[Ffmpeg-devel-irc] ffmpeg-devel.log.20130305

[burek]

[01:04] <barque> I know this is the wrong channel to ask but.... is there some parameter in ffmpeg to reduce stalls in videos? I'm downscaling 1 ogg video to another and the output video is ending up having slight stalls in the middle
[01:14] <barque> both #ffmpeg-devel and #ffmpeg are dead ... is that too much to ask at 7:11 PM EST?
[01:25] <burek> hi guys :) am i missing something or there is no docs entry for internal aac codec ?
[01:25] <burek> i was looking here: http://www.ffmpeg.org/ffmpeg-codecs.html
[01:28] <barque> burek, do you ever get stutters in your videos after converting with ffmpeg?
[01:29] <saste> burek, yes
[01:29] <saste> and why do you want internal documentation?
[01:29] <barque> what about you saste?
[01:29] <burek> saste, for the -cutoff param info
[01:29] <burek> i didn't know it was present at all
[01:29] <saste> barque, ask on #ffmpeg
[01:30] <barque> channel's dead
[01:30] <burek> barque not quite
[01:30] <burek> (continue to #ffmpeg)
[01:31] <saste> burek, oh i see
[01:32] <saste> burek, you may create a ticket
[01:32] <burek> ok
[01:32] <saste> i have no idea what those options are good for
[01:32] <burek> i guess it cuts freq above X KHz
[01:53] <Compn> wow
[01:53] <Compn> this program...
[01:53] <Compn> durandal_1707 better be happy
[01:55] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:3dbc0ff9c3e6: iff: fix integer overflow
[02:06] <Compn> had to install vista sp2, install a dx hotfix, install dx11, enable aero just to get the stupid screen recording program to work
[02:24] <burek> i installed xp on a quad core + ssd hard disk
[02:24] <burek> it runs 10x (literaly) faster than any other vista/win7/8 crap
[02:37] <Compn> obviously
[02:37] <Compn> vista is designed to sell computers, because it slows them down and you now need a 'new one'
[02:37] <Compn> :)
[02:37] <burek> that's so true, no matter how funny it sounds :)
[02:38] <Compn> (i had vista on this box, just had to install the vista sp2...)
[02:38] <Compn> but . also yay
[02:40] <burek> http://www.youtube.com/watch?v=FVbf9tOGwno
[02:40] <burek> lolita :)
[02:40] <Compn> i got this codec working in mplayer
[03:02] <Compn> highgod : can you compile mplayer for me with your patch applied to ffmpeg ? i want to see if i can then use dxva2 decoder in mplayer :)
[03:03] <Compn> i guess i could do it. just have to find out which mplayer revision will compile with your patch and ffmpeg :\
[03:03] <Compn> stupid api changes....
[03:06] <burek> is it difficult to sit down and create an appropriate interface
[03:06] <burek> for the future i mean
[03:07] <Compn> yeah but rename codec_id to av_codec_id ....
[03:07] <Compn> and something other rename that screws compilation
[03:08] <Compn> so not api changes, but api name changes
[03:08] <burek> yes i see
[03:50] <highgod> Hi,Compn,are you online now?
[03:51] <highgod> I can offer you the mpc version, but it is the official version
[03:52] <highgod> Compn:sorry I am out and I didn't see your message
[03:54] <Compn> its ok
[03:55] <Compn> i am here
[03:55] <Compn> but i think i'm asking for too much. dont worry about it :)
[03:55] <Compn> we should focus on getting your patch into ffmpeg first.
[03:56] <highgod> OK,hehe,I re-submit the patch yesterday according the latest comments
[03:57] <highgod> about mpc, we only have the official version
[03:59] <highgod> do you want .exe file or the source code witch can be compiled?
[04:16] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:e398990eb877: msrledec: fix output_end checks
[04:34] <highgod> Compn:are you on line?
[04:51] <cone-714> ffmpeg.git 03Carl Eugen Hoyos 07master:d043461344f2: Add h264chroma dependency for mpegvideo / lowres to configure.
