[Ffmpeg-devel-irc] ffmpeg-devel.log.20150113

burek burek021 at gmail.com
Wed Jan 14 02:05:02 CET 2015


[00:03] <pross> i having nothing against it. ffmpeg is supernova, it has almost consumed mplayer
[00:17] <cone-265> ffmpeg.git 03Michael Niedermayer 07master:24222cc1d5ca: avcodec/opus_imdct: Use av_malloc_array()
[00:17] <cone-265> ffmpeg.git 03Michael Niedermayer 07master:13871a95d0e4: avcodec/snow: Use av_malloc_array()
[00:17] <cone-265> ffmpeg.git 03Michael Niedermayer 07master:56c7e1059ab9: avcodec/snow: Fix av_malloc* failure checks
[00:17] <cone-265> ffmpeg.git 03Michael Niedermayer 07master:fbe8672e158a: avcodec/shorten: use av_reallocp_array()
[00:24] <aetasx> will --disable-encoder take a list or is this going to be one really long command line?
[00:25] <ubitux> i like the idea of dynamic loading for distro
[00:25] <ubitux> because this sucks typically:
[00:26] <ubitux> Depends On     : alsa-lib  bzip2  fontconfig  gnutls  gsm  lame  libass  libvdpau  libbluray  libmodplug  libpulse  libtheora  libva  libvorbis  libvpx  opencore-amr  openjpeg  opus  rtmpdump  schroedinger  sdl  speex  v4l-utils  libx264  xvidcore  zlib  x265  fribidi
[00:26] <aetasx> yeah optionals would be nice
[00:26] <ubitux> everytime an app has a small ff dep, all of this is pulled; having all of them optional would be cute..
[00:27] <ubitux> now how to do that properly in practice, i have no idea :)
[00:28] <aetasx> how about my configure list question?  any clue?
[00:28] <aetasx> trying to do stripped down builds
[00:30] <llogan> i believe it will accept a comma separated list
[00:33] <ubitux> you might need a 's'
[00:33] <ubitux> but yeah, comma separated list
[00:33] <ubitux> you can also --disable-everything and --enable-encoder=...
[00:36] <aetasx> yeah that was what I was going to do but that doesn't really help the situation
[00:37] <aetasx> I see a --disable-decoder=NAME and --disable-decoders, lemme check their implementation but their description doesnt really allude to a list
[00:41] <aetasx> looks like it'll take a list
[01:29] <BBB> Compn: even in the hypothetical case that you werent wrong, the fact we had to use a fork is kinda telling, isnt it
[01:29] <BBB> (wrong about rtmpdump vs. librtmp; I dont actually know)
[02:13] <cone-265> ffmpeg.git 03Mark Reid 07master:b23a8668448d: libavformat/mxfdec.c: export the full UMID as metadata
[02:53] <compn> BBB : is like saying xorg 'didnt work' because it was a fork :P 
[02:54] <compn> pross : vfw / binary codec in ffmpeg is a requested feature and many users would appreciate it.
[02:54] <compn> especially youtube would appreciate binary codec loader :P
[03:00] <BBB> I just went back to my re wmavoice code, just to see how I did that again
[03:00] <BBB> such a massive hack :D
[03:00] <BBB> (loading the binary dll into ffplay for comparison)
[03:55] <compn> BBB : did you use current wine or ancient mplayer crap ? 
[03:55] <BBB> ancient mplayer crap
[03:55] <compn> ehehe
[03:55] <BBB> it worked[tm]
[03:55] <compn> yes, mplayer crap works forever. thats why its still around, nothing major has broke
[03:57] <compn> i never understood why vlc didnt go loading binaries, even on win32 they only load dmo , and only wmv3 and something else
[04:02] <BBB> arent all codecs REed nowadays?
[04:05] Action: BBB zzz
[04:09] <wm4> <compn> yes, mplayer crap works forever. thats why its still around, nothing major has broke <- because it was already broken?
