[Ffmpeg-devel-irc] ffmpeg.log.20121004

[burek]

[01:13] <Diogo> anyone works with red5 (i know that is not the right channel to ask it.. sorry for that)
[08:04] <thresh> michaelni: ping
[09:21] <praveenmarkandu> Hi. FFMPEG 1.0 was released a few days ago
[09:21] <praveenmarkandu> there is a change "SubRip encoder and decoder without embedded timing"
[09:21] <praveenmarkandu> what does this mean?
[09:23] <ubitux> praveenmarkandu: it's an api change, this changelog entry was moved
[09:24] <ubitux> (or maybe not this one in particular)
[09:24] <ubitux> praveenmarkandu: do you want the technical details?
[09:24] <praveenmarkandu> @ubitux, not sure if I would understand the technical details
[09:25] <praveenmarkandu> but sure, why not
[09:25] <ubitux> ok
[09:25] <praveenmarkandu> is that documented anywhere?
[09:25] <ubitux> git log
[09:25] <ubitux> basically, when you have a .SRT file, the demuxer split that file into chunk including the timing information
[09:25] <ubitux> (the hh:mm:ss.xxx --> hh:... thing)
[09:26] <praveenmarkandu> yup
[09:26] <ubitux> along with the payload (the markup text, like <i> blablabla </i> ...)
[09:26] <ubitux> so the decoder receives these two information
[09:26] <ubitux> in a "text" payload
[09:26] <ubitux> the problem is, formats like matroska allow some SubRip markup
[09:26] <ubitux> (the <i> blablabla... )
[09:27] <ubitux> so the matroska demuxer needs to send this to the subrip decoder
[09:27] <ubitux> but it can't send chunks like the SRT demuxer
[09:27] <praveenmarkandu> ok. i actually understand so far
[09:27] <ubitux> since the timing information is stored differently
[09:28] <ubitux> previously it was constructing and adding the "hh:mm:ss.xxx --> hh:..." string into the packet before sending it to the decoder
[09:28] <ubitux> which was quite ugly
[09:28] <ubitux> now the matroska demuxer can just say it's subrip markup
[09:29] <ubitux> (the subrip decoder has now two branches: one for "SRT" packets, and one for subrip - without timing in payload so -)
[09:29] <ubitux> is that clear enough?
[09:30] <praveenmarkandu> yes
[09:30] <praveenmarkandu> thanks. lol. from the description, i thought it would solve my mpegts subtitles streaming problem
[09:30] <praveenmarkandu> hahahaa. hopeful & naive
[09:30] <ubitux> :)
[09:33] <ubitux> i'm really curious to know if you can put some actual *text* subtitles in mpeg-ts
[10:05] <praveenmarkandu> @ubitux, yes i am very curious myself :D
[10:05] <ubitux> i'm afraid you might be able to put one of the N TTML flavors
[10:05] <thresh> where do you guys want commits notifications come to?
[10:05] <thresh> this channel or ffmpeg-devel?
[10:05] <ubitux> devel channel obviously
[10:35] <smellynoser> Hi, with vlc I can use video encoder of mp2v, but with ffmpeg I can't even though vlc says to use ffmpeg - is the name different with ffmpeg?
[10:36] <ubitux> parse error
[13:14] <copper> hi
[13:14] <copper> http://ffmpeg.org/pipermail/ffmpeg-devel/2012-September/131785.html
[13:14] <copper> Apparently the TAK demuxer only works on TAK files that don't have APE tags
[13:15] <copper> Also, I can't get the ffmpeg plugin in deadbeef to load TAK files
[13:16] <JEEBsv> I recommend you either comment on the mailing list, or on #ffmpeg-devel
[13:16] <JEEBsv> the guy's nickname begins with durandal
[16:57] <diroots> hi there, I'm trying to convert a H264 video to DVCPRO with ffmpeg 7:0.11.2-dmo1. I get a strange color problem : see http://pastebin.com/Zvjx3whP, there is a screenshot here : http://dev.diroots.info/H2642DVC.jpg any idea? would that happen because of the pix_fmt conversion from yuv420p to yuv422p?
[17:31] <khali> hi there
[17:32] <khali> I have a ffmpeg conversion problem and am looking for advice
[17:32] <khali> I have a bunch of avi files with 1 xvid video stream and 2 mp3 audio streams, and I need to remove the second audio stream from them
[17:33] <khali> I am using: ffmpeg -i file.avi -codec:v copy -codec:a copy new.avi
[17:33] <Mavrik_> khali: you need to use -map parameter to set which streams get copied
[17:33] <khali> but the resulting file can't be played by ffplay or vlc... either the video is very slow, of the sound is very choppy
[17:34] <Mavrik_> and please put your ffmpeg output to a pastebin so we can actually help you.
