[Ffmpeg-devel-irc] ffmpeg.log.20130308

burek burek021 at gmail.com
Sat Mar 9 02:05:01 CET 2013


[01:19] <barque> How do you scale down a video by 50%?
[01:19] <barque> through ffmpeg
[01:20] <llogan> barque: -vf scale=iw/2:-1
[01:20] <llogan> http://ffmpeg.org/ffmpeg-filters.html#scale
[01:20] <barque> what if I don't know initial width
[01:20] <llogan> -vf scale=iw/2:-1
[01:21] <llogan> it will find the input width automatically
[01:21] <barque> hmmm scale=iw/2:ih/2
[01:21] <barque> that works too?
[01:25] <llogan> probably
[01:25] <llogan> If the value for width or height is -1, the scale filter will use, for the respective output size, a value that maintains the aspect ratio of the input image.
[05:12] <Catoptromancy> anyone have a good reason not to use mic jack for a cassette deck input?
[05:12] <Catoptromancy> heh
[05:13] <Catoptromancy> when I use mic jack I can fine tune volume
[05:13] <Catoptromancy> but line-in I have to lower alsamixer as low as possible
[05:13] <Catoptromancy> and it still clips
[10:14] <brontosaurusrex> are there statical linux builds with x11grab?
[10:16] <relaxed> no
[10:17] <brontosaurusrex> is there another way to pipe "screen" to ffmpeg?
[10:18] <Mavrik> not really
[10:18] <relaxed> brontosaurusrex: Why don't you build ffmpeg with x11grab support?
[10:18] <brontosaurusrex> relaxed, is there a guide?
[10:19] <brontosaurusrex> --enable-x11grab < thats it?
[10:26] <relaxed> yes
[10:26] <brontosaurusrex> mkay, thanks
[10:59] <transplant> hi. I'm on a low CPU device (raspberry pi) and i want to download and view movies
[11:00] <transplant> i thought to download and then use ffmpeg to convert from HD to some other format that takes less cpu do decode
[11:00] <transplant> and when ready, view
[11:00] <transplant> Any suggestions for what that fmt may be?
[11:01] <brontosaurusrex> does raspberry has any kind of gpu accelerated decoders?
[11:01] <transplant> yes but you have to pay MPEG fees to unlock
[11:01] <Mavrik> not for the H.264 one
[11:01] <Mavrik> rPi decodes HD H.264 with AAC audio just fine out of the box
[11:02] <transplant> ok. good to know.
[11:02] <Mavrik> 1080p works, but crashes the decoder regulary
[11:02] <Mavrik> make sure you have enough I/O bandwith :)
[11:02] <transplant> how about just converting to mjpeg or something?
[11:03] <klaxa> but... H.264 hardware decoding is far superior
[11:03] <klaxa> it's done in hardware and has higher quality, right?
[11:03] <Mavrik> transplant, no matter what you do
[11:04] <Mavrik> transplant, the puny ARMv5 in rPi is getting nowhere close to the dedicated hardware H.264 decoder
[11:04] <klaxa> have you tried installing that xmbc thing?
[11:04] <Mavrik> transplant, so use H.264
[11:04] <Mavrik> transplant, because either you'll run out of CPU power for decoding or you'll bog down the memory bandwith with badly compressed video
[11:04] <transplant> really, that bad?
[11:06] <Mavrik> well, you do have a 700MHz ARMv6 processor with a rather narrow memory bus :)
[11:06] <transplant> still, i may download something from torrents and H264 may not be an option...
[11:07] <transplant> and i would be OK to leave the box decoding while i'm away to "avi" or some other lighter format
[11:07] <transplant> might even convert stereo to mono and scale in advance...
[11:08] <brontosaurusrex> transplant, if it has a h.264 decoder, thats your probably best bet, just figure out what resolution/x.264 preset works best and  convert to that
[11:08] <Mavrik> transplant, I don't get it
[11:08] <Mavrik> transplant, if you're gonna be transcoding
[11:08] <Mavrik> what difference does it make
[11:08] <Mavrik> if you do an "avi"
[11:08] <Mavrik> or an "mp4"?
[11:08] <brontosaurusrex> avi is a "container", has nothing to do with lightness
[11:08] <transplant> I had the impression that decoding "avi" or "mjpeg" is requires much less CPU
[11:09] <transplant> brontosaurusrex, i'm not an expert :(
[11:10] <brontosaurusrex> transplant, well, long hours of reading in front of you :P
[11:10] <Mavrik> transplant, but you're not gonna be doing decoding on the "CPU"
[11:10] <Mavrik> if you have dedicated hardware made to decode H.264 video
[11:10] <Mavrik> right baked into the chip
[11:10] <brontosaurusrex> and listen to Macey
[11:10] <brontosaurusrex> Mavrik even
[11:10] <transplant> ok. i will definitelly try to download h264
[11:10] <transplant> but torrents for that may not be available
[11:11] <brontosaurusrex> transplant, thats where you would fire your legit copy of ffmpeg and do the proper transcode (i imagine)
[11:12] <transplant> can anybody suggest a transcoding format to something lighter?
[11:13] <transplant> brontosaurusrex, i suggest reading fabrice bellard's ffmpeg legal FAQ
[11:18] <brontosaurusrex> transplant, i had no questions :) ... lately
[11:19] <transplant> guess i have to bother the mailing list...