[04:59] <compn> BBB : but what about prores 2? vc99? h266? etc etc. to answer your question, i think theres about 500 binary codecs (video) of which 200 are in mplayer (and about 100 are RE'd)
[05:02] <compn> wm4 : it seems to "just work" for a lot of people
[05:03] <compn> with ubuntu using pulse  not to mention composite desktop features, mplayer does get broken :P
[05:16] <wm4> compn: they typically use the subset that happens to work, while ignoring everything else
[05:16] <wm4> while everything else makes maintenance actively harder (and still gets fixes even though development of it stopped years ago)
[08:16] <Timothy_Gu> didn't know irssi crashed until now
[08:17] <Timothy_Gu> is there a way to make irssi start on boot in Ubuntu?
[08:20] <ubitux> https://github.com/GutenYe/systemd-units/blob/master/irssi/irssi%40.service
[08:24] <JEEBsv> while you can switch to a systemd-based boot with 14.10 i don't think that's the default :)
[08:26] <Timothy_1u> :quit
[08:26] <Timothy_1u> too much vim recently
[10:32] <ubitux> why is avctx->me_method on "MUST be set by user" for encoding?
[10:34] <rcombs> almost certainly "historical reasons"
[10:34] <nevcairiel> histerical reasons?
[10:34] <rcombs> hysterical raisins!
[12:04] <anshul_mahe> mailing list of libnut
[12:05] <anshul_mahe> sry any one know here mailing list of libnut, it points out ffmpeg, libav and mplayer on google
[13:26] <kierank> anshul_mahe: have you hought about what a decoder does when it sees captions in the ES and captions in nut?
[13:34] <cone-202> ffmpeg.git 03Stefano Sabatini 07master:57ede2a5070f: lavfi: use ff_norm_qscale(), factorize
[13:34] <cone-202> ffmpeg.git 03Arwa Arif 07master:8c7e512ad9f8: lavfi: remove mp=pp7
[13:34] <cone-202> ffmpeg.git 03Stefano Sabatini 07master:c648879827d3: Changelog: mention the port of various mp PP filters
[13:36] <wm4> wow only 4 filters left
[13:36] <wm4> 2 of them doing the same (vf_eq/vf_eq2)
[13:36] <ubitux> and ilpack being useless apparently
[13:36] <ubitux> ...and softpulldown as well
[13:42] <kierank> ubitux: i think you can remove ilpack now
[13:43] <wm4> what replaces ilpack?
[13:43] <ubitux> scale apparently
[13:43] <ubitux> automatically/transparently?
[13:44] <ubitux> so mmh... -vf format=yuv422p will work as expected?
[13:44] <kierank> with forced interlaced scaling yes
[13:45] <saste> ubitux, about eq/eq2, James Darnley plan was to merge it with lut
[13:45] <ubitux> lut?
[13:45] <ubitux> you mean hue?
[13:45] <ubitux> kierank: -vf scale=interl=1,format=yuv422p ?
[13:45] <wm4> somehow I think not using a lut would be faster for any such transformation...
[13:45] <kierank> ubitux: i believe yes
[13:46] <saste> ubitux, yes
[13:46] <ubitux> is he working on it or..?
[13:47] <saste> ubitux, I don't think so
[13:47] <saste> I have his reply to arwa, he will leave if arwa or someone else will do it in his place
[13:47] <saste> i don't know if he wrote code or something
[13:48] <iive> i agree about ilpack.
[13:48] <saste> i think we can safely do the work in his place if we want to have eq and drop mp
[13:48] <iive> as for eq vs eq2, I think that eq2 is strictly superset of eq.
[13:49] <iive> meaning eq have no benefits over eq2
[14:30] <anshul_mahe> kierank: yes It works well, I am using that with bmdcapture
[14:31] <anshul_mahe> I export cc608 stream in nut
[14:31] <kierank> yes but what happens if nut contains h264 with captions in the ES and also a caption stream
[14:32] <anshul_mahe> in caption is in h264 then, you have to specifie it with movie command
[14:32] <anshul_mahe> *command = filter
[14:34] <anshul_mahe> so if user have both and does not use movie filter then he would get one in nut stream
[14:34] <anshul_mahe> so I dont see any case when it get jumbeled up
[14:35] <kierank> for ffmpeg yes
[14:35] <kierank> but for vlc for example
[14:35] <kierank> which stream should vlc decode?
