[Ffmpeg-devel-irc] ffmpeg-devel.log.20151218

burek burek021 at gmail.com
Sat Dec 19 02:05:02 CET 2015


[00:01:28 CET] <durandal_1707> I want to relicense stereo3d to lgpl, I already modified bunch of code and plan to more, can I do that?
[00:02:04 CET] <nevcairiel> Everyone that contributed to it needs to agree
[00:02:05 CET] <durandal_1707> I sed nothing relevant to keep it gpl
[00:02:13 CET] <durandal_1707> *see
[00:02:33 CET] <durandal_1707> People goes missing
[00:03:51 CET] <durandal_1707> Btw I can write same thing from scratch, even anaglyph coefs can be found on the web
[00:05:23 CET] <durandal_1707> I tried to contact original creator but without success
[00:09:40 CET] <J_Darnley> Then it sounds like you're out of luck.
[00:10:04 CET] <J_Darnley> You could try tracing him through other things he might have worked on.
[00:55:25 CET] <cone-926> ffmpeg 03Michael Niedermayer 07master:ef8f6464a55d: avcodec/h264_slice: Simplify ref2frm indexing
[00:55:26 CET] <cone-926> ffmpeg 03Michael Niedermayer 07master:c8ea57664fe3: avcodec/h264_mc_template: prefetch list1 only if it is used in the MB
[02:08:40 CET] <cone-926> ffmpeg 03Michael Niedermayer 07release/2.8:cfbf608c0af6: avcodec/h264_slice: Simplify ref2frm indexing
[02:08:41 CET] <cone-926> ffmpeg 03Michael Niedermayer 07release/2.8:cabd9ae5be87: avcodec/h264_mc_template: prefetch list1 only if it is used in the MB
[04:11:09 CET] <cone-926> ffmpeg 03Marton Balint 07master:b25adbaaf67f: doc/errno: fix description typo for ENAMETOOLONG
[04:11:10 CET] <cone-926> ffmpeg 03Marton Balint 07master:752ab408b456: lavu/error: add missing error messages for errors supported on all platforms
[10:57:02 CET] <cone-452> ffmpeg 03Paul B Mahol 07master:a2a217b35119: avfilter/vf_stereo3d: add fast path for interleave rows to alternating
[10:57:02 CET] <cone-452> ffmpeg 03Paul B Mahol 07master:ec1b95dda4a8: avfilter/vf_stereo3d: fix interleave rows output
[12:45:50 CET] <rcombs> >ffserver
[14:28:02 CET] <cone-452> ffmpeg 03Vittorio Giovara 07master:be00ec832c51: lavc: Deprecate coder_type and its symbols
[14:28:03 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:2630f7f70904: Merge commit 'be00ec832c519427cd92218abac77dafdc1d5487'
[14:34:39 CET] <nevcairiel> atomnuker: ping, i found something odd in aacenc .. http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/aacenc.c;h=ec09063e84b9d1707b220fa17bc44a6ba2237d4f;hb=HEAD#l652 this line uses avctx->frame_bits before its ever written to much later in this function, maybe it was meant to use the local frame_bits variable?
[14:36:41 CET] <nevcairiel> unless it specifically wants the frame_bits from the previous frame
[14:39:24 CET] <cone-452> ffmpeg 03Vittorio Giovara 07master:16216b713f9a: lavc: Drop exporting 2-pass encoding stats
[14:39:25 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:362028cac913: Merge commit '16216b713f9a21865cc07993961cf5d0ece24916'
[14:39:41 CET] <nevcairiel> (the reason I found it is that avctx->frame_bits got deprecated)
[14:40:41 CET] <durandal_1707> still merging?
[14:40:45 CET] <nevcairiel> yes
[14:41:46 CET] <atomnuker> nevcairiel: avctx->frame_bits is the bit count over from the last frame
[14:41:49 CET] <nevcairiel> if you want to push somethign feel free, need to run fate again anyway
[14:42:02 CET] <atomnuker> unless it gets reset every frame I don't see a problem
[14:42:20 CET] <nevcairiel> atomnuker: ok then, it should probably move into AACContext, since frame_bits in avctx is deprecated now
[14:42:22 CET] <durandal_1707> why sunrastenc have utvideo class?
