[Ffmpeg-devel-irc] ffmpeg-devel.log.20141007

burek burek021 at gmail.com
Wed Oct 8 02:05:02 CEST 2014


[00:05] <cehoyos> J_Darnley, arwa: Porting eq / eq2 is definitely welcome, if you are searching for easier filter tasks, please see the OPW site, it contains three tasks and a link to a patch that should make one of the tasks easier as qualification tasks for the VDPAU filter project.
[02:40] <cone-703> ffmpeg.git 03Rémi Denis-Courmont 07master:153fadc390d0: ff_get_format: fix infinite loop
[02:40] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:a52384dcdab2: Merge commit '153fadc390d05aa47e5e2c56290401898fe41a23'
[02:46] <cone-703> ffmpeg.git 03Martin Storsjö 07master:b76249443864: sdp: Simplify parsing/conversion of H264 extradata
[02:46] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:4db0e8fd3366: Merge commit 'b76249443864c88ffb2d41ab8d1de7432e985dc7'
[03:04] <reynaldo> can someone get in touch with merbanan see whether he is going to be mentoring the DTS/DCA Improvements ?
[03:04] <reynaldo> I dont have his current email address
[03:04] <reynaldo> merbanan: assuming you are not around to answer of course :)
[03:07] <cone-703> ffmpeg.git 03Martin Storsjö 07master:28816050e47b: lavf: Set the stream time base hint properly for chained muxers
[03:07] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:3726d07a2af8: Merge commit '28816050e47b6dba430a52e429d21a864cffda8e'
[03:31] <cone-703> ffmpeg.git 03Mark Reid 07master:5c50214eed19: avformat/mxfdec: read reel_name and source timecode from physical source package
[03:49] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:b60938e4fd05: avcodec/avpacket: simplify freeing pkt->data
[03:49] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:c9eac8062e6c: avcodec/avpacket: use av_freep(), do not leave stale pointers in memory
[03:49] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:1c5647f419db: avcodec/proresenc_kostya: use av_freep(), do not leave stale pointers in memory
[03:49] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:f0d1b3acdfb8: avcodec/escape124: use av_freep(), do not leave stale pointers in memory
[04:22] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:30ffb80dcaf7: avcodec: fix aac/ac3 parser bitstream buffer size
[04:22] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:48443598ed51: avcodec/utils: add GBRP16 to avcodec_align_dimensions2()
[04:22] <cone-703> ffmpeg.git 03Timothy Gu 07release/1.2:0f578ea70521: bktr: Fix Fabrice's name
[04:22] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:1482253790c6: avcodec/snow: check coeffs for validity
[04:22] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:248a2fca9084: avformat/swfdec: Use side data to communicate w/h changes to the decoder
[04:22] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:6f02de4ee8ca: avformat/swfdec: Do not change the pixel format
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:0fc229450fd9: avcodec/mpegvideo: Use "goto fail" for all error paths in ff_mpv_common_frame_size_change()
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:0f1882b9b2e4: avcodec/mpegvideo: check that the context is initialized in ff_mpv_common_frame_size_change()
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:d49b57fe0f74: avcodec/mpegvideo: Set err on failure in ff_mpv_common_frame_size_change()
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:b9b97900c19c: avformat/m4vdec: Check for non startcode 00 00 00 sequences in probe
[04:23] <cone-703> ffmpeg.git 03Katerina Barone-Adesi 07release/1.2:70dc7bb893db: apetag: Fix APE tag size check
[04:23] <cone-703> ffmpeg.git 03James Almer 07release/1.2:9b9048a3ddfe: x86/dsputil: add emms to ff_scalarproduct_int16_mmxext()
[04:23] <cone-703> ffmpeg.git 03Gianluigi Tiesi 07release/1.2:e28c27e25f30: avcodec/libilbc: support for latest git of libilbc
[04:23] <cone-703> ffmpeg.git 03lvqcl 07release/1.2:2144ce08c1de: avutil/x86/cpu: fix cpuid sub-leaf selection
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:e9d85860d0f3: avcodec/ac3enc_template: fix out of array read
[04:23] <cone-703> ffmpeg.git 03Reimar Döffinger 07release/1.2:997bf49b1c88: configure: add noexecstack to linker options if supported.