[11:19] <transplant> brontosaurusrex, yes it's insightful :)
[11:31] <transplant> can ffmpeg use the GPU for transcoding outside X Windows?
[11:31] <brontosaurusrex> afaik it can't use GPU at all
[11:31] <transplant> shait
[11:32] <brontosaurusrex> its possible to feed it with gpu decoded data thought, but no clue on how to that on nix
[11:32] <brontosaurusrex> unless you would pipe mplayer to it
[11:32] <brontosaurusrex> or something
[11:34] <tyf> hi all, i have followed the exact instruction at https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide to compile and install libfdk-aac and the x264, but I can't compile the ffmpeg successfully, saying that libfdk_aac not found, and similarly to the x264 as well
[11:37] <tyf> this happens when i run the ./configure command before compiling it
[11:37] <tyf> i have make sure that the x264 and the libfdk-aac is installed before running the ./configure for the ffmpeg
[11:38] <tyf> the last few lines of the config.log reads like this:
[11:38] <tyf> Being in IEM Student Chapter, we are strongly supported by the IEM, Malaysia.
[11:38] <tyf> Towards achieving a global mindset with competitive added advantage, this is what we have in stored for you.
[11:38] <tyf>     Closing the gaps between universities and colleges through joint projects.
[11:38] <tyf>     Enhance relationships with various companies through partnership and sponsored events.
[11:38] <tyf>     Seminars and camps will be held to boost passion and leadership in the engineering field among students prior to graduation.
[11:38] <tyf>     Implementing real engineering projects among students.
[11:38] <tyf>     More fun filled activities will be introduced into our student chapter to reward our loyal members.
[11:38] <tyf> gcc -Wl,--as-needed -o /tmp/ffconf.pBzlZJw0 /tmp/ffconf.EnkYE14P.o -lx264 -lvorbisenc -lvorbis -logg -ltheoraenc -ltheoradec -logg -lspeex -lrtmp -lz -lgnutls -lopencore-amrwb -lopencore-amrnb -lmp3lame -lfaac -lass -lm -pthread -lz -lrt
[11:38] <tyf> /usr/bin/ld: error: cannot find -lx264
[11:38] <tyf> /tmp/ffconf.EnkYE14P.o:ffconf.9EBRw4B1.c:function main: error: undefined reference to 'x264_encoder_encode'
[11:38] <tyf> collect2: ld returned 1 exit status
[11:38] <tyf> ERROR: libx264 not found
[11:39] <tyf> Oops, pasted some wrong thing over here....
[11:39] <tyf> anyone can help?
[11:39] <relaxed> where did you install x264?
[11:40] <tyf> the x26
[11:40] <brontosaurusrex> tyf, I'am followin my old thread and its still working, using debian wheezy thought ; http://crunchbang.org/forums/viewtopic.php?pid=274379
[11:40] <brontosaurusrex> dunno how far away ubuntu is this days
[11:41] <tyf>  i pulled the x264 into my home folder and installed using the checkinstall as told on the  https://ffmpeg.org/trac/ffmpeg/wiki/UbuntuCompilationGuide
[11:41] <relaxed> tyf: In the future use pastie.org to paste text and then provide the link. Pasting it in an irc channel can get you kicked.
[11:41] <tyf> ok, thanks for that tip
[11:41] <relaxed> what prefix did you use?
[11:42] <tyf> i didn't use any prefix
[11:44] <relaxed> ./configure --extra-cflags="-I/usr/local/include" --extra-ldflags="-L/usr/local/lib"
[11:45] <relaxed> er, `make clean` first
[11:45] <relaxed> also add your --enable-*
[11:47] <tyf> actually what's the difference between the CXX_INCLUDE_PATH and the --extra-cflags in the command above?
[11:51] <relaxed> where do you see that?
[11:52] <JEEB> if there are system variables that you could poke for similar effect for header paths and linker library paths, that would probably have similar results, but that seems quite complicated for a case like this :P
[11:52] <tyf> actually i tried to fix the problem by looking at the config.log, thought that i can solve it by adding the C_INCLUDE_PATH environment variable, but it is not successful
[11:53] <JEEB> just use the straightforward cflag and ldflag
[11:54] <relaxed> the variables would be CFLAGS and LDFLAGS
[11:54] <JEEB> yeah, and they would have to contain the defaults as well, methinks? like CFLAGS="${CFLAGS} -I/usr/local/include" or so
[11:55] <JEEB> so welp, I wouldn't generally recommend :V Unless you have some very specific needs
[11:56] <relaxed> I would hope they're appended.
[11:57] <JEEB> at least for PATH you have to do that
[11:57] <relaxed> true
[11:57] <JEEB> if CFLAGS is empty by default then I guess you could just add stuff there
[11:57] <tyf> what's the difference between the git://git.videolan.org/ffmpeg and git://source.ffmpeg.org/ffmpeg
[11:57] <JEEB> tyf, latter points towards to the first one
[12:01] <tyf> ok, this is weird, i run "make clean" but i get http://pastie.org/6419828
[12:01] <relaxed> that's fine
[12:02] <tyf> oops, i have wasted my time for cloning the git repo again...
[12:04] <tyf> now it seems it is compiling fine
[12:14] <tyf> thanks a lot!
