[Ffmpeg-devel-irc] ffmpeg-devel.log.20121030

burek burek021 at gmail.com
Wed Oct 31 02:05:02 CET 2012


[00:04] <durandal_1707> merging considered evil
[00:11] <durandal_1707> who reviewes merges?
[00:13] <kierank> nobody
[01:19] <durandal_1707> Daemon404: you have bactrace for such crash? I'm curious why ubitux cant reproduce it, is it reproducible using ubitux commmand?
[02:44] <cone-186> ffmpeg.git 03Michael Niedermayer 07740959fdbfbf: mpeg12parser: cleanup bitrate parsing.
[02:44] <cone-186> ffmpeg.git 03Heesuk Jung 07c349177b8cc0: mpeg12video_parser: Wrong bit rate in MPEG2 video file (Ticket #1862)
[02:44] <cone-186> ffmpeg.git 03Michael Niedermayer 074b7f34a34e45: mpeg12videodec: dont pass mpeg1 VBR special case value on as bitrate.
[04:35] <cone-186> ffmpeg.git 03Paul B Mahol 075f614bcf2c5d: vqavideo: return more meaningful error codes
[09:21] <ubitux> michaelni: i'd like to use sws (or something else) to do some various convolution filtering experiments with different kernels
[09:21] <ubitux> is it appropriate to use that?
[09:22] <ubitux> last time you pointed me out some particular functions for gaussian
[09:22] <ubitux> but i'd like to do similar things with "users" kernels
[09:23] <ubitux> i can at least re-use the SwsVector and related, but not sure about the rest
[09:30] <ubitux> mmh i should look at smartblur.
[11:21] <ubitux> http://b.pkh.me/sharpen.png :))
[11:25] <funman> re: ticket 1869: who is richardpi ?
[11:25] <funman> or richardpl
[11:26] <cbsrobot> funman: paul b mahol
[11:26] <cbsrobot> wait
[11:26] <funman> v210 is such a weird packing format that indeed swscale support would be hard
[11:26] <cbsrobot> on irc: durandal_1707
[11:26] <funman> ok
[11:30] <av500> ubitux: I have a better filter: http://i.imgur.com/JHGZL.png
[11:32] <cbsrobot> looks like the new hotten filter& lol
[11:32] <av500> will it be merged soon?
[11:52] <cone-18> ffmpeg.git 03Carl Eugen Hoyos 078bf44d5c9681: Partly revert f174fba: indeo3 decoder does not support direct rendering.
[12:10] <cone-18> ffmpeg.git 03Carl Eugen Hoyos 07d3de3a16d1e4: Partly revert f174fba: yop decoder does not support direct rendering.
[12:12] <ubitux> haha av500 
[13:22] <cone-18> ffmpeg.git 03Michael Niedermayer 0760e957476e33: rawdec: check avpicture_get_size() return value
[13:51] <ubitux> michaelni: i have some trouble with simple filters and swscale
[13:51] <ubitux> it seems to /0 in the filter init in some with some particular filters
[13:54] <ubitux> maybe i'm doing something wrong&
[13:54] <ubitux> but i don't see how i'm suppose to change my input
[13:56] <cone-18> ffmpeg.git 03Luca Barbato 0707585ffa62ee: nutenc: verbosely report unsupported negative pts
[13:56] <cone-18> ffmpeg.git 03Luca Barbato 07d4bff9f1ab59: nut: support textual data
[13:56] <cone-18> ffmpeg.git 03Martin Storsjö 07f21d5c905dd5: rtsp: Avoid a cast when calling strtol
[13:56] <cone-18> ffmpeg.git 03Diego Biurrun 0772c758f1fd06: aacenc: Drop some unused function arguments
[13:56] <cone-18> ffmpeg.git 03Diego Biurrun 07ab799664755c: cmdutils: Conditionally compile libswscale-related bits
[13:56] <cone-18> ffmpeg.git 03Diego Biurrun 0770478746a94e: avconv: Drop unused function argument from do_video_stats()
[13:56] <cone-18> ffmpeg.git 03Diego Biurrun 07e519990ced06: fate: Add shorthands for acodec PCM and ADPCM tests
[13:56] <cone-18> ffmpeg.git 03Diego Biurrun 07381dc1a5ec09: fate: ac3: Place E-AC-3 tests and AC-3 tests in different groups
[13:56] <cone-18> ffmpeg.git 03Michael Niedermayer 07cd37963684d8: Merge commit '381dc1a5ec0925b281c573457c413ae643567086'
[13:59] <ubitux> nyuhu: it seems you forgot to call ff_draw_slice() in the end_frame() of smartblur
[13:59] <durandal_1707> michaelni: it sucks that nut demuxer cant display stream duration
[13:59] <durandal_1707> michaelni: is that format limitation?