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:0bd9b78fb6c4: avcodec/jpeglsdec: Check run value more completely in ls_decode_line()
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:0ffa44340f24: avcodec/mjpegdec: check bits per pixel for changes similar to dimensions
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:e90d620cb93e: avcodec/utils: Add case for jv to avcodec_align_dimensions2()
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:76601e4ab8f5: avcodec/mmvideo: Bounds check 2nd line of HHV Intra blocks
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:f56095c4d7e5: avcodec/tiff: more completely check bpp/bppcount
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:12c8e4021c2d: avcodec/pngdec: Check bits per pixel before setting monoblack pixel format
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:665f70209a95: avcodec/pngdec: Calculate MPNG bytewidth more defensively
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:75f5fe165ced: avcodec/cinepak: fix integer underflow
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:24a5cd720d57: avcodec/gifdec: factorize interleave end handling out
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:891376d26108: avcodec/qpeg: fix off by 1 error in MV bounds check
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:4865948d2ea9: avcodec/smc: fix off by 1 error
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:c4f0f3c52d97: avformat/mpegts: Check desc_len / get8() return code
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:7fc97160c27d: avcodec/h264: Check mode before considering mixed mode intra prediction
[04:23] <cone-703> ffmpeg.git 03Michael Niedermayer 07release/1.2:65889b62b3eb: swresample/swresample: fix sample drop loop end condition
[04:36] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:0026c4ecfd7b: avformat/hls: use av_freep(), do not leave stale pointers in memory
[10:33] <hima> Hi ubitux. here is my first attempt for the demuxer for kate format. I am a unsure about the get_header function please have a look
[10:33] <hima> http://fpaste.org/139805/70713141/
[10:34] <ubitux> kate_read_header() is where all the parsing of the format happens
[10:35] <ubitux> you need to look for "events" (or something like that IIRC)
[10:35] <ubitux> and when you focus on these events, you try to extract the timestamp and text
[10:36] <ubitux> and use ff_subtitles_queue_insert() to append the event
[10:36] <ubitux> i'm not sure a "line reading" logic will be very helpful, but you can start with that if that's simpler for you
[10:37] <ubitux> like, you could ff_get_line() and look for the "event" string in it
[10:37] <hima> ubitux i have used that ff_subtitles_queue_insert()
[10:37] <hima> yeah umm did you see the funtions probe and get_pts 
[10:38] <ubitux> the current probe function can't work
[10:38] <ubitux> sscanf needs to focus on the beginning on the timestamp
[10:38] <ubitux> it's the same issue with get_pts
[10:38] <ubitux> you need to make sure it focuses on the timestamp before that
[10:39] <hima> timestamp before that i did not get you 
[10:40] <ubitux> how does an event look like in kate?
[10:41] <ubitux> event { 00:00:05 --> 00:00:10 "This is a text" } 
[10:41] <ubitux> it looks like this, right?
[10:41] <hima> yes 
[10:41] <ubitux> so you need to sscanf on "00:00:05 --> 00:00:10 ..."
[10:42] <ubitux> and skip the "event {"
[10:42] <ubitux> note that it can be "event           { \n\n\n             " or something like that
[10:42] <ubitux> so you need to move your pointer at the beginning of the timestamp
[10:42] <ubitux> otherwise it won't work
[10:43] <hima> ohh i thought event is a format of kate 
[10:43] <hima> http://fpaste.org/139807/12671428/
[10:44] <hima> this is should work then?
[10:44] <ubitux> only if ptr is pointing on "00:00:05 --> 00:00:10 ..."
[10:44] <ubitux> and not "event { 00:00:05 --> 00:00:10 ..."
[10:45] <ubitux> errh, you put the { in it
[10:45] <ubitux> but you need to skip the "event" at least
[10:45] <ubitux> but you should take it out
[10:45] <hima> okay yeah i see :|
[10:45] <ubitux> because "event      {          00:00:05 --> 00:00:10 ..." won't work
[10:46] <ubitux> the syntax looks kind of flexible, so you need a code capable of dealing with that
[10:49] <hima> ok i will take care of that 
[10:50] <hima> ubitux : if i complete this task then only i am capable of writing a proposal for OPW is it?