[12:14] <tyf> although it hasn't finish compiling :)
[12:22] <dekanx> hi
[12:22] <dekanx> i need to convert a flv file to mp4 file
[12:22] <dekanx> but mp4 file must be available to html5
[12:23] <dekanx> which command and parameters i have to use
[12:23] <dekanx> anybody?
[12:27] <relaxed> run ffmpeg -i on the flv
[12:30] <dekanx> asd
[12:30] <dekanx> anybody?
[12:32] <dekanx> relaxed http://pastie.org/6420001
[12:33] <relaxed> I require the output from the command as well.
[12:35] <dekanx> http://pastie.org/6420012
[12:35] <dekanx> here is input and output http://pastie.org/6420012
[12:37] <relaxed> dekanx: First of all your ffmpeg is way too old. You can download a recent version here: http://ffmpeg.zeranoe.com/builds/
[12:39] <masa-> does anyone know if there is work being done to support the (newer?) dxtory format?
[12:40] <masa-> i'm getting a "Unknown frame header 21000001" error when trying to encode some clips recorded with Dxtory with ffmpeg and libx264
[12:40] <dekanx> @relaxed downloading and try again
[12:41] <dekanx> but i need windows binary
[12:42] <relaxed> dekanx: you'll need 7zip to unzip the archive containing the binary. http://ffmpeg.zeranoe.com/builds/win32/static/ffmpeg-latest-win32-static.7z
[12:43] <dekanx> @relaxed downloading and try again
[12:43] <dekanx> but i need windows binary
[12:43] <relaxed> I know, the archive contains a 32bit windows binary.
[12:44] <dekanx> now i have it
[12:45] <dekanx> what i need to now to convert?
[12:46] <relaxed> defaultro: ffmpeg -i input.flv -c:v libx264 -c:a libvo_aacenc -b:a 128k output.mp4
[12:46] <klaxa> * dekanx has quit (Read error: Connection reset by peer)
[12:46] <klaxa> heuhuehue
[12:47] <dekanx> now i have it, how i can convert the file now ?
[12:47] <relaxed> defaultro: ffmpeg -i input.flv -c:v libx264 -c:a libvo_aacenc -b:a 128k output.mp4
[12:47] <relaxed> klaxa: I ignore joins/parts :)
[12:48] <dekanx> relaxed i lost connection , what i should type now for html5 supported mp4 file from flv
[12:49] <relaxed> I used the wrong nick, sorry defaultro
[12:49] <relaxed> dekanx: ffmpeg -i input.flv -c:v libx264 -c:a libvo_aacenc -b:a 128k output.mp4
[12:49] <dekanx> okay i did but here is some problem about http 200
[12:49] <dekanx> http://muzikbenimhayatim.com/201062010183049.flv.mp4
[12:50] <dekanx> okay i did but here is some problem about http 200
[12:50] <dekanx> here is file url http://pastebin.com/R9V60fnd
[12:51] <dekanx> Status Code:206 Partial Content
[12:51] <dekanx> it works but it says that.
[12:51] <dekanx> Will it make problem in future ? , what do you think ?
[12:52] <relaxed> dekanx: That's not an ffmpeg error so I don't know.
[12:52] <dekanx> relaxed did you see that?
[12:53] <dekanx> Status Code:206 Partial Content
[12:53] <klaxa> * dekanx has quit (Read error: Connection reset by peer)
[12:53] <klaxa> <relaxed> dekanx: That's not an ffmpeg error so I don't know.
[12:53] <klaxa> :V
[12:53] <relaxed> :/
[12:53] <dekanx> it's playing anyway
[12:53] <dekanx> it it makes problem i'll re connect and ask
[12:54] <dekanx> thank you so much guys
[12:54] <dekanx> saved me ^_^
[12:54] <relaxed> you're welcome
[13:00] <Bor0> what's the most efficientway to implement frame stepping backwards?
[13:07] <Doxin> I've got around 15000 png files, and a flac file. I'd like to combine these into a 60fps movie file. doing "ffmpeg -i render/%04d.png -i darp.flac -acodec copy derp.mp4" gives this error: "Could not write header for output file #0 (incorrect codec parameters ?)", what am I doing wrong?
[13:08] <klaxa> png is not a video format, how about encoding to h264? try: -c:v libx264 and then play with the settings, also take a look at this guide: http://ffmpeg.org/trac/ffmpeg/wiki/x264EncodingGuide
[13:09] <Doxin> klaxa: right, thanks,
[13:10] <Bor0> is there a specific channel for lib* dev, or any documentation at all
[13:10] <Bor0> or anything more than dranger's tutorials
[13:10] <Doxin> klaxa: "Unrecognized option 'c:v'"
[13:10] <klaxa> mmh for now use -vcodec consider updating your ffmpeg version
[13:11] <Doxin> unknown encoder in that case,
[13:12] <klaxa> ugh... update ffmpeg and you'll need libx264 (should be in the x264 package of your distribution)
[13:12] <juanmabc> Bor0: #ffmpeg-devel and my http://code.google.com/p/openmedialibrary example usage implementation, more update than dranger's stuff
[13:13] <klaxa> Doxin: you could also just grab a static build
[13:13] <Doxin> klaxa: where?
[13:13] <klaxa> either here http://ffmpeg.gusari.org/static/ or here http://dl.dropbox.com/u/24633983/ffmpeg/index.html
[13:15] <Doxin> klaxa: what container format should I use?
