[Ffmpeg-devel-irc] ffmpeg.log.20130219

burek burek021 at gmail.com
Wed Feb 20 02:05:01 CET 2013


[01:00] <teratorn> is command like this possible? ffplay -f rawvideo -pixel_format yuv420p -s 352x288 '/tmp/%04d.yuv420p.raw'
[01:24] <dionoea>  /win 0
[03:45] <relaxed> teratorn: I think only the image2 format can use printf's %d naming as input.
[03:46] <relaxed> but, you can cat raw yuv in as input.
[03:49] <defaultro> hey guys, how can we mimic a warm filter?
[03:50] <defaultro> it's different from hue filter
[03:51] <defaultro> or maybe another solution is to superimpose a png image with 50% or less for alpha
[03:51] <defaultro> is that possible?
[08:53] <mista_D> anyway to run libass on same file using two srt files (eng and spa)? without retranscoding twice?
[09:14] <juanfont> hi. Is there any way to reduce max_analyze_duration at runtime? or discard those 5M of data? I'm trying to reduce the lag between server and client as much as I can
[09:14] <juanfont> thanks :)
[09:29] <eli___> hello
[09:33] <eeli_> hello
[09:33] <eeli_> would you please help me `md?
[09:34] <eeli_> what can i do for can not chatting freenode channel?
[10:52] <jjido> Hello, I learnt that .VOB files in a VIDEO_TS folder are MPEG2. Do you know if it is possible to replace the video and audio streams in it to occupy less space? I want to do what DVDShrink does to fit a 2H DVD on a 4.7GB disk
[11:51] <invariant> I have a wav file which is the result of a cdrip. I want to turn that in a sequence of flac files. How can that be done?
[11:53] <relaxed> you need to rip the cd into a sequence of wav files
[11:55] <invariant> relaxed, alternatively, how can I make the wav file available as a rippable device?
[11:56] <invariant> I don't get why all the rippers work on a device file myself.
[11:56] <relaxed> Does it have a cue sheet?
[11:59] <invariant> relaxed, yes
[11:59] <invariant> relaxed, I have a cue file.
[12:00] <relaxed> Then you need software to extract the tracks using that file. ffmpeg can't do it.
[12:01] <durandal_1707> ffmpeg can do it but not in friendly way like some tools designed to do just that
[12:02] <relaxed> invariant: mplayer2 has cue://$file support, use that with -dumpstream to extract a wav for each track.
[12:04] <durandal_1707> http://www.etree.org/shnutils/shntool/
[12:04] <durandal_1707> it have cue mode
[12:05] <relaxed> real men use mplayer
[12:05] Action: relaxed flexes
[12:15] <durandal_1707> this one can create cue files
[12:41] <Aktau> Hey guys. I've been using ffmpeg to decode videos very well, but I'd like to know if it does (maybe through libavcodec) support HW accelerated decode through VA-API on linux?
[12:41] <Aktau> Does anyone know about that?
[14:00] <JEEB> WmA, sending me private messages for nothing that needs to be private or is specifically related to me will only lead to /ignore :P
[15:35] <otters> being a total noob, what option do I pass to map_metadata to convert a FLAC to Apple Lossless and keep per-track metadata intact (like album name etc.)
[15:37] <durandal_1707> they are usually completly copied and encoder is added
[15:38] <otters> so adding map_metadata is redundant?
[15:38] <durandal_1707> dunno, did you tried it ? does it work as expected?
[15:38] <otters> yeah, I tried it, and the tracks don't have *any* metadata when I open them in itunes
[15:39] <otters> oh, hey, yeah, omitting the option works.
[15:40] <otters> I must be some sort of moron
[15:46] <otters> thanks everyone
[15:53] <e2b> Is there any plan to support the Raspberry Pi GNU-acceleration for en-/decoding? The CPU itself is pretty slow, but thanks to the GPU it can already play HD videos witch optimized players.