[04:51] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:715c2308110d: Merge remote-tracking branch 'cehoyos/master'
[05:15] <burek> does ffmpeg have a filter that implements loudness normalization like this one http://en.wikipedia.org/wiki/ReplayGain
[05:17] <ubitux> there is a patchset pending for this
[05:17] <ubitux> but you can already do it in two pass
[05:17] <ubitux> using ebur128
[05:19] <burek> cool
[05:19] <burek> we could create a wiki article for such thing
[05:19] <burek> it might be useful
[09:21] <cone-714> ffmpeg.git 03Michael Niedermayer 07release/1.1:7c8beec48ce7: buildsys: only include log2_tab per library for shared builds
[10:56] <burek> is this ok to close bug trac tickets like this, without any explanation: https://ffmpeg.org/trac/ffmpeg/ticket/2326#comment:2
[10:56] <durandal_1707> saste: last oportunity to comment stereo3d
[10:56] <burek> it kinda makes me not want to submit any more tickets when they get handled like that
[10:56] <durandal_1707> burek: going to check it
[10:58] <saste> burek, +1
[10:58] <nevcairiel> it would've been funnier if he asked for uncut ffmpeg output
[10:58] <nevcairiel> :D
[10:58] <av500> Options which do not take arguments are boolean options, and set the corresponding value to true. They can be set to false by prefixing with "no" the option name, for example using "-nofoo" in the command line will set to false the boolean option with name "foo". 
[10:58] <av500> -stats (global)
[10:58] <saste> burek, ^^ that's the explanation
[10:59] <av500> Print encoding progress/statistics. On by default. 
[10:59] <av500> thus, nostats turns it off
[10:59] <burek> saste, that's great :) but still doesnt approve such a behavior on trac
[10:59] <burek> also this https://ffmpeg.org/trac/ffmpeg/ticket/2324#comment:2
[11:00] <burek> i sometimes have a feeling that guy is so eager to close tickets, that he might not even read them properly
[11:00] <burek> i've spent half an hour with a guy on #ffmpeg yesterday and managed him to create the tickets with media samples
[11:00] <burek> only to see today those got closed as "duplicates" not even referencing what is it a duplicate of
[11:01] <burek> that kind of handling tickets really makes people not wanting to spend their time anymore submitting tickets...
[11:01] <burek> and should be discouraged
[11:01] <saste> burek, carl is sometimes around here on irc
[11:01] <saste> did you try talking to him
[11:02] <burek> over email couple of times..
[11:02] <cbsrobot_> burek: just get over it. cehoyos does a graet job and so do you - somedays you make the extra effort and someday you don't, nothing to be mad about
[11:02] <av500> +1
[11:02] <av500> trac is not to exchange pleasantries
[11:02] <burek> im nod mat about it, im almost ignorant right now, but i kinda make fool of myself
[11:03] <burek> buncing people to somewhere where they will get treated like that
[11:03] <burek> it would be the best that tickets at least dont get closed if there is no proper explanation/resolution to them
[11:04] <cbsrobot_> I guess he values your input and he's challanging you to discover the dark places of ffmpeg &
[11:04] <burek> :)
[11:04] <burek> i've seen he has helped a lot but this is really not helping anyone
[11:05] <burek> im not trying to blame him directly, just the attitude that could be better, that's all
[11:05] <cbsrobot_> burek: I agree.
[11:06] <cbsrobot_> but it take effort, that somedays he/you and me don't have &.
[11:07] <cbsrobot_> burek: maybe he was looking at swscale for a few minutes & try to do the same and be polite afterwards :-D
[11:08] <burek> :)
[11:13] <durandal_1707> there some kind of cult developing
[11:20] <durandal_1707> saste: before you disappear again, you gonna comment or not?
[11:22] <saste> durandal_1707, yes
[11:22] <saste> durandal_1707, what kind of cult?
[11:24] <durandal_1707> saste: i mean to comment/review patch...
[11:27] <cone-714> ffmpeg.git 03Paul B Mahol 07master:f5ce359ce454: doc/filters: fix short options for noise filter
[11:35] <durandal_1707> wow, i made decoder twice faster
[11:38] <cone-714> ffmpeg.git 03Paul B Mahol 07master:d7f9f34dc683: zmbv: stop doing colorspace conversion
[11:47] <durandal_1707> saste: i give you 8 hours to reply, i have other patches pending for that filter....