[14:36] <anshul_mahe> vlc will never know that caption are there in h264
[14:36] <kierank> yes it does
[14:36] <kierank> vlc has a caption decoder
[14:36] <anshul_mahe> sry, vlc will never know if it using ffmpeg decoder
[14:37] <kierank> you are totally missing the point
[14:37] <anshul_mahe> may be
[14:37] <kierank> what is a hypothetical player meant to do if it sees two captions streams
[14:37] <kierank> one in the ES and one in the container
[14:39] <anshul_mahe> it should see both of them as different caption stream, for same program
[14:40] <anshul_mahe> I know that ATSC caption specs does not specify anything about muxers, it say only about h264
[14:40] <kierank> exactly
[14:40] <kierank> so you're breaking it already
[14:40] <kierank> the only saving grace is that nobody uses nut
[14:41] <anshul_mahe> but I have already seen lots of video with caption as stream even in ts
[14:41] <kierank> nope
[14:41] <anshul_mahe> in market, may be I can find one for u
[14:41] <kierank> in ts it's in the ES
[14:41] <kierank> apple may have their own proprietary crap
[14:41] <anshul_mahe> yes, they are in ts, but to prove I have to give u video
[14:42] <anshul_mahe> I have very old version of caption spec, so u have any new version that u can share with me, I would re read it
[14:43] <anshul_mahe> may they have added something
[14:43] <kierank> as I said captions go in an ES
[14:43] <kierank> whether ffmpeg likes it or not
[14:43] <anshul_mahe> ISDB(japanese standard) caption always go in ts 
[14:44] <anshul_mahe> I can share that spec with u, if u want
[14:44] <anshul_mahe> but its not e608
[14:48] <anshul_mahe> its arib and also make no point
[15:46] <akira4> ubitux, do you have a minute?
[15:46] <ubitux> akira4: i'll have time in about 5 hours
[15:47] <akira4> oh. Okay then.
[15:47] <ubitux> you can still ask your question
[15:48] <ubitux> akira4: about the repo, for the whole ffmpeg repository
[15:48] <ubitux> s/for/please clone/
[15:48] <akira4> okay. 
[15:49] <akira4> well, I was trying the protocol code and when I set the analyzeduration and probesize values very high only then do I get the subtitle streams
[15:49] <ubitux> i see; yeah it's probable the subtitles appear later
[15:50] <akira4> so then its not something that we can detect with the code then?
[15:53] <ubitux> it's done somewhere in libavformat/mpeg.c, don't worry about it
[15:53] <ubitux> so are you able to remux a few minutes of the dvd with the subtitles already?
[15:54] <ubitux> with something like... ffmpeg -i dvd://foo.iso -c copy -t 20:00 out.mkv
[15:54] <akira4> I haven't tried it out yet. I
[15:54] <akira4> I'll try it and clone the repo
[15:58] <wm4> the protocol approach won't work here
[15:58] <wm4> the dvd info tells you how many subtitle streams there are (and their language etc.), but you somehow have to do this on a slightly higher level
[15:58] <wm4> like on the demuxer level
[16:12] <ubitux> wm4: you can't actually
[16:20] <wm4> ubitux: huh?
[16:21] <compn> manually scan the .ifo ... ehehe
[16:21] <ubitux> wm4: the api is too low level
[16:21] <wm4> ubitux: ?
[16:22] <ubitux> afaiu you can't easily map what you get out of the libdvd* to AVPackets
[16:23] <ubitux> you can't even create proper av streams
[16:23] <wm4> I do it specifically for subtitles
[16:23] <ubitux> ah?
[16:24] <wm4> while leaving audio/video to the mpeg demuxer
[16:24] <ubitux> mmh
[16:26] <ubitux> how do you check if you get subtitles data when reading?
[16:26] <wm4> you mean mapping the ifo stream number ot the mpeg one?
