[Ffmpeg-devel-irc] ffmpeg-devel.log.20121221

burek burek021 at gmail.com
Sat Dec 22 02:05:02 CET 2012


[00:09] <cone-202> ffmpeg.git 03Stefano Sabatini 07master:28b96efab509: lavf/hlsenc: fix minor grammar inconsistencies in the option help messages
[00:09] <cone-202> ffmpeg.git 03Stefano Sabatini 07master:340b7caf5457: lavf/hlsenc: provide some feedback in case of invalid basename
[00:30] <cone-202> ffmpeg.git 03Carl Eugen Hoyos 07master:9eca649a92e0: Fix compilation with --disable-everything --enable-muxer=mov.
[01:00] <cone-202> ffmpeg.git 03Michael Niedermayer 07master:230bef0a89ed: ffplay: fix handling of resolution changes with h264-mt
[01:09] <saste> should i add an entry about the manuals modularization job?
[01:28] <llogan> saste: why did you change class="container" to id="container"? the css only has a .container, and not #container, AFAICT.
[01:28] <saste> llogan: now i realize it is wrong
[01:28] <saste> i started with doing init -> css and changing css at the same time
[01:29] <llogan> wait...now i'm confused because the next patch has #container
[01:29] <saste> ffmpeg.org define the container id, not the class
[01:29] <saste> i'll rework the patches
[01:29] <saste> btw what about moving default.css to ffmpeg.org?
[01:29] <saste> this way we would avoid duplication / desynching
[01:29] <durandal_1707> michaelni: the check is completly bogus widht/height is unsigned and i see what problems this could cause
[01:30] <durandal_1707> michaelni: *dont see what problems ...
[01:30] <llogan> saste: i was wondering about that.
[01:30] <saste> llogan, the problem with this approach is that you need to wait for a day before the css is applied to the website
[01:30] <saste> which is a big drawback
[01:33] <llogan> is it a bigger drawback than duplication?
[01:36] <cone-202> ffmpeg.git 03Paul B Mahol 07master:cfa530034f25: gifdec: remove dead store
[01:39] <saste> llogan, yes, for example if someone wants to hack the website css and can't find it
[01:40] <saste> while it is a nobrainer to diff website and doc css and synch them from time to time
[01:46] <michaelni> durandal_1707, someone could change the code setting w/h and forget updating the check, but as is the casts are useless
[01:47] <durandal_1707> check should not exist because width is uint16 and spec does not make it int16
[01:59] <durandal_1707> can we move banner contest to -user list?
[02:05] <saste> durandal_1707, why?
[02:07] <durandal_1707> spam
[02:09] <iive> is it single thread?
[02:13] <llogan> durandal_1707: if you are the tiebreaker it may end sooner.
[02:18] <durandal_1707> i'm disapointed in done artwork
[03:01] <Compn> i still want a snowflake zigzag :P
[03:15] <michaelni> Compn, why did you not submit a logo then ? ;)
[03:16] <Compn> i'm not very good with art
[05:12] <cone-708> ffmpeg.git 03Michael Niedermayer 07master:63753186cc27: lavf: Remove "Adjusting PTS forward" code
[10:29] <simonec77> Hi, I'm trying to use da AAC native Encoder, butavcodec_encode_audio2();
[10:29] <simonec77> with this setup:
[10:29] <simonec77> psAVCodecContext->channels = 2;
[10:29] <simonec77> psAVCodecContext->sample_fmt = AV_SAMPLE_FMT_FLT;  
[10:29] <simonec77> psAVCodecContext->sample_rate = 48000;
[10:29] <simonec77> psAVCodecContext->bit_rate = 128000;
[10:29] <simonec77> psAVCodecContext->profile = FF_PROFILE_AAC_LOW;
[10:29] <simonec77> psAVCodecContext->strict_std_compliance = FF_COMPLIANCE_EXPERIMENTAL;
[10:29] <simonec77> codecID : CODEC_ID_AAC or AV_CODEC_ID_AAC
[10:30] <simonec77> return error...
[10:32] <simonec77> Hi, I'm trying to use da AAC native Encoder, butavcodec_encode_audio2();
[10:32] <simonec77> avcodec_encode_audio2();
[10:32] <simonec77> with this setup:
[10:32] <simonec77> psAVCodecContext->channels = 2;
[10:32] <simonec77> psAVCodecContext->sample_fmt = AV_SAMPLE_FMT_FLT;  
[10:32] <simonec77> psAVCodecContext->sample_rate = 48000;
[10:32] <simonec77> psAVCodecContext->bit_rate = 128000;
[10:32] <simonec77> psAVCodecContext->profile = FF_PROFILE_AAC_LOW;
[10:32] <simonec77> psAVCodecContext->strict_std_compliance = FF_COMPLIANCE_EXPERIMENTAL;
[10:32] <simonec77> codecID : CODEC_ID_AAC or AV_CODEC_ID_AAC
[10:32] <simonec77> return error...