[11:51] <cbsrobot_> durandal_1707: what decoder ?
[11:53] <durandal_1707> cbsrobot_: ???
[11:54] <durandal_1707> cbsrobot_: ohh, I forgot it about in 4 seconds, zmbv
[11:54] <cbsrobot_> hehe
[11:54] <cbsrobot_> ok
[11:54] <durandal_1707> i just replaced bad swscale clone with memcpy
[12:17] <cone-714> ffmpeg.git 03Janne Grunau 07master:15ea1ac695d7: fate: remove last incomplete frame from mpeg2-field-enc test
[12:17] <cone-714> ffmpeg.git 03Diego Biurrun 07master:813b82b927b1: configure: Fix silly typo in logging command of check_struct()
[12:17] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:06ad309ab4b7: Merge commit '813b82b927b19532e3bc3397b972c8a27b9f6326'
[12:23] <cone-714> ffmpeg.git 03Diego Biurrun 07master:f868b24d0f28: configure: Add sanitize_var_name convenience function
[12:23] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:890b9c5c466c: Merge commit 'f868b24d0f280bb966e5c352028a343e45d20a02'
[12:28] <cone-714> ffmpeg.git 03Diego Biurrun 07master:9c7d85b3b45a: configure: Add check_builtin convenience function
[12:28] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:1cb3ca2a67fe: Merge commit '9c7d85b3b45a77742b1e69c48ab33f9c8a2e6f35'
[12:44] <cone-714> ffmpeg.git 03Diego Biurrun 07master:2a7ab2dfaa7d: configure: Use check_builtin() where appropriate
[12:44] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:1567095acf03: Merge remote-tracking branch 'qatar/master'
[12:47] <cone-714> ffmpeg.git 03Paul B Mahol 07master:d2b0b839b9e3: fate: add dependency for xface
[13:17] <cone-714> ffmpeg.git 03Alexander Kojevnikov 07master:29d8cd265a53: mp3dec: Fix VBR bit rate parsing
[13:22] <cone-714> ffmpeg.git 03Stefano Sabatini 07master:aba18c5af9e7: tests: add overlay rgb and yuv444 tests
[13:22] <cone-714> ffmpeg.git 03Stefano Sabatini 07master:7b9a3df5d716: lavu/opt-test: do not explicitly set default value in test_ctx
[13:22] <cone-714> ffmpeg.git 03Stefano Sabatini 07master:b1a268493648: lavu/opt: fix weird align
[13:44] <saste> michaelni,  av_reverse is deprecated
[13:45] <saste> how should we handle that?
[13:45] <saste> putting back av_reverse to its proper place?
[13:45] <saste> removing the deprecation?
[14:17] <cone-714> ffmpeg.git 03Nicolas George 07master:3ab25e45744c: Revert "lavf/concatdec: define "ffcat" and "ffconcat" extensions"
[14:34] <cone-714> ffmpeg.git 03Nicolas George 07master:622a6f6f3b8f: doc/muxers: specify that tee needs -map.
[14:35] <michaelni> saste, dunno, the only problem is 1 warning ? 
[14:39] <saste> michaelni, yes but when libav dump they will also remove the table from lavu
[14:40] <saste> and move it to lavc
[14:40] <saste> but wait, i need to look better
[14:40] <saste> uhm: *when libav bumps
[14:41] <durandal11707> just reaname it to ff_ then
[14:41] <michaelni> we can make the table static and move to eval.c
[14:42] <durandal11707> but does such table really help?
[14:49] <saste> av_reverse is used by root()
[14:50] <durandal11707> i know that, but i was not asking is it user or not....
[15:20] <nevcairiel> in any case i would advise against trying to load it from one lib for usage in another, so if you must use it in two libs, i would vote for duplication, maybe the same way ff_log2_tab works now
[15:22] <Daemon404> saste && ubitux: ping on ffprobe patch
[15:32] <cone-714> ffmpeg.git 03Stefano Sabatini 07master:ad4855613520: lavu/eval: add native support to lte and lt functions
[15:49] <cone-714> ffmpeg.git 03Nicolas George 07master:9b211c43dc5f: examples/muxing: fix memory leak.