[14:00] <ubitux> nyuhu: http://pastie.org/5137344 something like this seems to fix problems such as ./ffplay tests/lena.pnm -vf 'split[a][b]; [a]pad=iw*2[src]; [b]smartblur=1:1:1[filt]; [src][filt]overlay=w'
[14:01] <ubitux> i'll likely push this tonigh unless you want to fix it differently
[14:08] <durandal_1707> michaelni: about wav demuxer problem with some codecs, see this bug report: http://ffmpeg.org/trac/ffmpeg/ticket/1658
[14:10] <nyuhu> ubitux : thats fine by me :)
[14:10] <ubitux> okay
[14:11] <nyuhu> (btw, why is this call needed ? :x)
[14:16] <ubitux> because if you don't it doesn't work of course ;)
[14:16] <ubitux> i dunno the details honestly
[14:16] <ubitux> i just compared to other filters
[14:17] <ubitux> since draw_slice isn't set in your filter, i guess you have to call it explicitely in the next filter at some point
[14:25] <nyuhu> ok ok
[15:16] <cone-18> ffmpeg.git 03Diego Biurrun 072cbdd7c92958: fate: ac3: Add dependencies
[15:16] <cone-18> ffmpeg.git 03Anton Khirnov 07bff5e5f8b3c5: indeo3: remove duplicate capabilities line.
[15:16] <cone-18> ffmpeg.git 03Anton Khirnov 0701b760190d32: lavr: add general API usage doxy
[15:16] <cone-18> ffmpeg.git 03Anton Khirnov 07dc5793062e7a: lavr: document upper bound on number of output samples.
[15:16] <cone-18> ffmpeg.git 03Justin Ruggles 0739ef66f5300c: lpc: Add a function for calculating reflection coefficients from autocorrelation coefficients
[15:16] <cone-18> ffmpeg.git 03Martin Storsjö 078b25a20efbf4: lpc: Add a function for calculating reflection coefficients from samples
[15:16] <cone-18> ffmpeg.git 03Martin Storsjö 079b500b8f6c98: avcodec: Add a RFC 3389 comfort noise codec
[15:16] <cone-18> ffmpeg.git 03Anton Khirnov 0720dd41af8513: lavfi: add ashowinfo filter
[15:16] <cone-18> ffmpeg.git 03Anton Khirnov 0714f031d7ecfa: dv: use AVStream.index instead of abusing AVStream.id
[15:16] <cone-18> ffmpeg.git 03Michael Niedermayer 07e79c3858b35f: Merge commit '14f031d7ecfabba0ef02776d4516aa3dcb7c40d8'
[15:31] <durandal_1707> code stealers
[15:32] <cone-18> ffmpeg.git 03Anton Khirnov 07ede4cedf47f4: dv: fix indentation
[15:32] <cone-18> ffmpeg.git 03Ronald S. Bultje 07c285edd06ea6: Remove usage of INIT_AVX in h264_intrapred_10bit.asm.
[15:32] <cone-18> ffmpeg.git 03Ronald S. Bultje 0708b028c18dc3: Remove INIT_AVX from x86inc.asm.
[15:32] <cone-18> ffmpeg.git 03Michael Niedermayer 076f5f7f705d10: Merge commit '08b028c18dc31b6de741861b9555669dcca4d12a'
[15:48] <durandal_1707> michaelni: come avcodec entries that used MKBETAG dissapeared - i guess without mayor bump
[15:48] <durandal_1707> *some
[15:49] <michaelni> durandal_1707, what did when, where disappear exactly ?
[15:50] <durandal_1707> michaelni: for example MKBETAG for CLLC
[16:29] <cone-18> ffmpeg.git 03Paul B Mahol 078a4284adbf18: ff: fix case where image does not have alpha channel
[16:30] <durandal_1707> shit, that happens when you need to resurrect old patch
[16:57] <durandal_1707> what is that?