[10:51] <ubitux> i don't understand your question
[10:52] <hima> um we have to write a proposal for application in OPW before 22nd october right?
[10:52] <hima> and the we get selected according to the porposal and the interaction with the mentor 
[10:52] <ubitux> i really need to read these rules carefully...
[10:53] <ubitux> i'm going to say "yes probably" but i need to look it up
[10:54] <hima> um okay. Can i know how many applicants might get selected for this particula project?
[10:54] <hima> is there a number?
[10:54] <ubitux> we have only one slot, we might get more if we get another sponsor
[10:55] <hima> okay one slot for this subtitle project or including the projects that were mentioned?
[10:56] <ubitux> one slot for the FFmpeg project unfortunately
[10:57] <hima> okay. 
[10:59] Action: hawken read backlogs
[10:59] <hawken> guys...
[10:59] Action: hawken sighs
[11:09] <ubitux> hawken: did i say something wrong?
[11:13] <hawken> ubitux: No, I just read further
[11:13] <hawken> and got this link http://paste.ubuntu.com/8488924/
[11:13] <hawken> jeez -_-
[11:13] <hawken> I don't even think you were present at the place I read in my backlog :P
[11:13] <ubitux> haha
[11:13] <ubitux> yeah i wasn't there
[11:14] <hawken> :P
[11:26] <prachi> Hello! I am interested in "Subtitles" project of Outreach Programme for Women(OPW) and i would like to be a part of it. I am new to open source and i was looking for some guidance on how to get started with the qualificaton task. Thanks in advance :)
[11:28] <ubitux> hey guys i think i'll need some help with this
[11:28] <ubitux> anyone willing to back me up a little? :D
[11:28] <ubitux> hi prachi 
[11:29] <ubitux> you are the 4th candidate for the subtitles task, but i don't have much task/challenge to give you right now, i need to come up with one
[11:31] <prachi> okay sir,I'll wait.
[11:31] <ubitux> prachi: first, if you are interested in subtitles, you should read this: http://b.pkh.me/opw-subtitles.log
[11:31] <ubitux> prachi: this is the backlog of an explanation i had with the first applicant a few days ago
[11:31] <ubitux> she has to develop one demuxer, so this explains how stuff are stuck together
[11:32] <ubitux> meanwhile, i guess i'm going to try to find something for you
[11:32] <prachi> Sure,i'll read it in the meantime. Thanks :)
[11:49] <pross> when it rains, it pours.
[12:28] <cone-133> ffmpeg.git 03Anton Khirnov 07master:79a60c8e7792: hevc_mvs: initialize the temporal MV in case of missing ref
[12:28] <cone-133> ffmpeg.git 03Michael Niedermayer 07master:018ca8b7f6a8: Merge commit '79a60c8e779242b5ba4c531b2c706c871e8e6420'
[12:42] <cone-133> ffmpeg.git 03Anton Khirnov 07master:d208d1eba379: hevc_mvs: make sure to always initialize the temporal MV fully
[12:42] <cone-133> ffmpeg.git 03Michael Niedermayer 07master:33c8500db4e2: Merge commit 'd208d1eba3799c58fd6d3602d31de3e686f14aec'
[13:03] <cone-133> ffmpeg.git 03Josh Allmann 07master:9a03c2323593: h263dec: Force padding bug workaround for H.263.
[13:03] <cone-133> ffmpeg.git 03Michael Niedermayer 07master:aba873bbd057: Merge commit '9a03c2323593173a201cb75edd1b49887cf811ed'
[14:10] <pross> s/when it rains it pours/there is a lot of interest in subtitles here lately/g
[14:10] <benoit-> pross: :)
[14:11] <ubitux> it's just that the tasks look probably easier
[14:11] <Daemon404> eh
[14:12] <Daemon404> subtitles have proven to be anything but easy'
[14:12] Action: Daemon404 preaches to the choir
[14:13] <ubitux> it's not easy, and it's freaking boring as well
[14:13] <ubitux> though, i'm kind of happy to see interest in it
[14:14] <ubitux> i'm still waiting for someone taking over it
[14:15] <iive> but you know more than anybody else... you are our expert on the topic!