[13:15] <klaxa> mp4 or mkv
[13:15] <klaxa> mp4 probably won't take flac though, just use mkv then
[13:16] <Doxin> encoding, thanks.
[13:16] <Doxin> klaxa: doesn't ffmeg have some sort of "I don't care what format just /do/ it already" option?
[13:17] <klaxa> well... if you don't specify anything but input and output it should do something
[13:18] <klaxa> ah, if you want it to be 60 fps you need to specify that too, i forgot, use -r 60 somewhere
[13:18] <Doxin> oh
[13:19] <Doxin> let's just hope the ubuntu repos update soonish :P
[13:19] <klaxa> oh in that case you are not using ffmpeg but avconv
[13:20] <Doxin> in what case?
[13:20] <klaxa> you are running ubuntu, they package avconv
[13:20] <klaxa> or more specific, you are using the ubuntu repos
[13:20] <Doxin> ...
[13:20] <Doxin> that seems silly
[13:21] <Doxin> also, my pc sounds like it's going to go airborne any second now ._.
[13:23] <Doxin> klaxa: doing -r 60 didn't help the framerate it seems. it's taking the images at 30fps at least,
[13:24] <klaxa> hmm where did you put the -r 60? after the -i? try using it before specifing the input, i.e. before -i
[13:24] <klaxa> i'm not so sure about this myself
[13:24] <Doxin> hm
[13:25] <Doxin> that worked
[13:26] <tyf> @ relaxed, hmm....compiled the ffmpeg, tried some video conversion, i got libx264.so.130: cannot open shared object file: No such file or directory.
[13:26] <tyf> Not a big problem, I simply create a symlink to libx264.so.120 since this is what i have...
[13:27] <tyf> but now i still have another problem, ffmpeg: error while loading shared libraries: libfdk-aac.so.0: cannot open shared object file: No such file or directory
[13:27] <klaxa> have you installed fdk-aac?
[13:28] <tyf> yes, i built it from source with --enable-shared and --enable-static
[13:29] <tyf> "locate libfdk " got nothing at all
[13:29] <tyf> seems like the library is not installed?
[13:29] <Doxin> tyf: you done make install?
[13:30] <klaxa> won't change the fact that he doesn't have the lib
[13:30] <klaxa> oh wait
[13:30] <klaxa> no right, make install
[13:31] <tyf> ok, this is what i do...first autoreconf -fiv; ./configure --enable-shared --enable-static ; make ;sudo checkinstall --pkgname=fdk-aac --pkgversion="$(date +%Y%m%d%H%M)-git" --backup=no   --deldoc=yes --fstrans=no --default
[13:33] <tyf> any problem with this? the files should be installed
[13:34] <klaxa> no idea, i always do ./configure --some-options then make then make install
[13:35] <tyf> hmm....wait, i found the library is installed to /usr/local/lib
[13:41] <Mavrik> *grumbl*
[13:41] <Mavrik> why do mobile devices only support h.264 baseline profile -_-
[13:44] <tyf> klaxa : simply added /usr/local/lib to LD_LIBRARY_PATH works fine...
[13:45] <klaxa> Mavrik: depends on the device :P
[13:50] <Doxin> klaxa: woah, youtube does /not/ like that file format, it works fine in a local player, but on yt it's all white. any hints?
[13:51] <klaxa> use a pastebin-like site and post ffmpeg -i file.mkv output
[13:51] <JEEB> Mavrik, plenty of devices support main or high
[13:51] <JEEB> but yes, there's also plenty of cheap'o devices that only support baseline
[13:52] <Doxin> klaxa: http://paste.pound-python.org/show/31060/
[13:52] <JEEB> heck, my 2009 samsung galaxy spica (5700) supported up to baseline, level 3.0... and.... failed at weighted p-frame prediction (visual artifacts)
[13:52] <JEEB> meanwhile my 2005 PSP supports main, level 3.0 (with limitations of reference frames and ban on b-pyramid)
[13:52] <JEEB> :V
[13:52] <klaxa> yeah i think youtube won't like the video profile
[13:53] <klaxa> Doxin: re-encode with -profile:v high
[13:53] <JEEB> my newer huawei thankfully supports up to high profile, level 4.1
[13:53] <Doxin> klaxa: what does that do exactly?
[13:53] <klaxa> select the "high" video profile for x264 instead of "high444p"
[13:54] <Doxin> also "Unable to find a suitable output format for 'high' high: Invalid argument"
[13:54] <klaxa> uh... do paste logs
[13:54] <Mavrik> JEEB, yeah, but official docs require only baseline
[13:54] <JEEB> Mavrik, for what?
[13:54] <Mavrik> for decoding
[13:54] <JEEB> for what?