[17:34] <khali> Mavrik_: the streams I get in the output files are the ones I want, so I doubt it'll help, but I will try
[17:39] <khali> Mavrik_: http://pastebin.com/fkb0r6Wp
[17:40] <diroots> can someone help me with this :  I'm trying to convert a H264 video to DVCPRO with ffmpeg 7:0.11.2-dmo1. I get a strange color problem : see http://pastebin.com/Zvjx3whP, there is a screenshot here : http://dev.diroots.info/H2642DVC.jpg any idea? would that happen because of the pix_fmt conversion from yuv420p to yuv422p?
[17:40] <khali> ffmpeg plays sound OK but video is too slow, vlc plays video OK but sound is completely choppy
[17:41] <Mavrik_> hmm
[17:41] <khali> I found that I could get a playable output file by using mp4 instead of avi as the container, or by reencoding the audio stream, but I would like to avoid doing that if possible
[17:42] <Mavrik_> khali: it's probably an index problem
[17:42] <Mavrik_> khali: try using mencoder to reindex the file after conversion
[17:42] <khali> Mavrik_: I tried adding -map 0:0 -map 0:1 but the output file is exactly the same, to the bit - so that's not the solution
[17:42] <Mavrik_> mencoder output.avi -o output_fixed.avi -forceidx -ovc copy -oac copy
[17:44] <khali> Mavrik_: thanks for the suggestions, I tried that,but it doesn't solve the problem... actually it is worse, as video looks corrupted now
[17:45] <Mavrik_> corrupted video by copying the stream?
[17:45] <khali> Mavrik: yes I know it sounds odd
[17:49] <khali> Mavrik: there was a lot of "Skipping frame!" and the resulting file is ~8% smaller than the original
[17:50] <khali> Mavrik: but if I use mencoder directly on my source file, the output looks reasonable
[17:51] <khali> Mavrik: one thing looks wrong in my input file: audiocodec: framecopy (format=55 chans=2 rate=48000 bits=0 B/s=16000 sample-0)
[17:51] <khali> Mavrik: it should be bits=16
[17:51] <khali> I tried forcing it with -bits_per_coded_sample:a 16 but it did not seem to help
[17:52] <khali> Mavrik: another thing I noticed when using ffmpeg : if I do .avi -> .mp4 and then .mp4 to .avi, I get a playable avi file
[17:53] <Mavrik_> this looks like an avi muxer issue
[17:57] <khali> Mavrik: I was wondering if there was any option I could add to the command line to get the output file right directly
[17:58] <khali> Mavrik: I tried -vbsf dump_extra, -vbsf remove_extra and -bits_per_coded_sample:a 16, neither helped
[18:02] <khali> now tried -bits_per_raw_sample:a 16 too, but no luck
[18:02] <khali> Mavrik: are there options that affect the avi muxing process?
[18:30] <relaxed> khali: ffmpeg -i input -map 0:v -map 0:a:0 -c copy output.avi
[18:31] <relaxed> would remux video and the first audio stream
[18:33] <khali> relaxed: thanks for the suggestion, but the result is the same :(
[18:34] <relaxed> try with an updated version of ffmpeg
[18:34] <khali> except that now ffplay says: [avi @ 0x7f443c0008c0] non-interleaved AVI
[18:34] <khali> relaxed: I'm using ffmpeg version 0.11.1, is is considered old?
[18:35] <relaxed> When a problem occurs anything but git is old. You can try my static build http://goo.gl/DPrRY
[18:35] <khali> I'll be happy to rebuild my pwn ffmpeg from source as I was doing in the good old times
[18:35] <relaxed> That will work too
[18:57] <vlaad_> hello, anybody online?
[19:02] <relaxed> there are billions online
[19:02] <relaxed> BILLIONS
[19:03] <vlaad_> :)
[19:04] <vlaad_> and at least one in billion who accept error reports for ffmpeg-1.0 (git pull)?
[19:06] <vlaad_> http://pastie.org/4909806
[19:09] <vlaad_> for some strange reason, configure fails to find /usr/local/ version I've compiled; the one used is ubuntu-12.04 flite-dev package
[19:10] <vlaad_> but I don't think that breaks the test, anyway (correct me please, if I'm getting it wrong)
[19:11] <relaxed> you're running fate?