[14:17] <ubitux> really, not that much
[14:17] <ubitux> it's only limited to text subtitles, and i just basically know the main chain and where to go
[14:18] <ubitux> anyone familiar with ffmpeg can become "the text subtitles expert" in about an hour
[14:20] <Daemon404> yeah but noone wants to :D
[14:21] <pross> nods
[14:28] <cone-133> ffmpeg.git 03Michael Niedermayer 07master:2d6f31730c60: avfilter/af_adelay: use av_freep(), do not leave stale pointers in memory
[14:28] <cone-133> ffmpeg.git 03Michael Niedermayer 07master:158b3fb58485: avformat/tiertexseq: use av_freep(), do not leave stale pointers in memory
[15:14] <cone-133> ffmpeg.git 03Clément BSsch 07master:abaf4245a172: avfilter: add w/h parameters in timeline
[15:14] <ubitux> my git push seems stalled
[15:15] <ubitux> ah, it just took a very long time
[15:16] <BtbN> sounds like github
[15:17] <benoit-> Is there some "good practice" to follow to concatenate the data of two AVPacket packets?
[15:18] <benoit-> I'm still working on the EAC3 muxing in mov, and I need to concatenate syncframes when/if needeed
[15:18] <ubitux> wm4: the file from https://trac.ffmpeg.org/ticket/2176 doesn't seem to be decoded by your sup demuxer
[15:25] <ubitux> i guess SP ` PG
[15:43] <arwa> I read about xbr . I was curious if there exists a filter of similar kind(eg hqx etc) in ffmpeg so that I can use that as a base to build upon the new filter.
[15:45] <arwa> Also, I wanted to know what is my qualification task, and the site shows that the mentor is not yet decided :/ 
[15:46] <ubitux> i can help you with the filter thing
[15:46] <ubitux> hqx is available
[15:47] <ubitux> there is also super2xsai but it's under gpl
[15:48] <ubitux> so you should prefer hqx
[15:48] <arwa> okay :D
[15:48] <arwa> I will go through the hqx filter.
[15:49] <ubitux> note that hqx is already kind of advanced
[15:49] <ubitux> it has threading support typically
[15:50] <ubitux> you have tons of examples in the libavfilter directory anyway
[15:51] <reynaldo> anyone doing cloudopen/ELCE/lixcon europe/gstconf at dusseldorf next week? I plan to be there in case someone wants to go for a beer or 10 :)
[15:51] <reynaldo> linuxcon/
[15:52] <arwa> okay...So what is the best option to start with?
[16:00] <J_Darnley> arwa: what do you mean by "option"?
[16:00] <ubitux> arwa: i'd say to follow doc/writing_filters.txt and just take the config_output from libavfilter/vf_hqx.c
[16:00] <ubitux> and eventually look at the mainloop from hqx_filter
[16:01] <ubitux> edgedetect mentioned in doc/writing_filters.txt is mostly simple
[16:01] <reynaldo> arwa: did you take a look at https://trac.ffmpeg.org/query?status=new&status=open&status=reopened&component=avfilter&col=id&col=summary&col=status&col=type&col=priority&col=component&col=version&order=priority already? There are some interesting tasks there, like https://trac.ffmpeg.org/ticket/3302 although it can prove too involved as a quali, not sure
[16:02] <ubitux> ah note that i just suggested xbr because i find it interesting, but you might be short in time for it
[16:02] <ubitux> and would probably want to instead just work on porting one of the remaining filters
[16:03] <ubitux> what happened to the eq/eq2 port already? J_Darnley; you were the one working on it? i don't remember..
[16:06] <arwa> So, should I not go for xbr filter right now?
[16:06] <arwa> Which filter do I choose??