[13:54] <Mavrik> for Android devices :)
[13:54] <Doxin> klaxa: http://paste.pound-python.org/show/31061/
[13:54] <Mavrik> JEEB, so if I have to deliver video to them
[13:55] <JEEB> Mavrik, link please
[13:55] <Mavrik> I can't do Main or High no matter the fact that 80%+ devices have Main/High decoders
[13:55] <klaxa> Doxin: run: ./ffmpeg -r 60 -i render/%04d.png -i dirty_paws.flac -c:a copy -c:v libx264 -profile:v high dirty_paws.mkv
[13:55] <JEEB> but yes, if you want to support "most devices" you are really limited on both resolution and profile :P
[13:55] <Mavrik> JEEB, http://developer.android.com/guide/appendix/media-formats.html#core
[13:55] <Mavrik> JEEB, without having a way of knowing what profile the device supports yeah, I'm limited :\
[13:55] <Mavrik> and annoyed :)
[13:55] <Doxin> klaxa: "Error while opening encoder for output stream #0:0 - maybe incorrect parameters such as bit_rate, rate, width or height"
[13:56] <Doxin> klaxa: "x264 [error]: high profile doesn't support 4:4:4"
[13:56] <JEEB> -pix_fmt yuv420p
[13:56] <JEEB> for whatever reason you or ffmpeg want to encode 4:4:4 YCbCr (colloquially "YUV")
[13:57] <JEEB> I guess youtube's libavcodec hasn't been updated to support it
[13:57] <Doxin> works
[13:57] <JEEB> 4:4:4 YCbCr H.264 that is
[13:57] <Doxin> let's hope yt likes this encoding :P
[13:57] <JEEB> it should, 4:2:0 YCbCr has been around for quite a while
[13:58] <JEEB> for H.264
[13:58] <JEEB> you could try and poke youtube to update their software, I guess?
[13:58] <Doxin> if this works I aint complaining.
[13:58] <klaxa> i think i uploaded videos in high444p
[13:58] <JEEB> klaxa, high 4:4:4 profile itself is OK
[13:58] <JEEB> if the video is 4:2:0
[13:59] <JEEB> because lossless is in predictive lossless 4:4:4 profile, if I recall correctly
[13:59] <klaxa> hmm... i don't know if the video was 4:4:4 too
[13:59] <klaxa> well i ran it with that old lossless_ultrafast preset
[13:59] <JEEB> yeah
[13:59] <JEEB> that's predictive lossless 4:4:4 profile
[13:59] <klaxa> luckily those got removed, was kinda weird
[13:59] <JEEB> basically that's the only profile that contains lossless coding
[14:00] <JEEB> and before x264 got support for 4:2:2 and 4:4:4 coding, you could only encode 4:2:0 in it
[14:00] <JEEB> and that should be the default
[14:00] <JEEB> it's separate from high 4:4:4 profile
[14:00] <JEEB> IIRC
[14:00] <Doxin> gotta love the sound of a cpu fan hitting the resonant frequency
[14:02] <JEEB> hmm
[14:03] <JEEB> klaxa, anyways lossless seems to be under the same profile, but that has been supported in libavcodec and x264 for lossless coding for 4:2:0 YCbCr for ages, which is why it works (Youtube updated their decoder somewhere in either '09 or '10)
[14:03] <JEEB> but they still haven't updated it to one that supports 4:4:4 in that profile (can be lossy or lossless)
[14:04] <klaxa> i see
[14:04] <klaxa> wouldn't one think decoders should be updated more frequently?
[14:04] <JEEB> haha
[14:04] <JEEB> yeah...
[14:04] <JEEB> 'tube isn't exactly known for updating their software :)
[14:05] <JEEB> I kind of understand of course, it's quite a big system :)
[14:05] <JEEB> but yeah, expect a few years of lag
[14:06] <Doxin> I don't see why it ought to take a few years, a few months i'd understand. but afaik it's try in dev enviroment -> smoke tests -> roll out
[14:07] <JEEB> well, I think parts of high 4:4:4 predictive profile (basic new lossless for 4:2:0) was implemnted in x264 and libavcodec around... '08?
[14:07] <Mavrik> Doxin, because: "if it works, don't break a system used by tens of millions of people" ;)
[14:07] <JEEB> it was much later, like '10 or so when 'tube finally started decoding it
[14:08] <JEEB> yeah
[14:08] <Mavrik> especially if you're going to destroy all the gains by using 4:4:4 anyway
[14:08] <Mavrik> *of using
[14:08] <Doxin> Mavrik: it's trivial to unbreak at the first signs of trouble,
[14:09] <Doxin> Mavrik: I'm sure it could be automated too, sudden jump in upload failures -> rollback to known working.
[14:13] <Doxin> there we go, got it all working. thanks klaxa, JEEB. if you're curious as to what I did: http://www.youtube.com/watch?v=Lo9CBCN3fiE
[14:14] <klaxa> >This video contains content from UMG, who has blocked it in your country on copyright grounds.
[14:14] <klaxa> huehuehue
[14:14] <Doxin> germany?
[14:14] <klaxa> yup
[14:14] <Doxin> poor you.
[14:14] <Doxin> klaxa: only country it's blocked in.
[14:14] <klaxa> don't worry i can has private us proxy
[14:14] <Doxin> yay
[14:14] <tyf> can i do on-the-fly transcoding with ffmpeg and then stream via rtsp directly? Quite confused with ffmpeg manpage
[14:15] <Trashlord> not sure about that
[14:16] <Trashlord> it'd take a lot of processing power though
[14:19] <klaxa> it's possible, but it's not easy
[14:19] <klaxa> the rtsp streaming that is
[14:19] <klaxa> encoding on the fly isn't that hard
[14:19] <tyf> so, what do you recommend me to read on?
[14:20] <klaxa> dunno ffserver?
[14:20] <tyf> the ffserver admits itself to be broken....as can be seen in its manpage.... :P
[14:21] <JEEB> ffserver can work, as long as you don't poke the broken parts ^^;
[14:21] <Mavrik> ^^
[14:22] <JEEB> just don't ask me how to set it up or use it
[14:22] <JEEB> ok?
