[Ffmpeg-devel-irc] ffmpeg-devel.log.20140331

burek burek021 at gmail.com
Tue Apr 1 02:05:02 CEST 2014


[00:14] <BBB> did I really just see a review where the comment was that the variable naming of x265 is poor?
[00:15] <BBB> I really just hope that didn't happen
[00:33] <jnvsor> I have a wierd problem. Audio coming into ffmpeg is gradually getting out of sync
[00:34] <jnvsor> If I set volume=1 or volume=0 with a zmq filter I hear the mic click on or off at the right time, but the audio from the mic ends up late, so the end gets cut off
[00:35] <jnvsor> Over time the whole thing gets desynced. But since the mic clicks on at the right time it must be the input right?
[01:37] <cone-372> ffmpeg.git 03Martin Storsjö 07master:3e4e2142d246: fate: Convert the paletted output in the brenderpix tests to rgb24
[01:37] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:8e8276e0bc50: Merge commit '3e4e2142d246699a1a3a0045ba7124b18bc34d7a'
[01:50] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:9bd4a9f4eec8: isom: Simplify mov_rewrite_dvd_sub_extradata()
[01:50] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:36d37edf0189: Merge commit '9bd4a9f4eec8c8843b54126253bb9d77af95bbb3'
[01:56] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:96138e8c8da4: isom: only write the size if its non zero in mov_rewrite_dvd_sub_extradata()
[01:56] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:20bc5924a0ca: Merge commit '96138e8c8da4b9397e627c934a33c22988d1192b'
[02:07] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:fce28c3c6d60: mov: fill in subtitle dimensions after parsing tkhd
[02:07] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:1ffbaa88c92b: Merge commit 'fce28c3c6d603b7f42a5dbe36bf240d0470469c8'
[02:13] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:12789d96361e: mov: call mov_rewrite_dvd_sub_extradata() after parsing dimensions from tkhd
[02:13] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:46ff8108f6b4: Merge commit '12789d96361ed8e9804bfbb9db31ea4110e75c58'
[02:17] <cone-372> ffmpeg.git 03Paul B Mahol 07master:729d821fd864: libmp3lame: allow joint stereo to be disabled
[02:17] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:801b94a78294: Merge commit '729d821fd864a0568f4068e64d3a70faa23b9309'
[02:19] <cone-372> ffmpeg.git 03Timothy Gu 07master:292dbe5e8a95: libmp3lame: add comments about CBR/VBR modes
[02:19] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:e4fc7e3d4de9: Merge commit '292dbe5e8a95d38c1324b0432fbaca67f8fe4b73'
[02:55] <cone-372> ffmpeg.git 03Timothy Gu 07master:09fda6bb503d: libmp3lame: add ABR support
[02:55] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:75621cdf5d7e: Merge remote-tracking branch 'qatar/master'
[03:43] <cone-372> ffmpeg.git 03Michael Niedermayer 07master:e818ee090ac5: avfilter/vf_pullup: fix gray8
[06:53] <haasn> Does anybody here have access to or knowledge about SMPTE 428-1-2006? In particular, I want to know the transfer characteristics they define to obtain the X'Y'Z' channels from XYZ (and vice versa)
[06:59] <haasn> Never mind, I can access them through the Google Cache
[12:28] <superware> I've been trying something similar to https://gist.github.com/krieger-od/1950618 (muxing without transcoding) but avformat_write_header returns -1094995529, any help will be highly appreciated.
[12:45] <superware> anyone?
[13:00] <nevcairiel> you should look at the log output, it might tell you whats wrong
[13:02] <superware> nevcairiel: thanks! I've managed that one after manually copying some stream/codec properties from input to output. when I play the mp4 file the video plays without any timing/syncing - as fast as it can. any ideas?
[13:03] <nevcairiel> time base of the input stream and the output stream may be different
[13:03] <nevcairiel> which makes the timestamps invalid
[13:03] <nevcairiel> you would need to rescale the timestamps
[13:03] <superware> output_stream->pts = (AVFrac) { input_stream->pts.val, input_stream->time_base.num, input_stream->time_base.den };
[13:03] <superware> why? I mean, I only want to copy the stream
[13:04] <nevcairiel> different formats may still use different timebases
[13:04] <nevcairiel> its like a one-line thing
[13:05] <nevcairiel> well two lines, one for pts one for dts
[13:05] <superware> so the line I wrote above isn't correct?
[13:06] <nevcairiel> timestamp is a field on the packet, and its a plain number, so not sure what you're trying to do there at all
[13:07] <superware> I'm trying to read an h.264 over RTP and dump it to an mp4 file
[13:07] <superware> no transcoding
[13:09] <superware> please see http://pastebin.com/MTCrZ0D5
[13:10] <superware> should it be possible to use the same AVPacket that came from av_read_frame in av_write_frame?