[15:55] <durandal_1707> no plans, people usually come with patch that implements such funcionality - that is how FOSS works
[15:58] <divVerent> if "nm -a binary" shows an entry:
[15:58] <divVerent>   000000010087ca70 T _avcodec_default_execute
[15:58] <divVerent> does that mean I have the right to the source code of the binary? ;)
[15:59] <divVerent> I found this in "iTap Mobile RDP.app" for Mac OS X
[15:59] <divVerent> *hint* want to bring that Hall of Shame back?
[16:00] <durandal_1707> only if you have enough money to sue them
[16:00] <divVerent> putting them on the HoS costs no money :P
[16:00] <durandal_1707> and pay lawyers and etc
[16:00] <divVerent> also, they violated none of MY copyright
[16:01] <divVerent> so I wouldn't have a case there
[16:01] <divVerent> but I mainly am wondering, did I read the "nm" output right and this is not an external reference?
[16:01] <divVerent> I THINK I did
[16:02] <divVerent> also, can I somehoe figure out if they used ffmpeg or libav?
[16:02] <divVerent> ah, it's PROBABLY libav
[16:02] <divVerent> I see an URL with upload.libav.org
[16:38] <MIG-> e2b: raspbian includes source code for x264 encoding using the gpu
[16:42] <jeje34> Hi to all ;-)
[16:43] <MIG-> hello
[16:44] <jeje34> I'm want to use FFMPEG library in a windows application to decode H264 video streams from camera. I have a question about the memalign-hack for the configure compilation. I'm not sure if it needed on Windows OS
[16:46] <JEEB> if it is enabled by default in the configuration for your way of building for windows on the current ffmpeg, then it is either needed or not
[16:46] <JEEB> you shouldn't have to set it manually
[16:46] <jeje34> yes but in the configure file there: --enable-memalign-hack   emulate memalign, interferes with memory debuggers
[16:47] <jeje34> So if I use Visual Studio VC++ debuggers, it can make me memory problems?
[16:47] <JEEB> I repeat
[16:47] <JEEB> you shouldn't have to enable it manually
[16:48] <JEEB> configure should look it up by the way you are compiling it and for which arch
[16:48] <jeje34> yes but you tell me it is enables by default
[16:48] <JEEB> DEPENDS ON THE CONFIGURATION AND WHAT YOU ARE BUILDING WITH
[16:48] <JEEB> AS I SAID
[16:48] <jeje34> I use msys and mingw to compile FFMPEG on my windows
[16:48] <JEEB> look at the gawd damn configure script's output
[16:48] <JEEB> seriously
[16:49] <JEEB> and it shouldn't make the thing any worse debug-wise
[16:49] <JEEB> if it is needed, it is needed
[16:49] <JEEB> if it isn't, it isn't
[16:49] <JEEB> that's it
[16:49] <JEEB> if you are building with mingw MSVS's debugger won't be able to load the damn debug symbols anyways, if you are building with MSVC then you are getting the debug symbols if you enable debug
[16:50] <jeje34> I use Visual Studio to make my application, but I use the library of FFMPEG I built and don't try to debug "inside" FFMPEG
[16:51] <jeje34> just use debugger to debug my application
[16:52] <JEEB> go over what I said, the answers to your question(s) are there
[17:14] <ozone89> hello :)
[17:21] <jeje34> I'm sorry but there is something I can't explain in my code. I have an AVPacket m_lpPacket and an AVPicture m_lpPict. In order, I call m_lpPict.data[0] and after decoding I use avpicture_alloc( &m_lpPict, AV_PIX_FMT_RGB32, dwWidth, dwHeight); to use swscale to convert my YUV420 to RGB32. But at the next frame, when I call av_init_packet(&m_lpPacket), m_lpPict.data[0] become NULL.
[17:41] <mpfundstein> jeje34: didnt know people still use hungarian notation
[17:44] <jeje34> I find it more clear when writing classes
[17:58] <fenduru> How can I avoid artifacts between segments when segmenting an audio file?
[17:59] <fenduru> I need to be able to play them back sequentially without gaps/clicks between them
[18:36] <sabton> hey guys, just got caught up and built for Android at head release/1.1, after making necessary adjustments for deprecated API everything is running stable... except certain codecs (ogg/mp3) are decoding to garbled audio; however FLAC for example is decoding perfect
[18:36] <sabton> does this immediately indicate anything to anyone?