[17:22] <cone-18> ffmpeg.git 03Paul B Mahol 07788afee86e28: lavfi: remove palette libmpcodecs wrapper
[17:25] <Compn> zygo audio
[17:25] <Compn> sounds like quicktime codec
[17:30] <durandal_1707> btw, gcc47 and lcov failed badly
[17:31] <ubitux> any reason we keep libmpcodecs/vf_hqdn3d.c?
[17:31] <durandal_1707> inline assembly
[17:32] <Compn> you'll have to ask filter man
[17:32] <ubitux> durandal_1707: there is a x86/hqdn3d.asm
[17:32] <durandal_1707> actuall i see no assemly there
[17:32] <durandal_1707> so just remove
[17:33] <durandal_1707> anyone porting any other mp filter?
[17:33] <ubitux> saste is waiting for a review on vf field iirc
[17:33] <ubitux> nyuhu is working on eq filter iirc
[17:34] <durandal_1707> noise looks useful for porting (gonna test it if for fun)
[17:34] <ubitux> and most of the remaining filters are telecine related
[17:34] <ubitux> assembly in this one hehe
[17:35] <durandal_1707> that kind of assembly is pointless
[17:35] <ubitux> why?
[17:35] <Compn> surprised you guys arent porting avisynth filters like some people want :P
[17:35] <Compn> ehe
[17:36] <durandal_1707> what avisynth filters?
[17:36] <ubitux> Compn: i'm looking for adding some kind of generic convolution filter which should cover quite a large range of filtering
[17:37] <ubitux> about the filter from avisynth, i had a list
[17:37] <ubitux> and basically, there were two kind of filters
[17:37] <durandal_1707> ubitux: ever heard hdmv_pgs_subtitle
[17:37] <ubitux> 1) telecine/interlacing filters
[17:37] <ubitux> 2) fft video image enhancement
[17:37] <ubitux> durandal_1707: i saw that name
[17:37] <ubitux> but that's all
[17:38] <ubitux> i'm not sure about the benefits of fft for image filtering
[17:38] <durandal_1707> ubitux: there is sample
[17:39] <ubitux> ?
[17:39] <durandal_1707> uhh ffplay works
[17:39] <durandal_1707> i mean subtitle is displayed
[17:42] <ubitux> wtf is down3dright filter
[17:42] <ubitux> it can be done with various filters in lavfi
[17:43] <ubitux> i wonder if we couldn't drop it as well
[17:43] <durandal_1707> stereo movie?
[17:45] <ubitux> dunno, but with a split/crop/scale/overlay it's do-able
[17:46] Action: ubitux wonders about the usefulness of dsize too, given the expression system we have in vf scale
[17:47] <ubitux> not sure if fil is useful too
[17:47] <ubitux> ah vf geq, i really should work on this one one day
[17:48] <ubitux> huhu vf ow, wavelet denoiser
[17:49] <durandal_1707> does any of mentioned filters actually work?
[17:49] <ubitux> down3dright seems to
[17:51] <ubitux> is rectangle useful?
[17:52] <durandal_1707> tried it. is it same as drawbox?
[17:52] <ubitux> rectangle seems to have an automatic color adjustment
[17:52] <cone-18> ffmpeg.git 03Ronald S. Bultje 07bad8e33dc92a: x86: use PRED4x4/8x8/8x8L/16x16 macros to declare intrapred prototypes.
[17:52] <cone-18> ffmpeg.git 03Ronald S. Bultje 0795c89da36ebe: Use ptrdiff_t instead of int for intra pred "stride" function parameter.
[17:52] <cone-18> ffmpeg.git 03Anton Khirnov 0702e636425970: FATE: fix (AD)PCM test dependencies broken in e519990
[17:53] <cone-18> ffmpeg.git 03Mans Rullgard 07f454e879238c: configure: rewrite print_config() function using awk
[17:53] <cone-18> ffmpeg.git 03Luca Barbato 0726b5ad254330: swscale: support gray to 9bit and 10bit formats
[17:53] <ubitux> i'm not sure drawbox supports this
[17:53] <cone-18> ffmpeg.git 03Luca Barbato 077658295ba353: pixfmt: support more yuva formats
[17:53] <ubitux> maybe an option could be added
[17:53] <cone-18> ffmpeg.git 03Michael Niedermayer 07a201639a0128: Merge remote-tracking branch 'qatar/master'
[17:53] <ubitux> tile filter might need some more options to be able to replace completely mp=tile too
[17:56] <ubitux> about yvu9, shouldn't it be part of swscale?