[14:22] <JEEB> :D
[14:22] <Mavrik> I've found that just sending MPEG-TS over UDP usually works the best
[14:22] <Mavrik> and it's simple to setup :)
[14:22] <klaxa> nobody seems to know that
[14:22] <Mavrik> (if you're on the same network)
[14:22] <JEEB> there are some protocols that work straight via ffmpeg tho
[14:22] <JEEB> and those are relatively easy to set up
[14:22] <JEEB> if you want something simple to stream rtsp with, I recommend VLC
[14:23] <JEEB> that's what I used that one time I was testing rtsp
[14:24] <tyf> my plan is to stream h264 videos to the infamous Raspberry Pi
[14:25] <klaxa> why not use http then?
[14:25] <JEEB> you could use a whole lot of stuff then
[14:25] <JEEB> rtsp is just one of the alternatives
[14:25] <klaxa> actually i wanted to say earlier that i wrote some scripts in python to do that
[14:26] <klaxa> server started on the pi, client started with a video as the first argument and it streams it to the pi
[14:26] <klaxa> friend has a pi and his computer is rather far away from his TV so he connects the pi to the TV and streams over wifi with http
[14:27] <klaxa> funny thing though, i don't even have the scripts here :V i didn't upload them when i wrote them at my friend's computer
[14:45] <tyf> does the omxplayer on the pi supports streaming over http?
[14:45] <klaxa> no
[14:46] <klaxa> it also doesn't support playing from stdin
[14:46] <klaxa> i used netcat and named pipes
[14:46] <tyf> so i still need to use rtsp if i want to stick with omxplayer?
[14:46] <klaxa> well not exactly netcat
[14:46] <klaxa> no you don't have to
[14:47] <klaxa> like i said, named pipes work too
[15:10] <alx-> hey, what's the best way to create a slideshow from images at 30fps? im trying to avoid using low fps like 0.2 because youtube has problems importing it
[15:11] <Mavrik> add an fps video filter which will duplicate frames
[15:12] <Absolute0> Can I pipe data into ffmpeg instead of using -i?
[15:12] <Absolute0> cat foo > ffmpeg - output.ogg breaks
[15:12] <Doxin> Absolute0: named pipes :D
[15:13] <Absolute0> Doxin: I have python code that gets an uploaded file in memory
[15:13] <Absolute0> I would like to convert the in memory file without saving it first
[15:14] <Mavrik> Absolute0, well, you HAVE to use -i since that's the parameter that tells ffmpeg what the input is :P
[15:14] <Doxin> Absolute0: use a named pipe. http://linux.die.net/man/7/fifo
[15:14] <Mavrik> Absolute0, "ffmpeg -i - ... ;) "
[15:15] <Absolute0> Mavrik: that worked :)
[15:15] <Absolute0> haha
[16:37] <efyx__> Hi, I'am using ffmpeg libs in my project, with custom I/O callback (via avio_alloc_context), in my project all my I/O are async. For reading everything works just fine
[16:37] <efyx__> but when I seek my media, ffmpeg try to read right after the seek and data are not available yet, what's the correct value to return in such case?
[17:20] <codecowboy> I tried my usual ffmpeg command for mkv files to convert a file for viewing on an xbox 360 (xbox can't play it). http://pastie.org/private/nefbu0ju4i0nrhhcyqy5ig. Anyone know how to change the command so this will work?
[17:22] <JEEB> [mp4 @ 0x7fdbc3829400] track 0: could not find tag, codec not currently supported in container
[17:22] <JEEB> is the error
[17:22] <JEEB> and the reason is
[17:22] <JEEB> Video: msmpeg4v3 (DIV3 / 0x33564944)
[17:22] <JEEB> which is not MPEG-4 Visual :)
[17:23] <JEEB> in other words, you would have to re-encode the video
[17:23] <JEEB> replace "-vcodec copy" with "-c:v libx264 -preset fast -crf 23"
[17:24] <JEEB> -c:v is same as vcodec, but a new way of setting it
[17:25] <JEEB> -preset sets a libx264 preset
[17:25] <JEEB> in this case, a preset called "fast"
[17:25] <JEEB> http://mewiki.project357.com/wiki/X264_Settings#preset <- list of presets
[17:25] <JEEB> -crf is a libx264-specific rate control setting
[17:25] <JEEB> it is pretty much the closest we all have to "constant quality"
[17:25] <JEEB> if it looks bad, make it lower
[17:26] <JEEB> if it looks good, make it higher
[17:26] <JEEB> and when you find the highest number that still looks good
[17:26] <JEEB> that's what you use :)
[18:08] <codecowboy> JEEB thanks, trying that
[18:38] <QOAL> Hello. I'm trying to encode a sequence of png files into a video. Unfortunately, thanks to the subtle (dithered) colour gradients used, I'm only able to produce a video which has rather obvious colour banding.
[18:40] <sacarasc> 10bit H264 is meant to help with that a lot, IIRC...
[18:40] <LithosLaptop> does the colour banding already show up in the source pngs?
[18:41] <LithosLaptop> if not then dithering would help or a higher bitdepth like 10bit
[18:44] <QOAL> It only shows up a tiny bit in the source images.