[10:32] <simonec77> any idea?
[10:38] <wm4> simonec77: you posted your question twice
[10:59] <simonec77> sorry... I apoligize...
[13:01] <durandal_1707> anyone writing new filter?
[13:55] <Tjoppen> i'm feeling stupid - does anyone know how to change the font size in the xfce terminal (ubuntu default terminal)?
[13:56] <Tjoppen> nm, I didn't notice the font size thingy to the right of the font chooser button
[14:11] <cone-67> ffmpeg.git 03Jernej Virag 07master:e30e8e311e14: sapenc: Pass the title on to the chained muxers
[14:11] <cone-67> ffmpeg.git 03Björn Axelsson 07master:1eaff98c8320: flvenc: Check whether seeking back to the header succeeded
[14:11] <cone-67> ffmpeg.git 03Martin Storsjö 07master:a925f723a915: rtp: Don't read priv_data unless it is allocated
[14:11] <cone-67> ffmpeg.git 03Michael Niedermayer 07master:29707f5ba6ff: Merge commit 'a925f723a915bc0255e2673f8817af5212131763'
[15:51] <cone-67> ffmpeg.git 03Paul B Mahol 07master:ae27b70b9d9d: alsdec: channel sorting
[15:51] <cone-67> ffmpeg.git 03Paul B Mahol 07master:fb3f28ee289b: alsdec: cosmetics after previous commit
[16:44] <wm4> is it generally safe to change the image data if either FF_BUFFER_HINTS_PRESERVE is not set, or if buffer hints are not valid, if AVFrame.reference is 0?
[17:25] <cone-67> ffmpeg.git 03Ronald S. Bultje 07master:8c53d39e7f06: lavc: introduce VideoDSPContext
[17:25] <cone-67> ffmpeg.git 03Luca Barbato 07master:6906b19346ae: lavc: add missing files for arm
[17:25] <cone-67> ffmpeg.git 03Michael Niedermayer 07master:a41bf09d9c56: Merge commit '6906b19346ae8a330bfaa1c16ce535be10789723'
[17:25] <michaelni> wm4, yes
[17:37] <cone-67> ffmpeg.git 03Diego Biurrun 07master:511cf612ac97: miscellaneous typo fixes
[17:37] <cone-67> ffmpeg.git 03Michael Niedermayer 07master:d27edc038a5d: Merge commit '511cf612ac979f536fd65e14603a87ca5ad435f3'
[17:42] <cone-67> ffmpeg.git 03Diego Biurrun 07master:69f086e0f90f: build: avfilter: Remove duplicate compilation entry for vf_scale
[17:42] <cone-67> ffmpeg.git 03Michael Niedermayer 07master:be2c0adc968d: Merge commit '69f086e0f90f23d89e5739b099a4f984fa6a7885'
[17:51] <cone-67> ffmpeg.git 03Diego Biurrun 07master:0a0e340f5b8d: configure: Make avconv depend on null, anull and resample filters
[17:51] <cone-67> ffmpeg.git 03Martin Storsjö 07master:90c784cc13f6: rtpdec: Pass the sequence number to depacketizers
[17:51] <cone-67> ffmpeg.git 03Michael Niedermayer 07master:bb3420d88e71: Merge commit '90c784cc13f6bf21a8eb69f3b88b50c7a70f6c59'
[17:53] <wm4> michaelni: thanks... just makes me wonder why the buffer hints are not always set
[17:55] <cone-67> ffmpeg.git 03Martin Storsjö 07master:c7d4de3d7304: rtpdec_vp8: Don't return known-broken packets
[17:55] <cone-67> ffmpeg.git 03Michael Niedermayer 07master:1b598e677618: Merge commit 'c7d4de3d730473167e31cf765e0333c965f934d3'
[18:13] <cone-67> ffmpeg.git 03Carl Eugen Hoyos 07master:0fb79f30db13: Support H264 over video4linux2.
[18:19] <cone-67> ffmpeg.git 03Carl Eugen Hoyos 07master:a3a22c2178f8: Revert "Always write all colours for animated gif files, including transparent ones."
[18:19] <durandal_1707> ^omg
[18:29] <ubitux> why is the ffmpeg.org/ffmpeg.html page still huge?