[17:57] <ubitux> same for yuvcsp :p
[17:57] <nevcairiel> it should, but are you going to touch swscale to add it?
[17:58] <durandal_1707> swscale does not have them
[17:58] <durandal_1707> ?
[17:58] <nevcairiel> swscale fails at converting between different yuv matrices
[17:59] <nevcairiel> what does yvu9 do anyway?
[18:00] <nevcairiel> yvu9 can apparently be replaced by swscale actually
[18:00] <nevcairiel> yuvcsp not
[18:00] <nevcairiel> or am i thinking about the wrong plugin again
[18:00] Action: nevcairiel checks
[18:00] <durandal_1707> < nevcairiel> swscale fails at converting between different yuv matrices >>> example?
[18:01] <nevcairiel> i was thinking of the wrong plugin again, yuvcsp seems also useless
[18:01] <nevcairiel> what swscale fails at is what the vf_colormatrix plugin does
[18:01] <nevcairiel> ie. convert Rec.709 to Rec.601
[18:02] <durandal_1707> ^ what is going on with alsa?
[18:09] <durandal_1707> i already have patch that removes yuvcsp and yvu9
[18:19] <durandal_1707> what is actual purpose of noise filter?
[18:24] <durandal_1707> <@ubitux> rectangle seems to have an automatic color adjustment >>>> where?
[18:24] <TimNich> nevcairiel:  What does seem to fail is RGB<->yuv with a specified colorspace (-colorspace)
[18:25] <nevcairiel> both really
[18:25] <nevcairiel> i think its not supposed to do yuv->yuv with a different matrix
[18:25] <TimNich> and color_range
[18:25] <nevcairiel> but it falls still in its domain
[18:26] <TimNich> I wouldn't  necessarily expect that but going from RGB to 709 means going rgb->601 (the default) then using vf_colormatrix, which brings in concatenation errors..
[18:27] <TimNich> also I think colormatrix only works on 8 bit
[18:28] <nevcairiel> that could be fixed pretty easily, i guess
[18:30] <TimNich> be nice if it was. I kept meaning to look at it, but other things needed up more important ;(
[18:30] <TimNich> s/needed/ended/
[18:33] <ubitux> durandal_1707: dunno, by default
[18:33] <ubitux> at least the color seems to change automatically
[18:34] <durandal_1707> ubitux: first color can not be set at all
[18:34] <ubitux> durandal_1707: try ./ffplay -f lavfi -i testsrc,mp=rectangle=200:200:10:10
[18:34] <ubitux> and look at the border colors
[18:35] <Compn> does ffmpeg have a speex decoder or just using libspeex ?
[18:35] <ubitux> <@durandal_1707> what is actual purpose of noise filter? // generate high bandwidth source?
[18:36] <ubitux> reproduce the noise effect of unavailable videos on youtube?
[18:36] <durandal_1707> ubitux: i think this is caused by colorspace conversion
[18:36] <ubitux> durandal_1707: ah? really?
[18:37] <durandal_1707> should try with thicker border
[18:38] <ubitux> it seems rectangle has a command thing too
[18:40] <ubitux> durandal_1707: so what's your current pending drop list?
[18:40] <ubitux> yuvcsp and yvu9 only?
[18:41] <durandal_1707> yea
[18:42] <durandal_1707> hmm, why drawbox does not allow to see how thick line would be....
[18:42] <durandal_1707> *set
[18:44] <ubitux> you don't want to add down3dright and hqdn3d? :)
[18:44] <durandal_1707> on remove list?
[18:44] <ubitux> yes
[18:44] <saste> yuvcsp and yvu9
[18:45] <durandal_1707> maybe, but first goes useless stuff
[18:45] <saste> what are they useful for?
[18:45] <ubitux> can we drop vf tile as well?
[18:45] <durandal_1707> everbody ask, nobody answer
[18:45] <ubitux> (mp=tile i mean)
[18:45] <ubitux> saste: i'm going to give you the backlog in a moment
[18:46] <saste> ubitux: how is mp=tile different from tile?