[14:42:56 CET] <atomnuker> nevcairiel: yeah, not a bad idea, but what's replacing avctx->frame_bits?
[14:43:04 CET] <nevcairiel> atomnuker: nothing, its going
[14:43:55 CET] <nevcairiel> durandal_1707: obviously the answer is because someone screwed up :D
[14:45:09 CET] <durandal_1707> lol, I thought it was you
[14:45:17 CET] <nevcairiel> nah, the original commit is broken
[14:46:02 CET] <nevcairiel> the context also doesnt have a avclass
[14:48:29 CET] <cone-452> ffmpeg 03Vittorio Giovara 07master:c34df422628e: sgienc: Make sure to initialize skipped header portions
[14:48:30 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:ef9ae0e748ed: Merge commit 'c34df422628e6b7b657faee241fe7bb2629e0f57'
[14:48:31 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:67ebc88fb5ba: lavc/sunrastenc: fix private codec options
[14:50:20 CET] <cone-452> ffmpeg 03Aaron Colwell 07master:febfb49a70e8: matroskadec: Fix sample_aspect_ratio for stereo matroska content
[14:50:21 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:5242cc878a7a: Merge commit 'febfb49a70e82f5ac46dc7ea34dabd4d56b19b31'
[14:52:11 CET] <cone-452> ffmpeg 03Luca Barbato 07master:34138ece23c8: log: Use a do {} while (0) for tlog
[14:52:12 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:a38b50c3ff0b: Merge commit '34138ece23c8ddae543269212a051c00d49e67d7'
[14:52:29 CET] <cone-452> ffmpeg 03Luca Barbato 07master:81c95eb8eee8: openh264: Directly include the deprecation guards header
[14:52:30 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:4a80f0bdb042: Merge commit '81c95eb8eee856d98d4ac37367dbc761f2faf875'
[14:53:05 CET] <cone-452> ffmpeg 03Dave Yeo 07master:b0b133b8c029: hevcdsp: use a macro for .rodata section
[14:53:06 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:3564dfd53574: Merge commit 'b0b133b8c02984ba0a50448f632a5dd8a50c9045'
[14:56:23 CET] <cone-452> ffmpeg 03Anton Khirnov 07master:458e53f51fc7: mpegvideo_enc: actually add the side data with vbv_delay to the packet
[14:56:24 CET] <cone-452> ffmpeg 03Hendrik Leppkes 07master:06d69a2d225a: Merge commit '458e53f51fc75d08df884f8e9eb3d7ded23e97b3'
[15:14:59 CET] <atomnuker> huh, something broke VC-2 decoding
[15:24:06 CET] <atomnuker> nope, nevermind
[15:24:59 CET] <cone-452> ffmpeg 03Paul B Mahol 07master:620b47576dec: avfilter/vf_stereo3d: add fast path for interleave rows to mono
[15:25:00 CET] <cone-452> ffmpeg 03Paul B Mahol 07master:f6d1c15b64c2: doc/filters: copy all input modes to output modes.
[15:28:55 CET] <cone-452> ffmpeg 03Rostislav Pehlivanov 07master:4386f17bbdf7: acenc: remove deprecated avctx->frame_bits use
[15:31:48 CET] <cone-452> ffmpeg 03Andreas Cadhalpun 07master:4d5c3b02e9d2: on2avc: limit number of bits to 30 in get_egolomb
[15:31:49 CET] <cone-452> ffmpeg 03Andreas Cadhalpun 07master:90b99a81071d: exr: fix out of bounds read in get_code
[15:37:42 CET] <J_Darnley> I know the chances are slim but if anyone is working on a yasm successor then make it tolerate unicode, please.