[18:35] <saste> ubitux, it takes 24h to update
[18:35] <saste> i pushed the docu patches around midnight
[18:36] <saste> so you need to wait still a few hours
[18:37] <ubitux> ok :(
[18:40] <durandal_1707> nobody writing new filter?
[18:41] <ubitux> i am somehow writing 3 currently
[18:41] <saste> ubitux, ah i fixed overlay, will post a patch soon
[18:42] <saste> ubitux, which ones?
[18:42] <saste> so we avoid duplicating work
[18:43] <ubitux> vf pp, experiment with a fft 2d filter, and i plan to work on an ivtc one at some point
[18:43] <ubitux> i think mateo` is on vf spp
[18:43] <ubitux> i don't remember some other ppl working on particular filters
[18:43] <ubitux> saste: but i'm sure you're hidding some!
[18:46] <saste> ok, i have in my local repos: psnr, perspective, rotate, fish, elbg
[18:46] <saste> but i'm not going to work on them anytime soon
[18:46] <durandal_1707> elbg?
[18:46] <saste> also i don't know if nyuhu wants to finish his ones
[18:46] <ubitux> fish?
[18:46] <saste> or i'll update them myself
[18:47] <saste> nyuhu ^
[18:47] <saste> elbg -> posterize
[18:47] <saste> fish, well this is an ancient filter, it was a port of the vhook/fish filter
[18:47] <ubitux> source filter of an aquarium?
[18:48] <saste> no it was a fish detector
[18:48] <ubitux> haha ok
[18:48] <saste> you specify the color, and the filter selects a frame containing that color
[18:48] <ubitux> ok :)
[18:48] <saste> used to "capture" fish swimming in front of the camera
[18:48] <saste> really useful if you deal with fish
[18:49] <ubitux> :)
[18:51] <saste> and more secret filters
[19:14] <durandal_1707> saste: more filters?
[19:15] <saste> durandal_1707, not hold your breath, i'm working on several unrelated items at the moment
[19:18] <durandal_1707> saste: ok, i gonna write histogram filter
[19:19] <saste> durandal_1707, you can fix the one posted by nyuhu 
[19:19] <saste> or use it to borrow ideas, i can send you the patch
[19:19] <durandal_1707> that one does really too much
[19:19] <durandal_1707> saste: histogram is not rocket science
[19:20] <saste> i know but it may save you some time
[19:20] <durandal_1707> i doubt so
[19:22] <durandal_1707> i also would write spinning cube source if I know how to render it fast
[19:23] <wm4> vf_opengl when!?
[19:24] <durandal_1707> wm4: more like vf_yafaray or vf_blender
[19:33] <durandal_1707> i wonder how i should name file, isn't it V->V filter?
[20:11] <cone-67> ffmpeg.git 03Paul B Mahol 07master:a8ebb77493d6: lavfi/swapuv: use FFSWAP
[20:52] <durandal_1707> bunch of filters nees updater for more pix formats
[21:19] <burek> one stupid question
[21:19] <burek> what exactly --enable-postproc enables
[21:20] <nevcairiel> building libpostproc
[21:20] <burek> i mean what features does it enable
[21:20] <nevcairiel> on its own, probably none. you can use vf_pp now!
[21:21] <burek> ok thanks :)
[21:21] <burek> im trying to find a minimal configure line
[21:21] <burek> which will still be useful/functional :)
[21:21] <nevcairiel> at least i thought there was such a thing as vf_pp anyway
[21:22] <cone-67> ffmpeg.git 03Piotr Bandurski 07master:0e3a4da106ee: yop: set video bit rate
[21:22] <nevcairiel> oh its part of the mp  wrapper
[21:22] <cone-67> ffmpeg.git 03rogerdpack 07master:d828bae9d244: better failure message for img2enc
[21:22] <nevcairiel> i see
[21:27] <ubitux> < nevcairiel> on its own, probably none. you can use vf_pp now! // i haven't pushed it yet
[21:27] <ubitux> i need to figure out how to make dering behave properly
[21:27] <ubitux> because there is a mismatch between mp=pp and native pp because of this
[21:28] <ubitux> and& that's a bit tricky :p
[21:29] <iive> what kind of mismatch?
[21:31] <ubitux> http://ffmpeg.org/pipermail/ffmpeg-devel/2012-December/135613.html
[21:43] <michaelni> ... just drop dering from the tests as this doesnt affect anything but such test
[21:51] <durandal_1707> how to list formats that filter supports?