[18:44] <QOAL> This is a frame: http://i.imgur.com/nhEO521.png
[18:52] <LithosLaptop> don't know if this might help or not: http://doom10.org/index.php?topic=1526.0
[19:00] <QOAL> Hmm, well I'll give it a test. but I'm only using 8bit images.
[19:06] <LithosLaptop> oh
[19:25] <QOAL> I think I'm just going to deal with what I have, for now. But thank you for your time :)
[19:31] <Mista_D> can I create a .ts and .mp copying encoded stream to both files through two different muxers?
[21:20] <FlowRiser> hey guys, i am trying to screencast my desktop (using kubuntu 12.10 with alsa and pulse); my problem is that the resulting video is very (like it's jumping frames) sped up. If i don't specify any audio options it works just fine. Here's what i use: ffmpeg -f alsa -ac 2 -ab 192k -i pulse -f x11grab -s 1360x760 -r 30 -i :0.0 -acodec libmp3lame -vcodec libx264 -preset ultrafast -threads 0 /home/user/me.avi
[21:21] <FlowRiser> I get this extra line with the audio options: lookaheadless mb-tree requires intra refresh or infinite keyint
[22:23] <fjodor> hi, how do i tell ffmpeg to make the output file approximatly as big as the input file?
[22:23] <fjodor> i want to convert from m4a to mp3 and file size always explodes
[22:25] <sacarasc> Look at the bitrate of the input file and then do -b:a 123k
[22:25] <sacarasc> Where 123 is the bitrate of the input file.
[22:25] <fjodor> mmh ok
[22:25] <fjodor> where can i see the bitrate of the file?
[22:25] <sacarasc> ffmpeg -i foo.m4a
[22:31] <fjodor> ffmpeg -b:a 49k -i Skream_and_Benga_-_Baauer_is_on_the_phone_b01qtc97_default.m4a Skream_and_Benga_-_Baauer_is_on_the_phone_b01qtc97_default.mp3
[22:31] <fjodor> oh disconnect
[22:32] <fjodor> sacarasc: the file still blows up from 40mb to 160mb that was my command: ffmpeg -b:a 49k -i Skream_and_Benga_-_Baauer_is_on_the_phone_b01qtc97_default.m4a Skream_and_Benga_-_Baauer_is_on_the_phone_b01qtc97_default.mp3
[22:32] <sacarasc> Put the -b:a bit just before the output file.
[22:32] <sacarasc> It goes ffmpeg [input options] -i input [output options] output
[22:33] <fjodor> mmh thx
[22:36] <fjodor> sacarasc: sry, but same story over again 40mb to 160mb
[22:36] <fjodor> ffmpeg -i Skream_and_Benga_-_Baauer_is_on_the_phone_b01qtc97_default.m4a -b:a 49k Skream_and_Benga_-_Baauer_is_on_the_phone_b01qtc97_default.mp3
[22:37] <fjodor> http://pastebin.com/DZiAdmrc
[22:42] <sacarasc> Other than you're not actually using ffmpeg, I am not sure what is going wrong, there...
[22:42] <fjodor> :D
[22:42] <fjodor> haha
[22:43] <fjodor> mmh i dont want to be a kind of dickhead, but this shouldnt be a problem at all. i am kind of pissed that this is such a big deal
[22:44] <sacarasc> Try changing -b:a to just -b
[22:44] <sacarasc> Also, read that.
[22:47] <fjodor> sacarasc: haha thx
[22:48] <fjodor> now it changed the bitrate
[22:48] <ubitux> the ffmpeg/libav situation link is broken
[22:48] <sacarasc> Is it? Aww.
[22:48] <ubitux> at least not pointing to what was wanted
[22:59] <fjodor> mmh so they dislike michael niedermeyer? why that?
[23:12] <iive> fjodor: you want their public reason or the real reason?
[23:18] <fjodor> real reason
[23:18] <fjodor> iive:
[23:19] <iive> first is taking over the fftech foundation that had collected some money from (l)gpl violations. the second is that at least half of ffmpeg was written by michael himself.
[23:21] <iive> libav have been rewriting and refactoring code for years now, to get rid of everything holding the copyright of michael.
[23:22] <JEEB> huh? Then why have they been merging mn's commits just fine?
[23:22] <iive> even now when they need to port back something from ffmpeg, they redo the fix rather than just mergin it.
[23:22] <JEEB> naturally they have a derpy stance towards "if you want something merged from ffmpeg you will have to bring it up to us", but I do note quite a few commits by mn
[23:22] <iive> JEEB: link?
[23:22] <JEEB> sec
[23:23] <JEEB> my gpu chose not to co-operate too much today on this damn laptop
[23:23] <iive> and there are a number of cases where where copyright notices got lost...
[23:23] <JEEB> http://git.libav.org/?p=libav.git;a=search;h=HEAD;s=Michael+Niedermayer;st=author
[23:23] <JEEB> quite a few, so I don't see them exactly trying to get rid of everything by him :)
[23:26] <fjodor> iive: so they want to make money from libav and try to get rid of MN, thats why?
[23:26] <JEEB> also at least elenril's opinion on copyright notices in the file was that git's history does it better, but I have no idea about that in general
[23:26] <iive> fjodor: that's what I suspect.
[23:27] <fjodor> mmh but if they want to get rid of MN, whats with the other copyright holders? they do have a right too
[23:28] <fjodor> or is a company behind libav?