[18:48] <JEEB> sabton, sounds like you're expecting the decoder to output interleaved int while it is giving you planar float?
[18:48] <JEEB> although if flac works then it's the float vs int difference
[18:48] <JEEB> most lossy decoders were switched to output float by default, as that's how it basically gets encoded
[18:49] <JEEB> lossless stuff like flac naturally should output exactly what is put in
[18:49] <durandal_1707> flac did not changed sample format
[18:49] <JEEB> yes
[18:49] <durandal_1707> so it is GrEaT sample format switch
[18:49] <JEEB> I already noted it
[18:49] <JEEB> if you didn't see my second line :P
[18:50] <JEEB> and then the third+fourth noting the fact that the sample format was changed for most lossy formats
[18:50] <JEEB> sabton, ffmpeg does provide you a library to then convert the float audio to integer audio that you probably want
[18:51] <durandal_1707> and from planar to interleaved
[18:51] <JEEB> yup, if that is needed
[18:53] <sabton> ooh
[18:55] <sabton> that sure does explain things
[18:55] <sabton> the sound of my garbled audio is consistent with that explanation
[18:56] <sabton> so in addition to converting the samples, i guess the codecs will tell me what sample format is used?
[18:56] <sabton> i would need to find that API as well
[18:56] <JEEB> yes, I would think the avframe tells you what the output format is
[18:57] <sabton> ah, thank you
[18:57] <sabton> man, im playing with compiler settings and such.. =\
[18:57] <sabton> thanks :)
[19:13] <Dmole> is there a repo rpm with a recent ffmpeg with all libs enabled or is everyone compiling for some reason?
[19:17] <sophie_> hi
[19:17] <Dmole> what's p?
[19:17] <sophie_> do you know it's possible to run ffmpeg with ubuntu 11.04 ?
[19:18] <Dmole> ye it is.
[19:18] <Dmole> *yes
[19:18] <sophie_> i have a problem with yasm
[19:18] <sophie_> i have difficulties to upgrade it
[19:19] <JEEB> there shouldn't really be anything hard in upgrading it
[19:19] <JEEB> build it and put it somewhere in your PATH
[19:20] <Dmole> might need to purge from apt before did you try compiling yasm?
[19:20] <sophie_> yes
[19:20] <Dmole> so what exactly is the issue?
[19:20] <sophie_> terminal tell me :
[19:20] <sophie_> Found yasm 1.1.0.2352
[19:20] <sophie_> Minimum version is yasm-1.2.0
[19:20] <sophie_> If you really want to compile without asm, configure with --disable-asm.
[19:21] <JEEB> that just means that you have 1.1.0 installed
[19:21] <sophie_> but i've already done apt-get upgrade yasm
[19:21] <JEEB> yes, because the repositories don't have 1.2.0
[19:21] <JEEB> you have to grab the tarball
[19:21] <JEEB> extract it
[19:21] <sophie_> so, i can also unabled i ?
[19:21] <sophie_> it
[19:21] <JEEB> what
[19:21] <Dmole> disable
[19:22] <JEEB> no, don't disable asm
[19:22] <JEEB> that will be SLOW
[19:22] <sophie_> ok
[19:22] <sophie_> the tarball ?
[19:22] <JEEB> http://www.tortall.net/projects/yasm/releases/yasm-1.2.0.tar.gz
[19:22] <JEEB> wget that
[19:22] <JEEB> tar xvf yasm[tab]
[19:22] <JEEB> cd yasm-[tab]
[19:23] <JEEB> do you want it to be available to everyone in the system automatically?
[19:23] <JEEB> or are you OK with a local install?
[19:23] <sophie_> local install it's ok
[19:24] <JEEB> ./configure --prefix=${HOME}/ownapps
[19:24] <JEEB> make
[19:24] <JEEB> make install (no sudo needed)
[19:24] <sophie_> ok, i try
[19:24] <JEEB> export PATH=${HOME}/ownapps/bin:${PATH}
[19:24] <JEEB> then try what yasm --version gives
[19:25] <JEEB> if that works, then you can add that thing into your dot-bashrc file
[19:25] <JEEB> so that directory gets added to the PATH every time
[19:35] <sophie_> thank you JEEB
[19:35] <sophie_> it has worked!