[16:11] <saste> durandal11707, is odd size supported by stereo3d?
[16:16] <durandal11707> depends...
[16:50] <cone-714> ffmpeg.git 03Nicolas George 07master:c31458c334b6: lavf/concat: compute duration if possible.
[16:56] <ubitux> http://imgur.com/w8sQ4gw pseudo-vintage effect with ffmpeg
[16:56] <ubitux> ’ mp=eq2=1:1.2,hue=h=20,curves=r='0.4/0.2 0.9/1':g='0.75/0.8':b='0/0.1 1/0.9'
[16:56] <ubitux> will send a patch soon
[16:57] <ubitux> Daemon404: are they the only settings that could be altered?
[17:04] <durandal11707> i will apply stereo3d and mp=stereo3d drop in single commit
[17:05] <ubitux> durandal11707: why?
[17:05] <saste> ubitux: nice
[17:06] <durandal11707> because of nits+++++++++++, that take exponential time to rebase
[17:06] <ubitux> durandal11707: i mean you are squashing the 2 commits?
[17:06] <saste> btw does someone know how to add salt&pepper noise?
[17:06] <saste> durandal11707, same here
[17:06] <saste> makes much more sense to push as separate commits
[17:07] <durandal11707> i loosing valuable time rebasing nits over additional work i already done
[17:07] <saste> Daemon404, is ffprobe patch related to #1386
[17:09] <durandal11707> it is related to chat, look at irc logs ....
[17:12] <durandal11707> also you cant get format names for stereo3d in/out without duplicating bytes for no valuable gain
[17:14] <saste> durandal11707, i'm not sure what you're talking about, but my remark about stereo3d long names is mostly related to backward compatibility concerns
[17:15] <durandal11707> it is not loooooong names but for log message
[17:15] <durandal11707> also feel free to add looooong names, i consider that waste of time
[17:19] <saste> durandal11707, i don't mind
[17:20] <saste> if they were only used in log, that's fine as is
[17:20] <durandal11707> salt&pepper noise?
[17:20] <ubitux> high contrasted dot i guess
[17:21] <ubitux> (± larges)
[17:22] <ubitux> saste: maybe you can just do it with a geq somehow btw :))
[17:22] <durandal11707> why? where that could be useful?
[17:22] <ubitux> to test median denoise filtering for instance
[17:22] <durandal11707> you can do bunch of stuff with geq but it is slow and unreadable
[17:22] <ubitux> or to make a degraded effect (see the vintage thing)
[17:24] <durandal11707> i guess one could add another flag to noise filter
[17:27] <ubitux> it would be nice to have something like proba=3/(w*h):minsize=2*2:maxsize=5*5
[17:28] <ubitux> -> you get on average 3 dots per frame, of a size between 4 and 25 (random shaped)
[17:37] <Daemon404> saste, ffprobe patch is relates to a bug i ezperienced
[17:38] <Daemon404> dunno if theres an open bug
[17:38] <Daemon404> related*
[17:39] <ubitux> it's the same bug you reported
[17:39] <ubitux> a vp6 res issue with ffprobe
[17:41] <Daemon404> ok
[17:42] <Daemon404> im sshing on my phone roght now
[17:42] <Daemon404> so looking it up was effort
[17:48] <saste> Daemon404, i wonder if it should be fixed in the VP6 decoder
[17:48] <ubitux> can it be?
[17:48] <Daemon404> no
[17:48] <Daemon404> not afaik
[17:57] <saste> Daemon404, did you know why vp6 is resetting the w/h values?
[17:58] <Daemon404> BBB can likely explain
[17:58] <Daemon404> or i can once i get homw
[17:58] <Daemon404> on a bus atm
[17:58] <Daemon404> touch keyboard ftl
[17:59] <saste> Daemon404, your patch seems acceptable, but i want to understand if there is a simpler fix (possibly at the lib level)
[18:00] Action: saste runs gdb
[18:06] <saste> in avcodec_open2() we call avcodec_set_dimensions(), which resets h/w to the coded w/h
[18:06] <saste> /We only call avcodec_set_dimensions() for non h264 codecs so as not to overwrite previously setup dimensions
[18:07] <saste> michaelni, do you know why?