[16:08] <ubitux> well you can do xbr if you feel like it, it's a cool and documented filter
[16:09] <ubitux> if you think it's too hard we can give you another task
[16:09] <ubitux> basically, pick something that interest you, and we'll say if it's appropriate for evaluating you
[16:09] <J_Darnley> ubitux: I was.  I got bored, stared playing minecraft, got bored of that, now I'm back on programming.
[16:10] <ubitux> haha
[16:11] <arwa> I dont know actually.....I have never written a filter in C. So, I am kind of confused what I should do right now :/
[16:12] <arwa> I have worked with opencv,python,c++, and matlab for writing filters
[16:13] <ubitux> arwa: well, do you feel like xbr is hard?
[16:13] <ubitux> plugging yourself into libavfilter is really trivial if you follow writing_filters.txt
[16:14] <ubitux> the difficulty lies into xbr, not really the framework here
[16:14] <ubitux> i don't know your skills and how confident you are with them
[16:14] <arwa> I went through the algo...it didnt seem that hard.
[16:14] <ubitux> well then go for it
[16:15] <arwa> Wish me luck :D
[16:15] <ubitux> read carefully the writing_filters.txt, and just ask specific questions
[16:15] <ubitux> sure, good luck :)
[16:15] <arwa> Okay :d
[16:28] <arwa> I am not getting a clear idea of "AVFilterLink"
[16:29] <arwa> Does it refer to the the link between frames?
[16:30] <J_Darnley> no between filters
[16:30] <ubitux> in the filtergraph
[16:31] <ubitux> arwa: if you do -vf edgedetect,scale,crop there is a link between edgedetect and scale, and a link between scale and crop
[16:31] <ubitux> (and there are actually one before edgedetect, and one after crop)
[16:33] <arwa> by link you mean that we are calling a filter in another filter?
[16:33] <ubitux> i mean AVFilterLink
[16:35] <arwa> As in are we trying to use an already existing filter from the filter we are making?
[16:36] <ubitux> filters are linked together in a filtergraph ("edgedetect,scale,crop" is an example of filtergraph)
[16:36] <ubitux> a filter has no knowledge of previous and next filter
[16:36] <ubitux> but they see "links"
[16:36] <ubitux> so when they need input frames, they look at the input link
[16:37] <ubitux> which will tell/give them the input frames
[16:37] <ubitux> and outlink is used to push the frames in the output
[16:37] <ubitux> the filter will push the frames to the next filter
[16:37] <ubitux> through that link
[16:40] <arwa> okay....!! What does flag and config_prop do?
[16:41] <ubitux> see line 164 in doc/writing_filters.txt
[16:42] <ubitux> you can leave flags empty for now
[16:42] <ubitux> see timeline section line 316 in doc/writing_filters.txt for an example of flag
[19:09] <ubitux> prachi: so about the qualification task, maybe someone else can comment but youtube probably has some subtitles format as discussed the other day
[19:10] <ubitux> i'm really not familiar with youtube, but someone may comment on this
[19:10] <ubitux> i saw several half-assed formats support, but nothing really about simply how to get samples
[19:10] <ubitux> wm4 might know more about it
[19:11] <ubitux> otherwise, maybe various other websites do have their own subtitles formats
[19:11] <ubitux> and we might find one in those
[19:21] <wm4> ubitux: that sup file is probably a hd-dvd sub
[19:25] <rcombs> ubitux: iirc it's something XML-based
[19:27] <rcombs> ubitux: here, have a sample: https://gist.github.com/11rcombs/ece7488b06e9a73a1057
[19:28] Action: wm4 finds it fascinating that ffmpeg has no xml parser yet
[19:31] <rcombs> but YouTube has an API to send back subtitles as SRT
[19:32] <ubitux> ah, so it's pointless then
[19:33] <ubitux> wm4: we have a utils to deal with those
[19:34] <wm4> rcombs: what about youtube captions
[19:34] <wm4> or whatever they're called
[19:34] <rcombs> you mean annotations?