[15:41:27 CET] <kierank> J_Darnley: lol
[15:41:30 CET] <mosb3rg> hey guys
[15:41:33 CET] <mosb3rg> https://ffmpeg.org/pipermail/ffmpeg-devel/2015-March/170397.html
[15:41:51 CET] <mosb3rg> i realize this issue is from march of this year. but im curious if this is on the way to being rectified 
[15:42:19 CET] <mosb3rg> im still completely unable to utilize the -header $ cookie: function for stored hls cookie 
[15:43:21 CET] <J_Darnley> I have no idea of anything other than that page says the some patch was applied.
[15:43:34 CET] <mosb3rg> well yes but.. the point im making is this function still doesnt work
[15:43:43 CET] <mosb3rg> hls -headers being noted for cookie value 
[15:44:06 CET] <mosb3rg> if im using ffplay, i can without any issues note the header for the cookies and it will playback fine.
[15:44:26 CET] <mosb3rg> but when it comes to the ffmpeg ability to dump and relay the signal theres no means any further to note the -headers 
[15:45:09 CET] <mosb3rg> since its a function issue, this seems more a development question. because its not a function in the manual or help anymore.. but at one stage it was a predominant feature
[15:45:46 CET] <mosb3rg> if you look around on some forums, there are some who have seemed to patch it. but are far to secretive and refuse to share the patch or show there work.
[16:28:18 CET] <cone-452> ffmpeg 03Clément BSsch 07master:91b4afd58d78: swscale/arm/yuv2rgb: fix slicing
[16:34:02 CET] <ubitux> do we have a FFALIGN macro equivalent in x86utils & friends?
[16:40:52 CET] <J_Darnley> ubitux: not that I'm aware of (for either preprocessor or instructions)
[16:41:23 CET] <J_Darnley> and searching in the file include files doesn't reveal one
[16:43:50 CET] <J_Darnley> What do you want to round up to the next power of 2?
[16:44:55 CET] <ubitux> well, ffalign, so just a ADD and a AND
[16:47:45 CET] <J_Darnley> Sure, that'll work fine with a constant
[16:48:15 CET] <J_Darnley> and I guess any other register
[17:47:41 CET] <andrey_utkin> Could anybody help me to analyze the brokenness of generated H.264 P-frames? I develop a driver for hardware encoder, and make up headers manually. Having NAL parsing results down to elementary property (each) would be great. Ready to pay for any help. Please contact me.
[18:08:38 CET] <wooo> I'm trying to work around or fix a "bug" where video recorded from a camera streamed through a live555 rtsp proxy has messed up timestamps that cause playback problems later.
[18:09:05 CET] <wooo> The behavior of the timestamps seems intentional by the proxy according to : http://www.live555.com/liveMedia/faq.html#rtcp-synchronization-issue
[18:09:25 CET] <cone-452> ffmpeg 03Michael Niedermayer 07master:1c878474fb1d: avcodec/ffv1enc: unbreak -coder option
[18:12:36 CET] <wooo> the result when using the hls segmenter for output is that the first segment has a crazy long duration because the ots timestamp sometimes jumps from the first few frames to a value sometimes 4-12 hours later
[18:15:00 CET] <wooo> My thoughts on trying to fix it are to add code that either throws out all frames received before the SR packet is received (assuming it is?) or to try to artificially set the ots for those frames once the stable pts is received.
[18:15:47 CET] <wooo> any thoughts on how to approach that would be appreciated!  and if the mailing list is a better place to ask, I'll do that as well.
[18:18:20 CET] <wooo> (Another vaguely related bug is that something isn't handling pts rollovers properly, which results in the last segment growing arbitrarily large as something keeps "correcting" the suddenly non monotonicly increasing pts by making it 1 unit more than the previous packet. )
[19:24:24 CET] <mjbshaw> I've got a patch on the mailing list that seems to have fallen through the cracks: http://ffmpeg.org/pipermail/ffmpeg-devel/2015-November/183428.html
[19:25:00 CET] <mjbshaw> Is there a good way to revive that thread? My last ping was ineffective and I don't want to unnecessarily spam the mailing list.
