Bz0tva3e
[17:23:09 CEST] <anoop_r> i got several warnings . how to get rid of it
[17:23:39 CEST] <anoop_r> i had used ffmpeg-20150531-git-11aa050.tar.xz for compiling
[17:24:25 CEST] <t4nk103> anoop_r did you read your warning ?
[17:24:50 CEST] <anoop_r> yes
[17:25:08 CEST] <t4nk103> and what did you understand ?
[17:25:18 CEST] <anoop_r> but i am not great programmer
[17:25:27 CEST] <c_14> anoop_r: either ignore them, remove the warning flags from the compiler flags, or fix the code
[17:25:43 CEST] <t4nk103> did you are english natural user ?
[17:25:55 CEST] <anoop_r> no
[17:26:07 CEST] <anoop_r> sorry for my bad english
[17:26:18 CEST] <t4nk103> i'm not english natural speaker
[17:26:28 CEST] <t4nk103> sorry for my bad english too
[17:26:38 CEST] <t4nk103> but did you understand deprecated ?
[17:26:41 CEST] <anoop_r> so these warnings will not make any problems ?
[17:26:59 CEST] <anoop_r> ya legendary function
[17:27:04 CEST] <anoop_r> sort of
[17:27:20 CEST] <c_14> Nah, as long as there are no errors everything's fine.
[17:28:25 CEST] <anoop_r> one doubt . I had used the latest ffmpeg source and still it contains deprecated functions
[17:28:48 CEST] <t4nk103> yes code is constantly moving forward
[17:28:53 CEST] <c_14> Because nobody's gone around and replaced them with the non-deprecated versions.
[17:28:55 CEST] <c_14> patches welcome
[17:29:06 CEST] <t4nk103> :)
[17:29:14 CEST] <anoop_r> i will try
[17:29:48 CEST] <anoop_r> i had used ./configure --target-os=mingw32
[17:30:07 CEST] <anoop_r> with mingw64bit shell
[17:30:10 CEST] <anoop_r> is it ok
[17:30:20 CEST] <t4nk103> deprecated functions are not a probleme
[17:31:01 CEST] <anoop_r> when i run ffmpeg it says libbz2-1.dll not found
[17:32:39 CEST] <c_14> Are you executing it in the same shell you compiled it in?
[17:32:57 CEST] <c_14> Because it found the library during compilation or it would have errored while linking.
[17:35:28 CEST] <anoop_r> yes
[17:35:49 CEST] <anoop_r> so i need more configuration paramets
[17:53:19 CEST] <t4nk103> i dont know how to tell windows to look at dll
[17:53:22 CEST] <t4nk103> directory
[17:53:52 CEST] <t4nk103> do you see libbz2-1.dll anywhere ?
[17:59:51 CEST] <anoop_r> no
[18:03:34 CEST] <t4nk103> c_14 i try to add '-lvorbis' to the linker argument bu without success : http://pastebin.com/2HRmgRbz
[18:07:43 CEST] <t4nk103> Yip ha
[18:08:15 CEST] <anoop_r> i got
[18:08:16 CEST] <t4nk103> i use pkg-config to get the libs
[18:08:42 CEST] <anoop_r> how to integrate all these dll into ffmpeg
[18:09:03 CEST] <t4nk103> first i export the PKG_CONFIG_PATH to my installation
[18:09:05 CEST] <t4nk103> export PKG_CONFIG_PATH=/home/romain/ffmpeg_build/lib/pkgconfig
[18:09:37 CEST] <t4nk103> and after
[18:09:38 CEST] <t4nk103> pkg-config --libs libavformat libavcodec libavutil
[18:09:55 CEST] <t4nk103> this take me a long lib line
[18:09:56 CEST] <t4nk103> -lavformat -lavcodec -lva-x11 -lva -lxcb-shm -lxcb-xfixes -lxcb-render -lxcb-shape -lxcb -lXau -lXdmcp -lSDL -lasound -lpulse-simple -lpulse -lX11 -lXext -lcaca -lx265 -lstdc++ -lrt -lx264 -lpthread -ldl -lvpx -lvorbisenc -lvorbis -ltheoraenc -ltheoradec -logg -lopus -lmp3lame -lfdk-aac -lass -lfontconfig -lenca -lfribidi -lexpat -lfreetype -lpng12 -lz -lswresample -lavutil -lm
[18:10:20 CEST] <t4nk103> i put them into "motion" makefile
[18:10:31 CEST] <t4nk103> and now it's build successfully
[18:10:45 CEST] <t4nk103> thanks c_14 to point me the way
[18:12:13 CEST] <t4nk103> anoop_r : try to temporary put the dll in the same forlder of your application
[18:12:40 CEST] <t4nk103> (i dont know how to tel Windows to seach into a special directory for dll
[18:13:09 CEST] <t4nk103> perhaps you can tell with a special arg in ./configure...