[21:53] <michaelni> libavfilter/filtfmts-test
[21:54] <ubitux> burek: it should be the default, as long as you have --enable-gpl
[21:56] <burek> ubitux, ok thanks :)
[21:58] <durandal_1707> michaelni: how you build that?
[21:59] <ubitux> make libavfilter/filtfmts-test
[21:59] <durandal_1707> found i
[22:04] <burek> something happened with latest ffmpeg, regarding libaacplus support
[22:04] <burek> [libaacplus @ 0x2b634e0] nb_samples (0) != frame_size (2048) (avcodec_encode_audio2)
[22:04] <burek> [0x2b62c58] avcodec encoder error: Encoding problem..
[22:04] <burek> this crashes latest vlc, when compiled with latest ffmpeg
[22:05] <durandal_1707> burek: can you reproduce with ffmpeg?
[22:05] <burek> let me try
[22:09] <burek> i did manage to reproduce this (it might be related)
[22:09] <burek> http://pastebin.com/SYj4QHA4
[22:10] <burek> notice that i didn't specify -ar 22050 anywhere
[22:11] <burek> same with: ffmpeg -f alsa -ar 44100 -ac 2 -i hw:0 -c:a libaacplus -ab 32 -ar 44100 -ac 2 -y out.mp4
[22:16] <ubitux> michaelni: ok, done
[22:16] <ubitux> dering dropped, now fate match both
[22:16] <ubitux> i'm dropping mp=pp
[22:17] <ubitux> gonna update the doc with the review from saste, and will apply
[22:17] <ubitux> any objection?
[22:18] <ubitux> i could send a new patchset though
[22:26] <durandal_1707> burek: that is another error and happens because you give wrong args to encoder
[22:27] <burek> durandal_1707, and what would the wrong params be in this case?
[22:29] <michaelni> ubitux, missing yuvj formats in pp
[22:29] <durandal_1707> burek: bitrate
[22:30] <michaelni> otherwise might be ok
[22:30] <burek> durandal_1707, it worked before
[22:30] <burek> 32k is a valid bitrate for aac+
[22:30] <burek> and if you pay closer look, it does not complain about bitrate
[22:30] <burek> but about sample rate
[22:30] <burek> 22050
[22:30] <burek> which i didnt even specify
[22:31] <durandal_1707> when it worked before? one year? two year?
[22:31] <ubitux> michaelni: ah, just sent the patches
[22:32] <burek> durandal_1707, month or two ago
[22:32] <burek> all my builds were made around 20th november
[22:32] <burek> not older than that
[22:32] <ubitux> michaelni: aren't they deprecated?
[22:33] <durandal_1707> burek: i see nothing important in libaacplus that changed
[22:34] <burek> im trying 1.0.1 now
[22:34] <durandal_1707> perhaps cutoff?
[22:35] <michaelni> IIRC libaacplus could conflict with some other lib somehow so it would fail at runtime
[22:37] <durandal_1707> in mysterious ways?
[22:37] <burek> how could i debug it
[22:38] <burek> same with 1.0.1 :/
[22:39] <michaelni> IIRC the issue was something along the lines of 2 libs being based on the reference impl and having the same symbols
[22:39] <burek> how can i checkout version from say 20th novemeber or something... git checkout right?
[22:39] <michaelni> and depending on how they are linked (luck and versioning) it could fail
[22:39] <michaelni> but its IIRC
[22:39] <durandal_1707> burek: you have other aac libs enabled?
[22:39] <burek> only libaacplus
[22:43] <burek> anyway, is there any chance strace could help
[22:43] <burek> or something similar?
[22:43] <durandal_1707> burek: what if you set different sample rate does it report same one?
[22:43] <burek> durandal_1707 yes
[22:43] <burek> [22:11:55] <burek> same with: ffmpeg -f alsa -ar 44100 -ac 2 -i hw:0 -c:a libaacplus -ab 32 -ar 44100 -ac 2 -y out.mp4
[22:44] <burek> it ignores my input -ar and sets 22050
[22:44] <burek> which aac+ does not support iirc
[22:44] <durandal_1707> does same happen if you use file as input?
[22:45] <burek> i forgot how to generate random wav
[22:45] <burek> :D
[22:45] <durandal_1707> you dont have any audio file?
[22:45] <michaelni> burek, you need 32k not 32
[22:46] <burek> oh snap..
[22:46] <burek> let me see
[22:46] <burek> yup.. im an idiot...
[22:46] <burek> sorry guys..