[16:26] <ubitux> i have a hard time guessing how you are actually doing this
[16:27] <wm4> adding 0x20 to the subtitle index gives you the mpeg id
[16:35] <cone-202> ffmpeg.git 03Michael Niedermayer 07master:482761b85a68: avcodec/snowenc-test: check for av_malloc*() failures
[16:35] <cone-202> ffmpeg.git 03Michael Niedermayer 07master:88a46fa8c9e5: avcodec/snowenc-test: use av_mallocz_array()
[16:35] <cone-202> ffmpeg.git 03Michael Niedermayer 07master:295b79b5d8c0: avcodec/roqvideoenc: Check for av_malloc_array() failure
[17:05] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:0775653b4c47: avformat/segment: Use av_freep() avoid leaving stale pointers in memory
[17:05] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:6ac156864ca5: avformat/cdxl: Fix integer overflow of image_size
[17:05] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:d9ec3070d18d: avformat/flvdec: do not inject dts=0 metadata packets which failed to be parsed into a new data stream
[17:05] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:54e3aff10e2d: avformat/flvdec: Increase string array size
[17:05] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:efdd30df0697: avfilter/vf_sab: fix filtering tiny images
[17:05] <cone-202> ffmpeg.git 03wm4 07release/2.4:e2e145db8991: avcodec/dvdsubdec: fix out of bounds accesses
[17:05] <cone-202> ffmpeg.git 03Dale Curtis 07release/2.4:134ff88c6a80: mov: Avoid overflow with mov_metadata_raw()
[17:05] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:0787163cf369: avformat/mov: fix integer overflow in mov_read_udta_string()
[17:05] <cone-202> ffmpeg.git 03Dale Curtis 07release/2.4:54b76eb59515: mov: Fix negative size calculation in mov_read_default().
[17:05] <cone-202> ffmpeg.git 03Paul B Mahol 07release/2.4:20a4cf0d0970: lavfi: check av_strdup() return value
[17:05] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:63007901446f: avformat/matroskadec: Use av_freep() to avoid leaving stale pointers in memory
[17:05] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:45069582897c: avformat/utils: Clear pointer in ff_alloc_extradata() to avoid leaving a stale pointer in memory
[17:06] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:fe457ce4d169: cmdutils: Use 64bit for file size/offset related variable in cmdutils_read_file()
[17:06] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:86f4ac93659f: avformat/mov: Fix mixed declaration and statement warning
[17:06] <cone-202> ffmpeg.git 03James Almer 07release/2.4:51e880fed97f: configure: bump year
[17:06] <cone-202> ffmpeg.git 03wm4 07release/2.4:2ba1af5791fc: avcodec/dvdsubdec: error on bitmaps with size 0
[17:06] <cone-202> ffmpeg.git 03wm4 07release/2.4:3b4e9dddcff6: avcodec/dvdsubdec: fix accessing dangling pointers
[17:06] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:07c9df792cd4: ffmpeg: Clear error message array at init.
[17:06] <cone-202> ffmpeg.git 03wm4 07release/2.4:1b3332659a77: vp9: fix parser return values in error case
[17:06] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:0b0293ceafd6: avformat/movenc: workaround bug in "PathScale EKOPath(tm) Compiler Suite Version 4.0.12.1"
[17:06] <cone-202> ffmpeg.git 03Andreas Cadhalpun 07release/2.4:9cc1d21e76b5: doc/examples: fix lib math dep for decoding_encoding
[17:27] <cone-202> ffmpeg.git 03Michael Niedermayer 07release/2.4:1fd7fb9036fc: Update for 2.4.6
[18:22] <cone-202> ffmpeg.git 03wm4 07master:921706691a87: qpeg: avoid pointless invalid memcpy()
[20:26] <llogan> michaelni: i'll be gone this month from 15-24. Not yet sure if I'll have internet access though, so you or compn may have to clear the ML queue.
[20:45] <cone-202> ffmpeg.git 03Michael Niedermayer 07master:e172f5e53ae4: avcodec/hevc: Fix handling of skipped_bytes() reallocation failures
[21:08] <llogan> i wonder why badcontent didn't block that last one. i'll probably never know now that the spam monitor is being flooded for the last week or so
[21:09] <llogan> they have quite the botnet
[22:35] <cone-202> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n2.4.6': unknown revision or path not in the working tree.
[22:35] <cone-202> Use '--' to separate paths from revisions
[22:35] <cone-202> refs/tags/n2.4.6:HEAD: avcodec/hevc: Fix handling of skipped_bytes() reallocation failures
[00:00] --- Wed Jan 14 2015


More information about the Ffmpeg-devel-irc mailing list