[18:12] <ubitux> saste: the for working on av_expr_is_const ;)
[18:13] <ubitux> thx*
[18:17] <ubitux> BBB-work, BBB: re-ping for the iconv patch test from Carl
[18:17] <Daemon404> i have a mac too
[18:17] <Daemon404> its just running windows atm
[18:17] <Daemon404> maybe ill boot it one day.
[18:18] <ubitux> if at some point you do it, please test http://ffmpeg.org/pipermail/ffmpeg-devel/2013-March/139944.html with --enable-iconv
[18:18] <ubitux> of course, assuming it's currently broken with your setup
[18:19] <Daemon404> well no
[18:19] <Daemon404> i dont have macports
[18:19] <ubitux> ok :)
[18:19] <Daemon404> all pots systes are SHIT on mac
[18:19] <Daemon404> ports*
[18:20] <ubitux> always the same story; it's just a shitty workaround for an initial bad design 
[18:21] <ubitux>  else if (strmatch(next, "lte"   )) { AVExpr *tmp = d->param[1]; d->param[1] = d->param[0]; d->param[0] = tmp; d->type = e_gte; }
[18:22] <ubitux> lte -> type = e_gte? oO
[18:31] <BBB-work> ubitux, pong
[18:31] <BBB-work> which patch?
[18:31] <ubitux> http://ffmpeg.org/pipermail/ffmpeg-devel/2013-March/139944.html
[18:31] <ubitux> this one
[18:31] <ubitux> with --enable-iconv
[18:32] <BBB-work> it looks correct to me
[18:32] <BBB-work> but don't you want to test for libiconv also?
[18:33] <BBB-work> otherwise you get this funny situation where I have libiconv but it's not detected
[18:33] <ubitux> this patch is meant to test properly the availability first
[18:34] <ubitux> basically doesn't false-positively detect iconv
[18:34] <ubitux> supporting mac os libiconv ports thing ’ i really don't care personnally
[18:34] <ubitux> as long as it doesn't break build (which is currently the case if you --enable-iconv)
[18:35] <cbsrobot_> Daemon404: homebrew is quite ok
[18:35] <BBB-work> I can test at home
[18:35] <BBB-work> probably over weekend
[18:35] <BBB-work> is that ok?
[18:35] <ubitux> sure, no hurry
[18:36] <ubitux> it's Carl's patch, and the current workaround is good-enough for me
[18:36] <cbsrobot_> Daemon404: but of course not as good as pkg_add
[18:45] <ubitux> durandal11707, saste: 'got some time for reviews, what do you want me to look at?
[18:47] <saste> ubitux, opencl patch?
[18:48] <ubitux> from highgod?
[18:48] <saste> wei gao is the name of the guy iirc
[18:48] <ubitux> yeah right that's him
[18:49] <saste> ubitux, also maybe you want to remove some sscanf from lavfi
[18:49] <saste> deshake is one of the few missing
[18:51] <ubitux> that's not review!
[18:51] <ubitux> that's monkey patching!
[18:51] <ubitux> but sure ok i can do this
[18:52] <wm4> is AVFormatContext.duration updated during playback? I'm wondering whether reducing max_analyze_duration to something small will break stuff, or whether it only matters for the ffmpeg binary
[18:53] <ubitux> saste: so you're not working on deshake, i can do it?
[18:55] <saste> ubitux, sure
[18:55] <durandal11707> wm4: it may break many stuff in many strange ways
[18:55] <wm4> like what
[18:55] <wm4> I'm fine with the playback time being updated during playback
[18:56] <wm4> I absolutely hate that libavformat is reading so much data on start
[18:56] <wm4> it makes starting streams etc. much slower
[18:56] <durandal11707> max analyze duration have nothing to do with duration of stream
[18:57] <durandal11707> it is used to find streams for brain dead formats
[18:57] <wm4> I thought it was used to estimate the bitrate
[18:58] <durandal11707> nope..... it just mean : -analyzeduration   <int>        .D.... specify how many microseconds are analyzed to probe the input
[19:05] <Compn> durandal11707 : http://samples.ffmpeg.org/V-codecs/ficvideo.fic
[19:05] <Compn> durandal11707 : are you working on the codec/format ?