[18:13:26 CEST] <anoop_r> ok
[18:15:54 CEST] <t4nk103> what tuto did you use to build ffmpeg with mingw ?
[18:18:28 CEST] <anoop_r> i want to build ffmpeg with x265 for my academic project
[18:18:46 CEST] <anoop_r> it a transcoder application
[18:19:17 CEST] <anoop_r> i tried to do it with msvc but failed
[18:19:47 CEST] <t4nk103> did you read this : https://trac.ffmpeg.org/wiki/CompilationGuide
[18:20:28 CEST] <anoop_r> ya
[18:33:57 CEST] <mr_test> hi
[18:34:08 CEST] <anoop_r_> hi
[18:35:09 CEST] <mr_test> i'm having an issue converting a video to h264. ffmpeg libx64 is way slower than x264, even with ultrafast preset. it seems ffmpeg is not using my cpu fully (~25%) while x64 uses more (~55%)
[18:35:54 CEST] <anoop_r_> i not an expert
[18:36:19 CEST] <anoop_r_> but i think ffmpeg is not compiled with enough threads
[18:36:44 CEST] <mr_test> anoop_r_: how can i change that? what flag is necessary when compiling?
[18:37:12 CEST] <anoop_r_> enabel pthread or w32threads
[18:37:19 CEST] <anoop_r_> enable
[18:38:27 CEST] <mr_test> you're talking about process threads right? because ffmpeg doesn't even use a single core fully
[18:39:11 CEST] <mr_test> so i doubt multithreading would resolve this
[18:40:12 CEST] <t4nk103> did you test "-thread" arg ?
[18:40:23 CEST] <t4nk103> (probably he wait for I/O)
[18:40:51 CEST] <mr_test> t4nk103: i tried -threads 0 yes
[18:42:52 CEST] <anoop_r_> try -threads 8
[18:44:49 CEST] <mr_test> anoop_r_: no change
[18:46:07 CEST] <anoop_r_> ffmpeg -i somefile.wmv -c:a libfdk_aac -c:v libx264  -threads 1 transcoded.mp4
[18:47:26 CEST] <mr_test> anoop_r_: chag
[18:47:41 CEST] <mr_test> changed libfdk_aac to aac
[18:47:45 CEST] <mr_test> still low fps
[18:48:01 CEST] <anoop_r_> i mean -threads 1
[18:48:01 CEST] <mr_test> and added -strict_experimental
[18:48:23 CEST] <mr_test> yep i used -threads 1
[18:49:20 CEST] <anoop_r_> do you know how to build ffmpeg
[18:49:57 CEST] <mr_test> anoop_r_: as usual? configure;make?
[18:50:40 CEST] <anoop_r_> try -- enable-pthreads
[18:50:55 CEST] <anoop_r_> i don't know is it effective
[18:50:57 CEST] <anoop_r_> but try
[18:51:14 CEST] <anoop_r_> try -ss
[18:51:18 CEST] <anoop_r_> in encoding
[18:51:27 CEST] <mr_test> anoop_r_: should i use git or 2.6.3?
[18:52:18 CEST] <klaxa> when possible use git
[18:52:20 CEST] <mr_test> with -ss it doesn't start transcoding
[18:52:48 CEST] <klaxa> depending on the position of -ss, ffmpeg will do fast or slow seeking
[18:54:08 CEST] <klaxa> if passed as an input option (i.e. ffmpeg -ss 10 -i somefile.mkv -c copy outfile.mkv) it will do inaccurate fast seeking, when used as an output option (i.e. ffmpeg -i somefile.mkv -c copy -ss 10 outfile.mkv) it will do accurate slow seeking which decodes all frames up to the point where you wanted to seek to
[18:54:31 CEST] <klaxa> you can also combine both, note that the amount of time is added, see also https://trac.ffmpeg.org/wiki/Seeking
[18:56:04 CEST] <mr_test> used it as output option, again doesn't start encoding
[18:57:24 CEST] <mr_test> but i doubt -ss would change anything anyways?