[23:28] <JEEB> anyways, I know that libav seemed to start because of dislike of mn and the slow pace of stuff getting merged into ffmpeg up to 2010 or so. ffmpeg has changed since that, though. Right now libav is the one that merges less because of stylistic reviews and heavy opinions on something having to 'look right'
[23:28] <JEEB> I really haven't seen a specific effort to remove copyright from anyone
[23:29] <klaxa> but that's great isn't it? if ffmpeg and libav merge commits from each other that's a good thing isn't it?
[23:29] <iive> fjodor: One of the instigators of the coup does make money by selling his own code to firms. He closely guards it so nobody have a major copyright on it.
[23:29] <klaxa> i mean both ends can develop in their preferred environment
[23:29] <klaxa> shit gets done and both projects profit
[23:29] <JEEB> klaxa, yup -- and we've since the split had a great speed-up in development
[23:29] <iive> JEEB: I stand corrected, but 1 year of commits on a single page is very few.
[23:30] <iive> the speed-up also comes at cost. 3 years of constant api and abi breakage.
[23:30] <JEEB> iive, that's because in general libav merges very little. the official stance is "we don't have enough time, if you want something in you post it on our mailing list"
[23:30] <JEEB> which is :effort:
[23:30] <JEEB> very much so
[23:30] <JEEB> so I'm not surprised
[23:31] <JEEB> that's pretty much the only anti-ffmpeg thing I've noticed in some of the folk at libav
[23:31] <JEEB> most just go on developing their stuff
[23:31] <fjodor> mmh that reminds me of torvalds git talk. forking and merging is a good thing etc... a little bit like the scientific method
[23:32] <iive> fjodor: the coup was initially in 2 steps. The first step was takeover of the project by the root administrators. developers tried to work together for 3 months, but it just didn't work out.
[23:32] <iive> constant flaming, these who set the new rules tend to ignore these who didn't joined the coup... and there was a constant flaming.
[23:33] <iive> until michael setted up a new server and set the dns to it. then the take-overs finally decided to announce a proper fork.
[23:35] <iive> well... everybody get their own playground and toys.
[23:37] <fjodor> yeah and the fittest gets their commits merged. like android after linus didnt accept their commits for a long time
[23:38] <fjodor> may i ask something ffmpeg unrelated. i am planing a small project and want to speak to a developer
[23:40] <brx_> fjodor, what type of dev you wanna talk to?
[23:40] <fjodor> it's actually pretty simple so i think every programmer would help me
[23:40] <brx_> go on, ask
[23:42] <fjodor> i was thinking about a new kind of document format. i've already found a project which almost does what i want and it only needs some kind of modification
[23:43] <fjodor> http://stacks.math.columbia.edu/ they basicly write a math book in latex and generate pdf/website+comment with git
[23:43] <fjodor> the extra thing they invented was a so called tag-system
[23:44] <fjodor> like every lemma/definition gets an identifier
[23:44] <fjodor> so you can link to it no matter what happens to that lemma later on
[23:45] <fjodor> so they are currently using two structure elements. 1. tags, which resembles a graph
[23:45] <fjodor> 2. latex chapters etc. which resembles a tree
[23:46] <fjodor> i want to add another thing to this project
[23:47] <fjodor> i want to be able to import this graph and tree structure into a graph analysing software
[23:47] <saste> fjodor, basically you need a latex parser
[23:47] <fjodor> something like gephi: https://gephi.org/. this should be relativly simple right?
[23:48] <saste> i guess they are using LaTeX for the source
[23:48] <fjodor> yes
[23:48] <fjodor> import it would be the first thing
[23:49] <fjodor> and then export it again to a latex file. i think of it as follows: this is a really big project and a big book. you import it to gephi, select a subgraph and export the subgraph again to latex.
[23:50] <fjodor> so you can have your own special interest book
[23:50] <fjodor> generated from this big book
[23:51] <saste> fjodor, what do you need gephi for?
[23:51] <saste> also, are you aware of any latex parser?
[23:51] <fjodor> no
[23:51] <fjodor> i would like to be able to get an overview of the whole book...
[23:52] <saste> what bothers me more about latex is that it can't be used as a library
[23:52] <saste> to render latex output
[23:52] <saste> fjodor, also keep in mind that Latex *is* a programming language
[23:53] <saste> so you might be able to create code which introspects the structure of the document, and send it to another tool
[23:53] <saste> then you could tell the latex interpreter to only render a subset of what you selected with your graphical tool
[23:54] <fjodor> mmh i would actually think of it as an cut and paste job
[23:54] <saste> but i wonder, how is this related to ffmpeg?
[23:54] <fjodor> all you need to do is to take care of hierarchy elements
[23:55] <fjodor> i politly asked for off-topic question...
[23:55] <saste> yes and you need a parser
[23:55] <saste> i suggest to have a look at org-mode for emacs, there is some interesting stuff which may be related to your quest
[23:56] <fjodor> oh i know org-mode
[23:56] <fjodor> i was thinking about using it too :D haha
[23:57] <saste> another possible approach, there is this pandoc which should have a latex parser
[23:57] <saste> if you like haskell
[23:57] <saste> so you may rely on that, and glue it with the graphic tool
[00:00] --- Sat Mar  9 2013


More information about the Ffmpeg-devel-irc mailing list