[19:34] <wm4> right
[19:34] <rcombs> those are a mess
[19:35] <rcombs> I once wrote a parser and renderer in JS for those (years before YouTube's official one)
[19:35] <rcombs> to convert them to ASS, you'd need to create drawings for a bunch of stuff
[19:36] <ubitux> i guess i could assign prachi to that format anyway
[19:36] <ubitux> for the exercise
[19:36] <ubitux> anyone has a better idea?
[19:37] <rcombs> ubitux: I wonder if ffmpeg _should_ link an XML parser at this point
[19:37] <wm4> no, it should NIH one
[19:38] <ubitux> ff_smil_* are enough for those
[19:38] <wm4> and not a full parser, just a scanner
[19:38] <rcombs> or that, but only if all the existing ones are really bad
[19:38] <wm4> they are all bad
[19:38] <rcombs> it'd also be useful for DASH
[19:38] <ubitux> for hds and stuff an xml parser is needed
[19:38] <rcombs> hds?
[19:38] <ubitux> but for these subs, really it's not really needed
[19:38] <ubitux> yeah, the adobe crap of hds/hls/...
[19:38] <ubitux> you know when it was a thing one or two years ago
[19:39] <rcombs> dunno
[19:39] <ubitux> hds/hls/dash/smoothstreaming
[19:39] <rcombs> ah, smoothstreaming
[19:39] <ubitux> adobe did its own as well
[19:39] <ubitux> it's probably the worst of all
[19:39] <rcombs> oh, joy
[19:39] <ubitux> wbs can probably testify
[19:40] <rcombs> I've never had to deal with it
[19:40] <ubitux> so well anyway prachi, let's just do this
[19:40] <ubitux> prachi: make a demuxer able to support https://gist.githubusercontent.com/11rcombs/ece7488b06e9a73a1057/raw/00b8c37773164861f47e403a42ee16081cbb9908/gistfile1.txt
[19:41] <rcombs> we do HLS and smooth streaming (though that's dying) and DASH (which is a pain)
[19:41] <ubitux> prachi: take libavformat/vplayerdec.c as a reference
[19:41] <prachi> okay sir,
[19:41] <rcombs> and more and more clients are capable of just playing a live-encoded MKV stream these days
[19:41] <ubitux> prachi: you must be able to do ffmpeg -i gistfile1.txt out.srt
[19:41] <ubitux> that's your task
[19:42] <rcombs> well, maybe with a different extension or a -f input arg :)
[19:42] <ubitux> <?xml version="1.0" encoding="utf-8" ?><transcript>  looks enough
[19:42] <rcombs> ubitux: maybe the ff_smil_* functions should be named something more generic
[19:42] <ubitux> like what?
[19:42] <rcombs> ff_xml_*?
[19:42] <ubitux> it's not really xml
[19:43] <rcombs> yuck
[19:43] <ubitux> i like smil better because it suggests that it's just a tag soup crap
[19:43] <ubitux> like html
[19:43] <ubitux> which SAMI typically exploits fairly well
[19:44] <ubitux> rcombs: no xml parser can actually handle SAMI
[19:44] <rcombs> wow
[19:44] <rcombs> ff_sgml_?
[19:44] <rcombs> or is it not technically that either :|
[19:44] <ubitux> dunno
[19:44] <prachi> sir,I am a little confused with so much of these heavy words.Can you please tell me finally what am i supposed to do?
[19:44] <wm4> <ubitux> rcombs: no xml parser can actually handle SAMI <- huh?
[19:45] <ubitux> wm4: well, it's more html than xml
[19:45] <ubitux> it also has inline css etc
[19:45] <ubitux> an xml parser actually requires a properly defined structure
[19:45] <ubitux> otherwise it just breaks
[19:46] <wm4> ok, I thought it was real xml
[19:46] <ubitux> http://blackhole.ubitux.fr/sami/Intro_to_SMTP_messaging.smi
[19:46] <ubitux> look
[19:46] <ubitux> tag aren't even closed
[19:47] <ubitux> http://blackhole.ubitux.fr/sami/testvideo.smi look at this one
[19:47] <ubitux> enjoy <SAMIParam>
[19:50] <wm4> who ever thought that sgml was a good idea
[19:59] <ubitux> i can't believe this guy is using the enable system with the resolution
[19:59] <ubitux> it's fun
[20:00] <ubitux> but reminds me of xkcd #1172
[20:04] <cone-769> ffmpeg.git 03Michael Niedermayer 07release/1.2:e0465d6fcf3a: Update for 1.2.9
[20:15] <Jianyong> Hello there
[20:15] <Jianyong> I found an issue with audio generation for webm-dash using ffmpeg
[20:15] <Jianyong> https://groups.google.com/a/webmproject.org/forum/#!topic/webm-discuss/4TAUht7t8DU
[20:15] <Jianyong> and I want to apply a fix for it
[20:16] <Jianyong> can anyone help to verify this to see if it is valid issue?