[18:57:35 CEST] <mr_test> about my low fps
[18:57:40 CEST] <klaxa> no, not really
[18:57:59 CEST] <klaxa> can you pastebin your full x264 command and your full ffmpeg command? both preferably with output included
[19:04:40 CEST] <mr_test> klaxa: ffmpeg -re -y -loglevel fatal -i input.flv -filter_complex scale=150:100 -map 0:v -map 0:a:0 -g 52 -c:v libx264 -preset ultrafast -c:a aac -strict experimental -v verbose -threads 8 -f mp4 output.mp4
[19:05:54 CEST] <klaxa> well how many fps do you get? the same amount input.flv has? because you are using -re to emulate real-time decoding
[19:06:12 CEST] <mr_test> oh alright :)
[19:06:53 CEST] <mr_test> klaxa: ~20fps
[19:07:02 CEST] <mr_test> lower than input
[19:07:46 CEST] <klaxa> what if you remove -re?
[19:10:34 CEST] <mr_test> normal fps, comparable to x264 then :)
[19:13:21 CEST] <mr_test2> i'm trying to play a video in html5 transcoded in realtime by mediatomb
[19:13:37 CEST] <mr_test2> however it lags when playing the video :/
[19:13:48 CEST] <mr_test2> maybe it's because of -re?
[19:22:10 CEST] <klaxa> hmm dunno
[19:39:39 CEST] <lilibox> hi
[19:40:12 CEST] <lilibox> please, is any chance to get the latest static build of ffmpeg for centos 5 - 2.6.18?
[19:52:21 CEST] <klaxa> 2.6.18 kernel? isn't that a bit very old?
[19:52:29 CEST] <klaxa> http://johnvansickle.com/ffmpeg/
[19:52:40 CEST] <klaxa> these static builds are for 2.6.32 and above
[20:02:55 CEST] <anoop_r_> why no body use cmake + msvc to build ffmpeg like x265 ?
[20:03:26 CEST] <JodaZ> and they really don't work for older kernels, klaxa?
[20:03:45 CEST] <klaxa> i don't know
[20:04:11 CEST] <JodaZ> can't imagine what ffmpeg could have that'd interfere
[20:04:32 CEST] <klaxa> well if you feel nostalgic, you can set up a 2.6.18 machine and try it out ;)
[20:06:48 CEST] <anoop_r_> can i change the output result of ffmpeg ?
[21:30:15 CEST] <kikobyte> Hi everybody, I'm looking into the mxf demuxer of libavformat, wondering what's the best way to get (in a client code) a piece of demuxer-private data like picture essence coding, etc. Can anybody help me?
[21:57:12 CEST] <user3123> Hi all
[21:57:22 CEST] <user3123> My HLS sttrean has 4-5 programs eatch containing streams with different widthxheight. ffmpeg automaticaly select the best (big) widthxheight but I want it to select the program (stream) with the 640x360 stream
[21:58:04 CEST] <user3123> If I do not know what program or stream number has the width 640 and height 360, how can I tell ffmpeg to select the program or stream that has the 640x360 because it default select the best stream.
[22:04:34 CEST] <lilibox> back
[22:04:57 CEST] <lilibox> i know, it is old but i kept stuck on this version from serious reasons
[22:05:05 CEST] <lilibox> soo, is chance to make it though?
[22:05:39 CEST] <user3123> what?
[22:07:04 CEST] <user3123> My HLS sttrean has 4-5 programs eatch containing streams with different widthxheight. ffmpeg automaticaly select the best (big) widthxheight but I want it to select the program (stream) with the 640x360 stream  If I do not know what program or stream number has the width 640 and height 360, how can I tell ffmpeg to select the program or stream that has the 640x360 because it default select the best stream.
[22:12:07 CEST] <llogan> user3123: you will probably have to script it
[22:12:23 CEST] <llogan> using ffprobe to get the stream info
[22:15:01 CEST] <anoop_r> can any one explain this " bit depth is not bit per pixel"
[22:16:55 CEST] <kikobyte> anoop_r: I think it means bits per component (per channel, e.g. R/G/B or Y/Cb/Cr)
[22:18:35 CEST] <anoop_r> i am confuse for example there is x265 16bpp preset .but if i encode video using x265 16bpp and check it with media info it only shows 10bit video
[22:18:51 CEST] <anoop_r> i thought it will give 16bit video
[22:22:07 CEST] <kikobyte> anoop_r: 16bpp = 16 bits per pixel, it can be, like, 4:2:2 subsampled 8-bit YCbCr; while 10bit video, depending on subsampling, may be 30 (for 4:4:4), 20 (for 4:2:2) or 12 (for 4:2:0) bits per pixel.
[22:22:58 CEST] <anoop_r> so nobody yet have a 16bit video
[22:23:27 CEST] <BtbN> What would be the point of 16 bit color components?
[22:23:40 CEST] <BtbN> 8 bit usualy is enough, 10 bit for special usecases.
[22:24:01 CEST] <anoop_r> it needs when you have 8k video
[22:24:08 CEST] <BtbN> what?