[18:46] <saste> i never checked
[18:46] <durandal_1707> also rectangle have some mystery feature
[18:46] <saste> then we have mp=rectangle and drawbox, which are somehow redundant but not equivalent feature-like-wise
[18:46] <ubitux> saste: http://b.pkh.me/filter-remove
[18:47] <ubitux> saste: mp=tile seems to be quite the same
[18:47] <ubitux> i thought there was some padding option at some point
[18:47] <ubitux> but might be a bug in my memory
[18:49] <ubitux> tile might be duplicating the same frame though
[18:50] <ubitux> but that's possible with some pad & overlay
[18:50] <saste> ubitux: thx for the backlog
[18:50] <saste> i don't know if nyuhu is still working on any filter
[18:51] <ubitux> hum vf tile is crashing in mplayer hehe
[18:51] <saste> i'd wait some time before duplicating work, in all cases
[18:51] <ubitux> fortunately it works with ffplay
[18:52] <saste> anyone wants to review field? it was the simpler to port, so I did it (took a few hours, though)
[18:52] <durandal_1707> what is purpose of it?
[18:52] <ubitux> mmh actually something is wrong, vf mp=tile in ffmpeg doesn't act as vf tile in mplayer
[18:52] <ubitux> but vf tile in mplayer works just like vf tile in ffmpeg
[18:54] <saste> durandal_1707, just select the bottom or top field
[18:54] <saste> so it turns interlaced material to non-interlaced
[18:54] <saste> i don't know if it is used in practice, i suppose can be useful for testing reasons at least
[18:54] <saste> or as a very cheap deinterlacer
[18:55] <cone-18> ffmpeg.git 03Michael Niedermayer 0765e0a3ccd823: sws: fix BE/LE handling for fillPlane16()
[18:55] <cone-18> ffmpeg.git 03Tomas Härdin 07a44c3ae3b07b: Change my GPG fingerprint
[18:55] <cone-18> ffmpeg.git 03Michael Niedermayer 0789d2f014bc57: sws_unscaled: drop fill_plane9or10()
[18:55] <cone-18> ffmpeg.git 03Michael Niedermayer 0770c25c62051f: sws: move fillPlane16() to header so it can be used from more than 1 place
[18:55] <cone-18> ffmpeg.git 03Michael Niedermayer 074d2f1d8c930a: swscale.c:remove fill_plane9or10() use fillPlane16()
[18:56] <durandal_1707> saste: it here makes video height 2x times smaler
[18:57] <saste> durandal_1707, yes, that's what it's supposed to do
[19:06] <durandal_1707> saste: doxgen comment @file looks wrong - should that be on next line?
[19:08] <durandal_1707> beside that i see nothing obviously wrong
[19:27] <durandal_1707> lavc/imgconvert stuff should be exposed in lavu
[19:27] <durandal_1707> the color_type stuff for pix_fmt
[19:30] <durandal_1707> michaelni: did you do anything to make extending libswsale with packet yuv formats trivial?
[19:30] <cone-18> ffmpeg.git 03Mans Rullgard 07674c4a11cbcc: configure: fix print_config() with broke awks
[19:30] <cone-18> ffmpeg.git 03Michael Niedermayer 075cac0656a3cb: af_ashowinfo: integrate config_input() in filter_samples.
[19:31] <cone-18> ffmpeg.git 03Clément BSsch 0703179c2426da: lavfi/smartblur: add missing draw slice call.
[19:32] <michaelni> durandal_1707, i wanted to improve alot in sws but its still on my todo list ...
[19:54] <saste> people inventing random terms, and then explaining them by example (look, you just acted like "groidism")
[19:55] <ubitux> groink~
[20:00] <saste> let's call it "goinkism"
[20:00] <saste> *groinkism
[20:03] <Compn> bikeshism
[20:03] <durandal_1707> saste: so compatibility with old syntax for drawbox is not important?
[20:03] <Compn> btw i have not seen example in history of mplayer of anyone ever using vf rectangle in mplayer ...
[20:04] <saste> durandal_1707, new syntax is backward compatible
[20:04] <durandal_1707> saste: wow
[20:13] <cone-18> ffmpeg.git 03Paul B Mahol 07b90d7840dff8: lavc: const correctness for avpicture_fill()
[20:26] <ubitux> saste: oh so there is indeed a color invert for the borders in mp=rectangle, right?