[13:17] <cone-739> ffmpeg.git 03Thilo Borgmann 07master:a9ccec527067: configure: Remove invalid option --disable-qtkit.
[13:29] <superware> nevcairiel?
[13:33] <cone-739> ffmpeg.git 03Peter Ross 07master:bf7d93e15783: avcodec/adpcm: also squelch 'mismatch in coded sample count' warning for AV_CODEC_ID_ADPCM_EA_R1
[13:55] <superware> nevcairiel: after adding the audio stream (to the video stream) the outputted mp4 plays the video rapidly, but the audio sounds perfect.
[14:50] <Daemon404> michaelni, ping me when you get a chance
[14:51] <michaelni> Daemon404, ping
[14:51] <JEEB> :D
[14:51] <Daemon404> lol
[14:51] <Daemon404> michaelni, ive run into an ... interesting jpeg
[14:51] <Daemon404> subsampled CMYK
[14:51] <Daemon404> im not so sure how to add it into mjpegdec
[14:51] <Daemon404> since it picks colorspace based on subsampling
[14:52] <Daemon404> and we dont have a subsampled rgb either
[14:52] <Daemon404> or it might actually be YUVA420P
[14:52] Action: Daemon404 isnt sure
[14:52] <Daemon404> imagemagick says cmyk
[14:52] <michaelni> hmm, can i get that jpeg from somewhere ?
[14:52] <Daemon404> yea sure
[14:53] <Daemon404> 1 sec
[14:53] <Daemon404> PMed it to you
[15:23] <superware> I'm trying to mux an h.264/RTP stream to mp4 without transcoding http://pastebin.com/KG7mHA7M but the resulting file video plays very fast, while the audio sounds normal. any ideas?
[15:25] <superware> michaelni can you please take a look?
[15:25] <ubitux> superware: this is not a support channel
[15:26] <superware> ubitux: I know, but I couldn't get help anywhere else (#ffmpeg)
[15:26] <ubitux> if it works with ffmpeg then you are misusing the api, then you should ask #ffmpeg or libav-user at ffmpeg.org
[15:27] <superware> retrying...
[17:22] <cone-739> ffmpeg.git 03Derek Buitenhuis 07master:32998ee9577a: libx265: Bump version requirement
[17:22] <cone-739> ffmpeg.git 03Marcus Gustafsson 07master:1b28d9b357db: libx265: Use the repeat headers flag when not using global headers
[17:28] Action: ubitux wonders why koda considers AVFILTER_DEFINE_CLASS problematic&
[17:29] <ubitux> the opw girls are going to waste so much time... :/
[17:31] <wm4> are they?
[17:31] <Daemon404> diego isnt even around
[17:31] <kierank> ubitux: really?
[17:31] <wm4> AFAIK the plan is to have them do more than cosmetics and refactoring
[17:31] <kierank> hopefully yes
[17:31] <wm4> and the current patches are just some sort of entry qualification
[17:31] <ubitux> well... so far i've seen reindents and asettb portage 
[17:32] <ubitux> asettb with requests of reindent & friends
[17:32] <ubitux> squashed into the same commit for the lulz etc
[17:32] <wm4> libav will be libav
[17:33] <ubitux> sure but well, can't they propose more fun stuff for them
[17:33] <wm4> they need to get done until tomorrow
[17:33] <wm4> 1st april is the deadline I think
[17:33] <ubitux> asking them to do the portage of ffmpeg is really ungrateful as a first task
[17:33] <wm4> lol
[17:34] <ubitux> deadline is a joke?
[17:35] <ubitux> lenovo is so sloooow at delivering hw that's insane
[17:35] <ubitux> more than one month for a cheap laptop
[17:36] <ubitux> that PoS is going around the world
[17:43] <ubitux> >  --target-path=DIR        path to view of build directory on target
[17:43] <ubitux> what does that mean?
[17:44] <ubitux> i get --target-{exec,samples} but not this one
[17:44] <Daemon404> e.g. if you build on a shared nfs dir
[17:44] <Daemon404> corss-build on x86 and run on arm
[17:44] <Daemon404> the fate run on arm will use $TARGETPATH/ffmpeg
[17:44] <Daemon404> to run stuff
[17:44] <Daemon404> for teh build dir
[17:44] <Daemon404> a lot of the fate instances rely on it
[17:44] <ubitux> but isn't the build dir on the host?