[19:41] <sophie_> JEEB what do you mean by "add that thing into your  dot-bashrc file"
[19:41] <sophie_> ?
[19:41] <sophie_> you mean these phrase :
[19:41] <sophie_> export PATH=${HOME}/ownapps/bin:${PATH}
[19:42] <JEEB> yes, that thing. dot-bashrc is a file that gets loaded every time bash gets loaded :P
[19:42] <JEEB> basically so you don't have to run it every time you start a terminal
[19:43] <JEEB> manually
[19:43] <sophie_> and where it is ?
[19:44] <JEEB> /home/your_username/ is where .bashrc should reside
[19:44] <JEEB> (that file name is dot-bashrc)
[19:45] <JEEB> although you generally should have one I guess
[19:45] <sophie_> when i want to ./configure ffmpeg
[19:45] <sophie_> it tells me :
[19:45] <sophie_> :~/ffmpeg-source/x264$ ./configure
[19:45] <sophie_> ./configure: ligne 241: rm : commande introuvable
[19:45] <sophie_> ./configure: ligne 243: dirname : commande introuvable
[19:45] <sophie_> ./configure: ligne 245: grep : commande introuvable
[19:45] <JEEB> uh-oh
[19:45] <JEEB> you did something wrong :P
[19:45] <sophie_> Unknown system , edit the configure
[19:45] <sophie_> yes
[19:45] <sophie_> i know
[19:45] <JEEB> your original ${PATH} got derped
[19:46] <sophie_> i just pick some command on the net
[19:46] <sophie_> when it wasn't good
[19:46] <sophie_> what can i do ?
[19:46] <sophie_> reinstall my all system ?
[19:47] <JEEB> no...
[19:47] <JEEB> you can just close the terminal and open a new one :P
[19:47] <sophie_> :)
[19:49] <sophie_> i was on this site :
[19:49] <sophie_> http://buildall.wordpress.com/2011/05/09/how-to-compile-and-install-ffmpeg-in-ubuntu-11-04/
[19:50] <sophie_> perhap's it will be more easy by installing new version of ubuntu ?
[19:50] <JEEB> no
[19:50] <JEEB> also that "guide" is a few years old :P
[19:51] <JEEB> https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide use this if you want to use some damn guide :P
[19:51] <JEEB> also this installs yasm on the whole system
[19:52] <sophie_> yes, it's ok too
[19:53] <sophie_>  This guide supports Ubuntu 12.10 (Quantal Quetzal), Ubuntu 12.04 (Precise Pangolin), and Ubuntu 11.10 (Oneiric Ocelot).
[19:54] <sophie_> but i have version 11.04
[19:54] <sophie_> natty
[19:54] <JEEB> shouldn't matter too much in most cases
[19:54] <sophie_> it will work either ?
[19:54] <sophie_> ok
[19:54] <JEEB> there are some things you actually will have to think about, yes
[19:54] <JEEB> now you're not afraid of thinking, right?
[19:55] <JEEB> just try to understand what the fuck is going on instead of mindless copy-pasting shit and you will be effing fine
[19:56] <sophie_> yeah!
[19:56] <sophie_> you know how to speak my godness!
[20:00] <JEEB> Do pardon me on my french, but I was kind of trying to think how to explain some very basic *nix things to you, but ended up just not thinking about it because it made my head hurt, and I'm not here to teach basics of *nix
[20:12] <Dmole> or if you don't want to think stick to apt
[20:29] <reactormonk> how do I drop the video stream? -vcodec null doesn't work
[20:30] <JEEB> if you want no video at all in the output, -vn
[20:30] <reactormonk> thanks
[20:45] <WmA> can anyone tell me what I am doing wrong please?
[20:45] <WmA> http://pastebin.com/zZpSwrEq
[20:46] <WmA> I would like to pipe a http ts stream into libx264
[20:46] <WmA> to record it straight to mp4
[20:49] <relaxed> WmA: why wouldn't you use the stream as ffmpeg's input?