[19:09] <durandal11707> Compn: not any serious work yet, i only had quick look on disassembly
[19:12] <saste> michaelni: ping on lavu/opt: make av_opt_get() return NULL in case a string is not set
[19:12] <saste> same for the other people around
[19:14] <Compn> durandal11707 : did you see the directshow demuxer for ficvideo ?
[19:14] <Compn> i mean .fic
[19:14] <Compn> i saw it in my searches
[19:15] <Compn> http://www.mirillis.com/forum/viewtopic.php?f=36&t=12625
[19:15] <durandal11707> nope, but it would be funny to know why they created yet another container
[19:16] <Compn> obviously they only wanted you to use their software for editing
[19:16] <Compn> what better way than to use special container
[19:20] <durandal11707> fic format seems obvious, it have FICM for header, FICV chunk for video and FICA for audio
[19:21] <durandal11707> Compn: can lates player play those old fic files?
[19:23] <Compn> i didnt test lol
[19:23] <Compn> if you want i can
[19:23] <ubitux> durandal11707: rgb48 is rgb 10 bits instead of 8, right?
[19:23] <nevcairiel> 16 actually
[19:24] <ubitux> ah? ok
[19:24] <nevcairiel> rgb48 is without alpha, 3x16, and 64 is with alpha, 4x16
[19:24] <ubitux> ok
[19:25] <ubitux> a random git grep gave me "-pix_fmt rgb48le -bits_per_raw_sample 10" so i was wondering
[19:25] <saste> ubitux, alpha curves would be intersting as well
[19:25] <ubitux> saste: i wonder in what case it really makes sense
[19:26] <ubitux> but yeah we could add various channels
[19:26] <ubitux> like a "value" one too (just like gimp)
[19:26] <durandal11707> ubitux: that is some ugly hack for dpx/openjpeg....
[19:26] <ubitux> durandal11707: ok :)
[19:51] <cone-714> ffmpeg.git 03Paul B Mahol 07master:3f35f36a2e7e: lavfi: port MP stereo3d filter
[19:51] <cone-714> ffmpeg.git 03Paul B Mahol 07master:1c808943b8ec: lavfi/mp: drop mp=stereo3d filter
[19:53] <saste> \o/
[19:54] <durandal_1707>  |
[19:54] <ubitux> yay.
[19:55] <michaelni> saste, aout null opt, i wasnt sure about ABI/API as its public function
[19:55] <durandal_1707> it have 2 eastern eggs hidden, find it fast
[19:55] <michaelni> saste, but also wasnt really sure it was a problem 
[20:01] <saste> michaelni, same here
[20:16] <cone-714> ffmpeg.git 03Stefano Sabatini 07master:f164228fd793: lavfi/overlay: add support for partial overlaying
[20:16] <cone-714> ffmpeg.git 03Stefano Sabatini 07master:8d9509377e7e: lavfi/overlay: improve log message warning in case of clipped overlay
[20:22] <ubitux> nice.
[20:26] <ubitux> saste: do you want me to look at the tinterlace perf issue?
[20:26] <ubitux> so we can get rid of the mp version
[20:36] <saste> ubitux, most tinterlace modes just perform stride arithmetic
[20:36] <saste> since tinterlace is more powerful than mp=tinterlace, it's a bit slower
[20:36] <saste> for example it calls a function, while mp=tinterlace use inline code to perform the same thing (with some code duplication)
[20:37] <saste> so if you compare routine by routine you get slower results
[20:38] <saste> what i didn't do: compare the overall performance of a command using mp=tinterlace and tinterlace
[20:38] <ubitux> ok
[20:38] <cone-714> ffmpeg.git 03Clément BSsch 07master:72d2e9437d7e: doc/showspectrum: add some examples.
[20:38] <saste> and see if there is significant difference
[20:38] <saste> i think a few optims can be done on tinterlace, but the cost is code duplication or some uglification
[20:39] <cbsrobot_> ubitux: the truepeak looks great to me, maybe just output it in dBTP and in the summary aswell
[20:39] <ubitux> saste: some modes in particular?