[19:12] <vlaad_> if you're asking for "make fate", yes (make check)
[19:14] <relaxed> I don't have any experience with fate but it looks all green here. http://fate.ffmpeg.org/
[19:15] <relaxed> What are you trying to do?
[19:16] <vlaad_> just to compile ffmpeg with as many possible conversions as possible (making it as useful as possible)
[19:19] <vlaad_> readelf gave me "6: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND cmu_lex_init" on /usr/lib/libflite_cmu_us_slt.so.1, that's why I wanted to use my own compiled library
[19:45] <vlaad_> update: I succeeded (executing all fate tests without errors) when I _omitted_ --enable-libflite; it seems that package flite1-dev on Ubuntu is broken; I'll now try to see in configure what blocks the recognition of libflite manually compiled (if anybody knows, please interrupt me), this could be helpful in resolution of this error on Ubuntu 12.04
[20:42] <khali> relaxed: wow, it worked!
[20:43] <khali> relaxed: using ffmpeg built manually from the git repository, your command line works
[20:47] <khali> relaxed: the output has only 11 bytes differing from the non-working variant, (at least) 5 of which are from the different version numbers of Lavf
[20:47] <khali> probably 6
[20:47] <khali> do that's just 5 different bytes fixing it :)
[21:03] <khali> relaxed: now I feel a bit stupid as obviously I should have tried this first :/
[21:04] <khali> Mavrik_: thanks to you too, BTW :)
[21:05] <khali> relaxed: thanks a lot for your help, I should be able to convert all my files and save some disk space now :)
[21:10] <vlaad_> I made it, do you accept patches? :)
[21:14] <vlaad_> hello, maintainers?
[21:17] <saste> vlaad_, patches are welcome :)
[21:17] <vlaad_> where to put them? (one patch, one pkgconfig file)
[21:17] <saste> anyway, what did you make?
[21:18] <vlaad_> I've made pkgconfig for libflite, then I had to change configure to use require_pkg_config instead of require2
[21:18] <saste> vlaad_, file a bug on trac, but patches get more attention on the ffmpeg-devel mailing list
[21:18] <vlaad_> in order to bypass ubuntu's dynamic build
[21:19] <saste> uhm, can you pastebin the patch?
[21:19] <vlaad_> yes
[21:19] <saste> i'm the author of flite so maybe i can help
[21:20] <vlaad_> http://pastie.org/4910416
[21:21] <saste> vlaad_, what's exactly the issue?
[21:22] <vlaad_> I was not able to compile with ubuntu's dev package, so I wanted to use the /usr/local one (mine), for some strange reason (later, I deduced that static build is the only solution, test returned errors for undefined symbols)
[21:22] <vlaad_> I'll be back in 15 minutes
[21:35] <vlaad_> I'm back, did I miss something important? I'm going to bugzilla to report
[21:39] <saste> vlaad_, https://ffmpeg.org/trac/ffmpeg
[21:40] <saste> also, can you explain the issue (on trac is even better)?
[21:40] <vlaad_> sure, basically this affects ubuntu (perhaps debian too) people only
[21:42] <saste> you create a new account, then you click on new ticket and file the bug
[21:42] <saste> if you have a patch you can add the patch file for it
[21:42] <vlaad_> I did, just reading the "tracker manual"
[21:42] <vlaad_> I want to be nice :)
[21:52] <saste> vlaad_, don't be too nice, save your time ;-), I won't be picky with you bug report
[21:52] <saste> what I need is just an explanation of the issue
[21:52] <saste> i'm on ubuntu 11.10, and it works fine here, so I need to understand why is failing on other systems
[21:53] <vlaad_> ar you on multiarch or i686 system?
[21:53] <vlaad_> s/ar/are/
[21:55] <vlaad_> basically, the only change to flite-1.4 is that I've made libflite.pc and put it in /usr/local/lib/pkgconfig (thus, modified ffmpeg's configure can find it, instead of always using /usr/lib/ shared version provided by ubuntu)
[21:57] <saste> vlaad_, that is you *created* a .pc file?
[21:57] <vlaad_> yes
[21:57] <saste> that's beyond of what is fixable in ffmpeg
[21:57] <saste> should be rather be addressed by debuntu/flite upstream
[21:58] <vlaad_> uhh... OK, but issue is also that without patching, ffmpeg's configure will not search (or at least is not finding on my machine) the /usr/local version
[21:58] <vlaad_> I can reproduce it
[21:59] <saste> vlaad_, did you try with --extra-ldflags=/usr/local/lib ?