[22:24:11 CEST] <kikobyte> anoop_r: why not, there is a lot of 16-bit media shot by cameras before it goes to production
[22:25:02 CEST] <BtbN> Why would the resolution matter at all?
[22:25:54 CEST] <anoop_r> see x265 include 16bit video in its profile
[22:26:02 CEST] <kikobyte> BtbN: There are 12-bit codecs already on the field. The key point in increasing bit depth is to provide wider dynamic range
[22:26:30 CEST] <anoop_r> better compression is also possibel
[22:26:48 CEST] <BtbN> better compression if each pixel is twice as big in bit-size?
[22:27:21 CEST] <anoop_r> that's what x265 algorithm does
[22:28:10 CEST] <BtbN> You have any benchmarks/tests about that? Cause on a first glance, it sounds plain wrong. It's much more information it has to store and process.
[22:28:40 CEST] <kepstin-laptop> x264, when using 10bit encoding, can get better compression efficiency in some cases with predicted frames because there's less precision loss over generations of frames.
[22:32:31 CEST] <kikobyte> anoop_r: what's the full name of the profile? I see "Main", "Main 4:2:2 10", "Main 4:4:4 16 Intra" and others, which one are you referring to?
[22:36:45 CEST] <anoop_r> anyway to filter ffmpeg output besides using -loglevel
[22:39:36 CEST] <kikobyte> kepstin-laptop: that's weird. I can believe in precision loss due to low motion vector accuracy (e.g 1/4 of pixel), but what's the difference in predicting 8-bit value using another 8-bit value and the same with 10-bit?
[22:43:01 CEST] <ChocolateArmpits> Is anyone familiar with MXF standard ?
[22:43:26 CEST] <kikobyte> a bit
[22:44:39 CEST] <ChocolateArmpits> How should I exactly define my own universal label key to use in the header metadata set ?
[22:45:08 CEST] <kikobyte> You can't.
[22:45:20 CEST] <kikobyte> It is managed by SMPTE
[22:45:25 CEST] <ChocolateArmpits> Well that part I know
[22:47:29 CEST] <ChocolateArmpits> So I should stick to a standard like Ebucore and use the key values provided by that or would it be safe to produce my own eyes through comparison with already existing ones ?
[22:47:38 CEST] <ChocolateArmpits> keys*
[22:47:41 CEST] <ChocolateArmpits> not eyes sorry
[22:48:25 CEST] <ChocolateArmpits> I guess the latter would be violating the spec
[22:49:30 CEST] <ChocolateArmpits> The file based on standard should still be playable. The file might get misinterpreted if the key ever matches a registered UL
[22:50:44 CEST] <kikobyte> Exactly. The labels are strictly defined, and the bytes do have their own meanings. In other words, not even matching to a whole registered UL might harm, but to a part as well.
[22:51:49 CEST] <ChocolateArmpits> Hmm I found an "Experimental UL"
[22:51:52 CEST] <ChocolateArmpits> "SMPTE Label identifiers used by Organizations for experimental use. Users may create structures consistant with their own specifications"
[22:52:12 CEST] <anoop_r> how to encode in 10 bit using ffmpeg libx265
[22:53:10 CEST] <ChocolateArmpits> The list of ULs that I'm refering to is dated 2008. Do you know where I can find any newer public lists, kikobyte ?
[22:56:34 CEST] <kikobyte> ChocolateArmpits: http://www.smpte-ra.org/mdd/
[22:57:08 CEST] <ChocolateArmpits> kikobyte: Oh major thanks, I'll try digging in
[22:58:34 CEST] <TheSchaf> http://www.npr.org/sections/therecord/2015/06/02/411473508/how-well-can-you-hear-audio-quality/ <- interesting, what do you guys score? i got one right, 4x 320 kbit/s and 1x 128 kbit/s :D
[22:58:55 CEST] <kikobyte> ChocolateArmpits: still I'm not sure that's the registry you need... Do you want to pack your own metadata?
[23:00:49 CEST] <ChocolateArmpits> kikobyte: Yes, an xml structured text to later on use as a reference.
[23:01:51 CEST] <ChocolateArmpits> I've already observed that with Sony's XDCAM files, they stick in XML data, so just wanted to know about the keys. What's legal what's not
[23:07:18 CEST] <kikobyte> Well, I'm not really aware of all the details, but what I see from wiki is that you need to use 380m, RP210 and RP224 - the last one can be found on that smpte-ra.org/mdd page
[23:08:12 CEST] <kikobyte> and actually, both RP210 and RP224 are there
[00:00:00 CEST] --- Thu Jun  4 2015