[20:39] <ubitux> cbsrobot_: ok, i keep that in mind
[20:44] <ubitux> saste: i'm adding some random "@subsection Examples" all over the doc/filters.texi ; you don't have a local patch for this already, right?
[20:52] <saste> ubitux, no
[21:08] <durandal_1707> saste: code dublicatio using inline is not code duplication
[21:08] <cone-714> ffmpeg.git 03Clément BSsch 07master:c0bfc2b90d3d: doc/filters: make examples a bit more consistent.
[21:11] <ubitux> sorry if the above commit broke some local patches
[21:19] <cone-714> ffmpeg.git 03Clément BSsch 07master:34d0ea524776: lavfi/deshake: support named options.
[21:41] <saste> michaelni, and ping about the avcodec_set_dimensions()
[21:41] <saste> if I should apply Daemon404 patch or if there is a better fix
[21:42] <Daemon404> saste, theres a small error in that patch
[21:42] <Daemon404> i accidentally removed a random comment
[21:42] <Daemon404> just fyi
[21:42] <Daemon404> (also i have push access)
[22:36] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:4c56b79a16b2: vf_stereo3d: Add AVOption table terminator
[22:36] <cone-714> ffmpeg.git 03Michael Niedermayer 07master:d92936481470: asfdec: dont truncate type 2-5 values
[22:50] <durandal_1707> code.google.com sucks, i cant get shallow copy of repo
[23:28] Action: ubitux wants to write another stupid filter
[23:28] <ubitux> what about a "vignette" filter?
[23:29] <durandal_1707> it does?
[23:31] <ubitux> add some darkness in the corners
[23:32] Action: ubitux just ffbombed its desktop with ffmpeg -i in1080p.mov -vf mp=tinterlace -f null -
[23:32] <ubitux> his*
[23:34] <ubitux> michaelni: i agree with saste that's it's safe to drop tinterlace...
[23:34] <ubitux> i mean mp=tinterlace
[23:37] Action: ubitux wonders when the oom killer will wake up
[23:39] <michaelni> ubitux, _remove_ some darkness from the corners may be more usefull
[23:39] <ubitux> to undo a vignette filter?
[23:39] <ubitux> i guess i can have a -1 to 1 setting
[23:41] <ubitux> michaelni: but you know, i'm not very fond of useful stuff
[23:41] <michaelni> the idea would be to correct vignetting of lenses
[23:41] <ubitux> (i like when i'm alone working on stupid stuff, so no one bother me with it and i'm the king of it)
[23:41] <michaelni> would need more than 1 parameter though probably
[23:41] <ubitux> lens correction is kinda advanced
[23:43] <michaelni> ubitux, what is advanced on that ?
[23:43] <ubitux> this particular task might not be
[23:43] <ubitux> but lens correction generally requires more advanced stuff
[23:43] <ubitux> image distorsion etc
[23:43] <michaelni> user provides a reference image from white wall and filter uses that to find parameters for vignetting correction and ...
[23:44] <michaelni> ubitux, fisheye <-> rectilinear converter filter
[23:45] <ubitux> right
[23:45] Action: ubitux just want to add some darkness in the corners
[23:45] <ubitux> :(
[23:55] <tab> ubitux: your configure patch for iconv works4me
[23:55] <ubitux> the patch from carl?
[23:56] <tab> http://ffmpeg.org/pipermail/ffmpeg-devel/2013-March/139944.html
[23:56] <ubitux> right, so when you ./configure --enable-iconv
[23:56] <ubitux> it doesn't enable it?
[23:56] <ubitux> (well, basically it fails to detect it)
[23:57] <tab> it does in both cases
[23:57] <tab> without patch, everything is fine up until the final link for ffmpeg/etc
[23:57] <tab> although i don't understand why that link cmd is causing trouble
[23:58] <ubitux> just to make sure, after the patch the configure --enable-iconv tells you it's not found, right?
[23:59] <tab> wait wasn't i supposed to have it installed
[23:59] <tab> it is found
[00:00] --- Wed Mar  6 2013