[20:58] <stephanedev> hi everybody. i wrote a program using ffmpeg C API which streams a video encoded with vp8 over rtp. when streaming the video with ffmpeg and playing it with ffplay and simulating a 5% datagram loss, the video still plays ok and most frames are not too degraded.
[20:59] <stephanedev> but when using my program, it works fine if no datagram is lost, but with a 5% loss, ffplay shows a much more degraded video with most frames showing big pixel blocks or sometimes unrecognizable images
[21:00] <stephanedev> of course i am not surprised that ffmpeg is doing a much better job than my program, but i would like to understand what it does to achieve that
[21:01] <stephanedev> i tried to play with gop size, bitrate and other settings, but to no avail
[21:04] <stephanedev> does anyone have any hint on what i could try to improve my streaming code?
[21:10] <stephanedev> is there any parameter in AVCodecContext or AVFormatContext which would help?
[22:15] <vulkanr> hello. need some help with libavfilter. anyone in the house?
[22:15] <saste> vulkanr, don't ask to ask, just ask
[22:16] <vulkanr> trying to transcode video and pass the raw frames though libavfilter. i am getting really bad quality output. i think i am doing something wrong with the frame pts.
[22:17] <vulkanr> looking at the output file - the frames seem to be ordered by the pts which is weird and not the case when not using libavfilter
[22:19] <vulkanr> i followed the example code and am converting the pts between timebases as showed
[22:34] <ezekiel> in recent versions of ffmpeg, has the "lowpass"(http://ffmpeg.org/ffmpeg-filters.html#lowpass) filter been removed? Or perhaps replaced with something better?
[22:35] <saste> ezekiel, it has been added one week ago or so
[22:35] <ezekiel> okay, so only master ?
[22:35] <saste> yes
[22:35] <ezekiel> saste: much appreciated!
[22:36] <ezekiel> I've been building from git, but I've had to step back a few versions to achieve good audio+video sync from v4l2
[22:36] <saste> ezekiel, how old? we recently fixed a few v4l2 issues
[22:36] <ezekiel> 1.0.4 works great - 1.1.X is out of sync :(
[22:37] <saste> ezekiel, try master
[22:38] <ezekiel> I will try again and hopefully come back with some better info
[23:19] <ezekiel> eck, bad news for me - build from master still doesn't capture synchronized audio
[23:33] <ezekiel> to AVI: http://pastebin.com/HHddvm8W
[23:36] <rsumi> hello, i'm trying to build mediainfo on my box but i'm getting the following errors - http://pastebin.com/ZEJ9sQNA - how can i fix this?
[23:38] <llogan> rsumi: this is #ffmpeg. mediainfo is not from the FFmepg project.
[23:38] <llogan> ezekiel: N-36755-gce153ee does not appear to be current git head
[23:39] <rsumi> llogan: yes i know but i thought it's likely that someone from ffmpeg has ran into this issue. thanks anyways
[23:39] <llogan> we usually just use ffprobe instead
[23:39] <ezekiel> hmm, git log tells me - commit ce153eef8ffdd10ead74ad4646eb784be1c5b107
[23:40] <saste> ezekiel, how can be sure that the timestamps are synched?
[23:40] <llogan> current is N-50103-ga5153b1
[23:41] <saste> use setpts/showinfo to make sure
[23:41] <llogan> rsumi: did you run ./autogen before configure?
[23:41] <ezekiel> llogan: yeah, that commit was pushed while I was building - so now my build is one commit behind
[23:42] <llogan> close enough, but your N- number is odd
[23:42] <llogan> i mean weird
[23:43] <rsumi> llogan: no. i just did now and i get these errors
[23:43] <rsumi> http://pastebin.com/gXNcxc94
[23:44] <llogan> i don't know. make sure you have libtool, automake, autoconf as build deps and libmediainfo as a depends.
[23:45] <llogan> other than that i don't know, especially for PPC.
[23:47] <rsumi> llogan: yeah i have all that stuff except for libmediainfo. i suspect libmediainfo is what i get after building
[00:00] --- Wed Feb 20 2013


More information about the Ffmpeg-devel-irc mailing list