[19:35:14 CET] <jamrial> mjbshaw: that thread has some replies, with michaelni's being the last it seems
[19:35:34 CET] <jamrial> it's also a month old, so you can ping it again just fine. it would hardly be considerd spamming
[19:36:37 CET] <mjbshaw> jamrial: I pinged it Dec 2 (http://ffmpeg.org/pipermail/ffmpeg-devel/2015-December/184407.html), but I'll go ahead an ping it again. Thanks.
[19:48:50 CET] <cone-452> ffmpeg 03Paul B Mahol 07master:aefcc77b9051: avfilter/vf_stereo3d: multiply linesize only once for interleaved row to mono
[20:04:39 CET] <ubitux> michaelni: any idea why sth like http://b.pkh.me/0001-avfilter-scale-WIP-brightness-contrast-saturation.patch seems to have no effect whatsoever on the output?
[20:04:49 CET] <ubitux> i tried random value in [0;1]
[20:05:05 CET] <ubitux> the settings are indeed passed to sws, but without any change
[20:05:13 CET] <ubitux> well i'm probably missing sth obvious
[20:27:37 CET] <durandal_1707> kierank: when you gonna sent cfhd code?
[20:28:29 CET] <kierank> durandal_1707: going to work on it over christmas
[20:28:38 CET] <Compn> RE takes a while 
[20:29:02 CET] <durandal_1707> but send wip?
[20:31:25 CET] <kierank> durandal_1707: github OK?
[20:32:04 CET] <durandal_1707>  yes
[20:32:20 CET] <durandal_1707> :)
[20:34:18 CET] <kierank> OK will send during weekend
[20:34:20 CET] <kierank> Pub now
[21:47:34 CET] <michaelni> ubitux, dont really know, you could look at sws_setColorspaceDetails() what/if anything is set and then what convertion path is taken
[22:05:28 CET] <cone-452> ffmpeg 03Paul B Mahol 07master:8d6b1ffba0ca: avfilter/vf_stereo3d: fix interleaved columns output
[23:19:18 CET] <Daemon404> nevcairiel, i love your users: forum.doom9.org/showpost.php?p=1749674&postcount=20083
[23:19:33 CET] <Daemon404> (that guy had some epic threads 10 years ago, i rememebr his nick(
[23:20:12 CET] <nevcairiel> is that the xp guy?
[23:20:27 CET] <nevcairiel> yes it is
[23:20:30 CET] <nevcairiel> what an idiot
[23:20:40 CET] <Daemon404> you knew before you even opened the link
[23:20:42 CET] <Daemon404> <3
[23:20:59 CET] <Daemon404> i was also unaware that installing SP3 == new os
[23:21:01 CET] <nevcairiel> well he was the highlight of recent idiots
[23:44:41 CET] <RiCON> how long does it take people to realize newer software isn't supposed to work with old OS forever?
[23:46:50 CET] Action: thardin submits bug report about ffmpeg not working on MS-DOS 6.2
[23:47:12 CET] <Mavrik> RiCON, hopefully as long as possible.
[23:47:13 CET] <Mavrik> :)
[23:47:38 CET] <Daemon404> thardin, i swear we used to have a buildbot on freedos...
[23:47:42 CET] <Daemon404> but i might be confused.
[23:47:51 CET] <thardin> (actually, DOS support would be kinda neat)
[23:48:03 CET] <thardin> I mean.. it's a shell
[23:48:52 CET] <thardin> I could resurrect my old DOS setup, convert those syndicate and UFO videos to some other format on it
[23:49:24 CET] <Mavrik> On a machine without MMX if possible :P
[23:49:45 CET] <thardin> of course. I had an SX, so not even float ops
[23:52:14 CET] <thardin> .. or I could get back to working on this programmable load
[23:57:10 CET] <cone-452> ffmpeg 03Rostislav Pehlivanov 07master:591b240ed486: oggparsedaala: sync with current bitstream syntax
[00:00:00 CET] --- Sat Dec 19 2015


More information about the Ffmpeg-devel-irc mailing list