[20:26] <saste> ubitux: kind of, 0xff-luma
[20:27] <ubitux> mmh right ok :)
[20:28] <saste> how can i scale preventing blurring (e.g. for scaling bitmaps?)
[20:29] <saste> ubitux: you used that when working on like, IIRC
[20:29] <saste> *on life
[20:29] <ubitux> flags=16
[20:30] <ubitux> i don't remember what it means
[20:30] <saste> which means?
[20:30] <ubitux> bitexact maybe
[20:30] <durandal_1707> saste: gonna make drawbox thickness custom?
[20:31] <ubitux> "point"
[20:31] <saste> durandal_1707, yes
[20:31] <ubitux> saste: "neighbor"
[20:31] <ubitux> flags=neighbor
[20:32] <cone-18> ffmpeg.git 03Michael Niedermayer 0728bcca46ed66: av_get_audio_frame_duration: fix IMC
[20:35] <durandal_1707> michaelni: lol it was trivial
[20:37] <saste> uhm should i make c an alias for color?
[20:37] <saste> ^^ in drawbox
[20:37] <saste> since then i have t and thickness
[20:41] <ubitux> i think we use "c" in several places as alias
[20:43] <saste> changed locally
[20:56] <ubitux> saste: hey btw, color_str[] isn't safe in drawbox
[20:56] <saste> why?
[20:56] <ubitux> no length limit
[20:57] <ubitux> overflow possible with > 1023 char
[20:57] <saste> it's allocated by av_opt
[20:57] <ubitux> char color_str[1024]; sscanf(...%s", color_str);
[20:57] <saste> it's fixed by my posted patches
[20:57] <ubitux> what patch?
[20:58] <saste> the first one, accept named options etc
[20:59] <ubitux> oh my bad, ok
[20:59] <ubitux> i was looking at the other one
[21:00] <cone-18> ffmpeg.git 03Michael Niedermayer 072f84f13806fa: fate: remove typo, reenable AC3 tests
[21:12] <durandal_1707> hmm muxing g723_1 into wav corruputs metadata
[21:14] <durandal_1707> ffmpeg -y -i [input] -ar 8000 -acodec g723_1 -ac 1 -b:a 6.3k out.wav
[21:15] <durandal_1707> only last 3 bytes of metadata
[21:16] <durandal_1707> michaelni: ^
[21:24] <saste> ubitux: yes process_command() and (dynamic) expressions may be useful in drawbox
[21:27] <ubitux> saste: reviewing vf field at the moment, give me a few minutes
[21:29] <saste> and i don't want to change drawbox behavior (yet)
[21:37] <durandal_1707> saste: shouln't thickness overflow in case it is 0? meaning it never complete
[21:38] <ubitux> arg sorry didn't review the last one saste 
[21:38] <saste> durandal_1707, thickness = 0  to me means the thickness should be 0, so don't write it
[21:39] <durandal_1707> saste: you tried that?
[21:39] <saste> sure
[21:40] <durandal_1707> but it is pointless
[21:41] <saste> no, can be useful for some (admittedly weird) scripting scenarios
[21:41] <saste> it's easier to tweak a param than to remove/insert a filter from a filtergraph
[21:42] <saste> for example you may want to disable drawing at some point (in combination with process_command)
[21:49] <michaelni> durandal_1707, locally fixed the metadata issue
[21:50] <durandal_1707> what caused it?
[21:50] <michaelni> hardcoded offset
[21:50] <durandal_1707> hah
[21:54] <cone-18> ffmpeg.git 03Michael Niedermayer 07f7a9a9966125: wavenc: fix curruption of metadata
[22:06] <ubitux> this is a \\\\\\'string\\\\\\'\\\\
[22:06] <ubitux> made my day
[22:07] <durandal_1707> ?
[22:12] <saste> ubitux, yeah, nobody wonders why people got it wrong most of the times
[22:28] <cone-18> ffmpeg.git 03Paul B Mahol 077b378c2fb763: lavfi: remove yuvcsp and yvu9 libmpcodecs wrappers
[22:37] <durandal_1707> ffplay -f lavfi -i testsrc=n=2:s=hd720
[22:37] <durandal_1707> gives strange results
[22:44] <ubitux> saste: FYI: 
[22:44] <ubitux>     new video filter to extract a single field using stride arithmetic,
[22:44] <ubitux>     i.e. without using cpu time
[22:44] <ubitux> this was the original commit description
[22:45] <ubitux> and no thread on the ml unfortunately
[22:45] <saste> durandal_1707, why?