[22:46] <burek> :S
[22:46] <burek> it works in 1.0.1 with 32k
[22:49] <burek> ok works in latest git too.. im really sorry for that :/ ill see how to reproduce the vlc's segfault now in ffmpeg only and let you know if i succeed
[22:51] <durandal_1707> perhaps something like supported_bitrates could be handy
[22:52] <durandal_1707> michaelni: libavformat/img2enc.c:81:127: warning: more '%' conversions than data arguments [-Wformat] ...pattern '%s' (either set updatefirst or use a pattern like %03d within the filename pattern)\n",
[22:55] <durandal_1707> should array inside function be really declared static?
[22:56] <durandal_1707> it really should not
[22:59] <cone-67> ffmpeg.git 03Michael Niedermayer 07master:2f436b1fce99: lavc: Warn in case the set bitrate is very low
[22:59] <cone-67> ffmpeg.git 03Michael Niedermayer 07master:165be91c2bd5: img2enc: fix typo
[23:05] <durandal_1707> hmm reading more static will move it to data segment
[23:05] <beastd> durandal_1707: where is that static array?
[23:05] <durandal_1707> af_volume
[23:07] <durandal_1707> there should be tool that will list all nonstatic and/or nonconst arrays
[23:09] <beastd> I am not sure about the problem. I can't currently imagine why it is a problem if it is "static const". Isn't it just hiding visibility to the scope of the function which may be legitimate?
[23:10] <durandal_1707> it is just moving it to another segment which may produce faster binary
[23:19] <beastd> durandal_1707: So I understand correctly you changed your mind and find it indeed ok for the arrays to be "static const" inside functions? And now you want to find other places in FFmpeg code base where "static const" would be an improvement?
[23:20] <durandal_1707> beastd: there are filters that already do that
[23:21] <durandal_1707> beastd: you have explanation that this is bad idea?
[23:28] <beastd> Don't know if I understand you correctly. From what I know now it might be a good idea for many places. But I am not sure it applies to all places... Anyway I have no clue if it is a bad idea really.
[23:36] <burek> durandal_1707, sorry for bugging you, does this mean that the segfault is in ffmpeg: http://pastebin.com/X20na5Cr
[23:36] <burek> im just not sure what exactly to type in cmd line to reproduce it (if it is)
[23:37] <burek> the cmd line in vlc just reads alsa input and encodes it to aac (i guess using libav libs) to out.mp4
[23:38] <burek> (in vlc ab=32 means 32k, just to avoid the same confusion again)
[23:38] <durandal_1707> you can reproduce crash with vlc?
[23:38] <burek> yes, every time
[23:40] <durandal_1707> dunno, perhaps vlc use older version or it uses it incorrectly
[23:40] <burek> i compiled latest ffmpeg from git and latest vlc from git
[23:40] <burek> no repository versions installed on the system at all
[23:41] <burek> #1  0x00007fffe9e9a572 in av_samples_copy () from /usr/local/lib/libavutil.so.52
[23:41] <burek> #2  0x00007fffea5cde09 in avcodec_encode_audio2 () from /usr/local/lib/libavcodec.so.54
[23:41] <burek> and from latest ffmpeg
[23:41] <burek>   libavutil      52. 12.100 / 52. 12.100
[23:41] <burek>   libavcodec     54. 81.100 / 54. 81.100
[23:41] <burek> so i assume it uses the latest ffmpeg
[23:42] <durandal_1707> vlc may use old api
[23:45] <durandal_1707> does vlc compiled with 2months old ffmpeg still crash?
[23:45] <burek> let me quick compile it
[23:46] <burek> ill try with FFmpeg 0.11.2 "Happiness"
[23:47] <durandal_1707> i think that is before new encoding api is introduced
[23:47] <durandal_1707> but it really looks like vlc problem
[23:48] <durandal_1707> can you pick some other codecs with vlc?
[23:48] <burek> yes
[23:48] <burek> aenc=...
[23:48] <durandal_1707> and others work?
[23:49] <burek> i didn't compile anything except aacplus :) and gpl stuff in ffmpeg that i dont use that much, so i dont know which one to try :)
[23:50] <durandal_1707> flac
[23:50] <durandal_1707> alac,mp2,ac3
[23:51] <burek> ill try it now
[23:56] <burek> durandal_1707, same thing with flac
[23:56] <burek> #0  0x00007ffff1fb4dea in ?? () from /usr/local/lib/libavcodec.so.54
[23:56] <burek> #1  0x00007ffff225dadb in avcodec_encode_audio2 () from /usr/local/lib/libavcodec.so.54
[23:57] <burek> ill try the older ffmpeg just to see what happens
[00:00] --- Sat Dec 22 2012


More information about the Ffmpeg-devel-irc mailing list