[21:59] <saste> also you may want to fix globally you environment, and add /usr/local to LD_FLAGS
[21:59] <vlaad_> no, I was relying on environment LDFLAGS
[21:59] <saste> then it shouldn't fail
[22:00] <saste> what's your LDFLAGS?
[22:00] <vlaad_> hmm, I'll try with --extra-ld-flags
[22:03] <vlaad_> configure correctly reads my LDFLAGS='-rdynamic -lpthread -lgcc -Wl,-Bsymbolic-functions -Wl,-z,relro -Wl,-z,now -lgomp -L/usr/local/lib -L/usr/lib -L/usr/lib/gcc/x86_64-linux-gnu/4.6 -L/usr/lib/gcc/x86_64-linux-gnu/4.6/32'
[22:03] <vlaad_> so local lib is there, and I'm getting "ERROR: libflite not found"
[22:04] <vlaad_> I'll put now my config.log
[22:04] <vlaad_> on pastie
[22:08] <vlaad_> it is here, I've copied just the part when configure tries to compile one libflite dependent test file: http://pastie.org/4910604
[22:09] <vlaad_> therefore, I added the following to libflite.pc in order to resolve this: "Libs: -L${libdir} -lasound -lflite_cmu_time_awb -lflite_cmu_us_awb -lflite_cmu_us_kal -lflite_cmu_us_kal16 -lflite_cmu_us_rms -lflite_cmu_us_slt -lflite_usenglish -lflite_cmulex -lflite"
[22:14] <saste> vlaad_, silly flite...
[22:14] <saste> so it is linking against libasound (which is never used in the current app)
[22:14] <saste> and failing
[22:15] <saste> the best solution would be to use a pkgconfig file, but since it is *not* provided by flite, the best we can do is to add a dependency to libasound
[22:16] <vlaad_> OK, should I report this as a bug? or not?
[22:16] <saste> vlaad_, wait, give me some minutes and i'll give you a patch to test
[22:17] <vlaad_> OK, cool :)
[22:27] <saste> vlaad_, http://pastebin.com/jqssR1w2
[22:31] <vlaad_> saste, this was not working; however, when I added unconditionally -lasound to LDFLAGS, it passed the test; so definitely, it requires libasound as prerequisite
[22:32] <saste> vlaad_, why is not working?
[22:32] <saste> now when you do --enable-libflite, lasound is automatically added to ldflags, so it should work if my understanding is correct
[22:32] <vlaad_> do you want me to put the config.log on pastie?
[22:33] <saste> wherever you want
[22:40] <vlaad_> here it is: http://pastebin.com/Y4dkppRw
[22:40] <vlaad_> I made it with shorter options, in order to make it more readable
[22:48] <saste> vlaad_, are you sure you applied the patch?
[22:49] <vlaad_> yes, that crossed my mind too :) I pulled the git again and compared, it differs by 7 lines you posted
[22:49] <saste> vlaad_, also, can you try with configure --disable-indevs --enable-libflite
[22:49] <saste> in your case even configure --enable-libflite should do
[22:50] <saste> vlaad_, you can patch configure by hand
[22:50] <saste> just replace: enabled_any alsa_indev alsa_outdev && ...
[22:51] <saste> with enabled_any alsa_indev alsa_outdev libflite && ...
[22:51] <vlaad_> this "configure --disable-indevs --enable-libflite" didn't work, now I'll do that change manually
[22:52] <saste> vlaad_, and make sure you didn't mess up configure in other parts
[22:52] <saste> do git stash to stash away all the local changes
[22:52] <saste> then patch by hand
[22:53] <vlaad_> did that, but again it fails :(
[22:54] <vlaad_> lasound is not added to the LDFLAGS
[22:58] <vlaad_> I've found what is going on :)
[22:58] <vlaad_> I'll post you different patch for configure now
[22:59] <vlaad_> I've just moved that line above the actual check :) http://pastie.org/4910828
[23:00] <vlaad_> indeed, to speak the truth, this makes it uglier and messier
[23:01] <saste> vlaad_, yes
[23:01] <saste> my patch was suboptimal
[23:01] <saste> i see why it was failing
[23:02] <saste> i'll try to think of a better fix, or i'll just push the quick'n dirty solution if it has no side effect
[23:02] <vlaad_> OK, but your patch was correct
[23:02] <saste> just tell me your name so i can mention it in the commit
[23:02] <vlaad_> Vladimir Kraljevic
[23:02] <saste> or i'll just call you "vlaad_", as you prefer
[23:02] <vlaad_> thanks :)
[23:03] <saste> you're welcome
[00:00] --- Fri Oct  5 2012