[17:45] <Daemon404> [16:44] <@Daemon404> e.g. if you build on a shared nfs dir
[17:45] <Daemon404> it's how the e.g. windows arm and aarch64 stuff runs
[17:46] <Daemon404> (on a hacked iphone)
[17:46] <ubitux> ah, mmh ok
[17:46] <Daemon404> http://fate.libav.org/armv8-apple-darwin-xcode5/20140331022134
[17:46] <Daemon404> here enjoy some eyecancer config likes
[17:47] <Daemon404> go go apple
[17:48] <ubitux> i see the trick, ok
[17:48] <ubitux> thx
[17:51] <Daemon404> :)
[17:52] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:f9059ce79441: avcodec/mjpegdec: upgrade upscale_h to support multiple planes
[17:52] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:798490812a05: avcodec/mjpegdec: upgrade upscale_v to support multiple planes
[17:52] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:559c8b9ee965: avcodec/mjpegdec: support subsampled 420 CMYK
[17:52] <michaelni> Daemon404, 420 cmyk should work
[18:16] <Daemon404> michaelni, cool
[18:50] <mankan> Hi all, I was told to push some libx265 changes but they require another patch to be pushed first. Should I push that too?
[19:00] <ubitux> https://pkgr.io/ let's try with ffmpeg \o/
[19:04] <wm4> let's rename ffmpeg to videodcdr before that
[19:36] <Daemon404> ubitux, yet another thing that is a paid webapp for no reason...
[19:36] <ubitux> :)
[19:37] <Daemon404> its not so hard to do nightly packages
[19:37] <Daemon404> or even CI
[19:39] <wm4> where does it say it's a paid service?
[19:40] <Daemon404> https://pkgr.io/pricing
[19:40] <Daemon404> kinda of
[19:40] <Daemon404> i dont really like the idea of relying on a random webservice to do that sort of stuff myself
[19:41] <wm4> if only debian packaging wasn't such an absurd clusterfuck
[20:03] <Plorkyeran_> it sounds like it's only for stuff that Heroku can handle automagically
[20:03] <Plorkyeran_> i.e. ruby libs and not much else
[20:34] <cone-739> ffmpeg.git 03Derek Buitenhuis 07master:cd60631752ca: libx265: Bump version requirement
[20:34] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:4a6dce9f9fac: Merge commit 'cd60631752cafe106b406f00966ac0323ad2ce1f'
[20:56] <Compn> wow first bayer, now cmyk? whats next???? :)
[20:56] Action: Compn gives thumbs up to michael, pross and others
[20:57] <Compn> probably drv in there and paul mahol and carl. working hard on random colorspaces. kostya did a few as well
[21:28] <Daemon404> carl hasnt done shit.
[21:30] <wm4> but what about the 43534 hacks to make shit work
[21:31] <Daemon404> if (thing_from_bug_report) return;
[21:32] <wm4> nah, not that descriptive
[21:32] <Daemon404> if (val & 0xF4A54) return -3;
[21:50] <cone-739> ffmpeg.git 03Marcus Gustafsson 07master:c042cbc3a68d: libx265: Use the repeat headers flag when not using global headers
[21:51] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:4ea333d7deb0: Merge commit 'c042cbc3a68d09a20cfd75fb4a4eff0534253081'
[21:59] <cone-739> ffmpeg.git 03Rémi Denis-Courmont 07master:332591181557: h263: improve hwaccel #ifdefs
[21:59] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:c43ebd94e423: Merge commit '33259118155764ce27a3433894659dda1d148b5f'
[22:05] <cone-739> ffmpeg.git 03Rémi Denis-Courmont 07master:654c7446a806: vc1dec: improve hwaccel #ifdefs
[22:05] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:0f714abb4985: Merge commit '654c7446a806c5dc8ec7800874c49a4c1ba2ce80'
[22:19] <llogan> only one trac spammer made it through out of that spam storm over the last few days.
[22:21] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:f94371b1db89: mov: Unbreak sbtl writing
[22:21] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:9d013fe84021: Merge remote-tracking branch 'qatar/master'
[22:33] <cone-739> ffmpeg.git 03Thilo Borgmann 07master:d5c0036d4adc: lavd/qtkit: Fix non-constant initializer element for some clang compilers.
[22:44] <llogan> Zeranoe: a "please include your full ffmpeg command and the complete console output" notice somewhere on the forum might be good to add for the 5 users who will actually read it
[22:44] <llogan> you should see how big it is on gusari forum, yet it still is missed quite often
[23:03] <superware> I'm dumping h.264 to mp4, is there a way to embed a secondary data-stream with my own custom data, so that a dedicated player will be able to extract it while playing?
[23:15] <superware> or... is it possible to create and embed a CODEC_TYPE_DATA stream while muxing to mp4?
[23:19] <cone-739> ffmpeg.git 03rogerdpack 07master:0cc8a62a78b3: dshow: fixup some COM objects, based on patches from Don Moir
[23:48] <cone-739> ffmpeg.git 03Michael Niedermayer 07master:1f361124d97c: avformat/img2dec: zero probe buffer
[00:00] --- Tue Apr  1 2014


More information about the Ffmpeg-devel-irc mailing list