[22:48] <durandal_1707> saste: numbers are wrong for me
[22:49] <ubitux> wow
[22:49] <ubitux> it's megafast
[22:49] <ubitux> @_@
[22:54] <cone-18> ffmpeg.git 03Paul B Mahol 077adc60021992: av_get_audio_frame_duration: add G723_1
[22:55] <saste> durandal_1707, numbers are correct here
[22:55] <saste> how so?
[22:58] <durandal_1707> bah, numbers are just too big to parse them correctly ...
[23:14] <ubitux> michaelni: i think it will be simpler for me to just apply the filter myself, without using sws
[23:16] <ubitux> i can workaround the /0 easily, but if i want to do stuff like signed ±128 post processing operation (conditionally), i think it might be simpler to do that in the filter itself
[23:16] <ubitux> unless you think it really belongs into sws?
[23:33] <michaelni> ubitux, no problem, if you prefer to do it outside sws ...
[23:33] <burek> hmmm, what is GMP4 video codec?
[23:33] <burek> Could not find codec parameters for stream 0 (Video: none (GMP4 / 0x34504D47), 720x288): unknown codec
[23:41] <llogan> burek: an un-re'd "proprietary" mpeg4 part-2 variant from geovision ip camera?
[23:42] <burek> yes, some surveilance crap
[23:42] <cone-18> ffmpeg.git 03Stefano Sabatini 07b420b87848c4: lavfi/tinterlace: remove poll_frame() callback
[23:42] <cone-18> ffmpeg.git 03Stefano Sabatini 072ed7e6cce56e: lavfi/drawbox: extend syntax, accept named options
[23:42] <cone-18> ffmpeg.git 03Stefano Sabatini 07b6248299f628: doc/filters: itemize examples for drawbox, add an example, fix grammar
[23:42] <burek> i was betting with my friend ffmpeg will play it :D
[23:42] <burek> now i must search for crappy codecs to install as dshow or something
[23:42] <llogan> heh
[23:43] <llogan> i had a private investigator bring a video to me. it was recorded by a camera with a proprietary codec by a company that no longer exists.
[23:44] <ubitux> michaelni: it's not that i prefer to do it outside sws, it just looks like simpler to me, but maybe i'm missing a magic trick :p
[23:44] <burek> yes, I guess they hope if they encode video in their own format, to bind users to buy all their crap for playback and stuff
[23:44] <burek> which never happens anyway :)
[23:44] <burek> they just waste money on obfuscating video data
[23:45] <llogan> did you ever get things working on your rasp pi?
[23:45] <burek> oh yes :)
[23:45] <burek> i compiled ffmpeg for pi
[23:45] <burek> and it started :D
[23:46] <burek> but didnt have time to test more, since i got some job to do and it took me a whole month :(
[23:46] <llogan> sounds like good wiki guide material.
[23:46] <burek> ill return to testing stuff as soon as i get some free time :)
[23:46] <burek> will do that too :)
[23:50] <cone-18> ffmpeg.git 03Stefano Sabatini 071730dcf093d2: doc/filters: fix typo in drawbox section
[23:52] <saste> llogan, we have spam in the wiki
[23:52] <llogan> link?
[23:52] <saste> https://ffmpeg.org/trac/ffmpeg/wiki/DeltaCast?version=1
[23:52] <saste> https://ffmpeg.org/trac/ffmpeg/wiki/BlackMagic?version=1
[23:52] <saste> from the same user, he should be banned
[23:53] <saste> check also today's timeline
[23:55] <llogan> odd. looks like he made a legit bug report (1869), are you sure he's just not retarded?
[23:56] <saste> well, you can't never be sure of that :)
[23:57] <llogan> usually the spam is more obvious since it has nothing to do with video, but i don't know why a user would make such wiki pages in the first place. i'll delete them and keep an eye on that user.
[23:57] <saste> but well, let's assume he did it not on purposes, or just by chance
[00:00] --- Wed Oct 31 2012


More information about the Ffmpeg-devel-irc mailing list