[20:22] <jamrial> if you think it's a bug, open a trac ticket or email Vignesh Venkatasubramanian (he's the dash maintainer)
[20:23] <Jianyong> ok, thanks, do you have his email?
[20:23] <jamrial> yeah, vigneshv at google com (not gmail)
[20:24] <Jianyong> cool!
[20:24] <Jianyong> thanks!
[20:37] <arwa> What is the use of implementing threads in hqx filter??
[20:38] <arwa> Do I need to use it in xbr also??
[20:39] <ubitux> no you can ignore the threading as i said
[20:39] <ubitux> and the use is obviously... performance
[20:42] <arwa> okay
[20:51] <arwa> What does " ctx->internal->execute" do??
[20:51] <arwa> is it used to execute threads?
[20:57] <ubitux> arwa: it's only thread related
[20:57] <ubitux> as i said, look at edgedetect for the logic
[20:58] <ubitux> and only pick config_props and look at the main hqx function hqx_filter() 
[20:58] <arwa> okay
[20:59] <cone-769> ffmpeg.git 03Clément BSsch 07fatal: ambiguous argument 'refs/tags/n1.2.9': unknown revision or path not in the working tree.
[20:59] <cone-769> Use '--' to separate paths from revisions
[20:59] <cone-769> refs/tags/n1.2.9:HEAD: avfilter: add w/h parameters in timeline
[21:06] <arwa> Where can I have a look on all the predefined structure in ffmpeg?
[21:06] <Compn> which structure ? api ? file structure ?
[21:06] <Compn> code paths ?
[21:07] <arwa> no
[21:07] <arwa> the "struct" used in C coding
[21:10] <ubitux> the public structures are split in different headers accross the code base
[21:12] <J_Darnley> ... so if you want a definitive answer you will have to tell us what struct you are looking at.
[21:24] <Compn> maybe arwa wants the full list of global strucs ?
[21:26] <J_Darnley> `find -name '*.h' -exec grep 'struct AV'`?
[21:29] <anshul_mahe> I use, grep -rn "struct AV" --include \*.h
[21:32] <ubitux> git grep 'typedef struct AV' -- '*.h'
[21:49] <arwa> thanks :D
[21:50] <arwa> btw I am getting too confused....how do I go about starting to write the filter?
[21:53] <anshul_mahe> there is null filter, look in vf_null.c
[21:54] <anshul_mahe> i would have started there
[21:56] <arwa> okay...i wrote the basic constructor ...after that ?
[21:57] <arwa> i am confused in the flow of data in the codes that i have looked so far
[22:05] <anshul_mahe> it all depend on what you want to do, or trying to do
[22:05] <anshul_mahe> btw what are you trying to do
[22:16] <arwa> I am trying to implement xbr filter
[22:17] <ubitux> arwa: did you really read doc/writing_filter.txt? :P
[22:19] <arwa> Not exactly ;)
[22:19] <arwa> Just skimmed through it
[22:20] <ubitux> maybe you should
[22:20] <ubitux> we could paste it linearly on irc but that won't make much difference
[22:24] <arwa> Actually I read it till AVFilterlink....after that I am blank
[23:28] <cone-769> ffmpeg.git 03Tobias Rapp 07master:143a8a44f4ec: ffprobe: add -show_pixel_formats option
[23:28] <cone-769> ffmpeg.git 03Tobias Rapp 07master:b36b2c89dfd7: ffprobe: add pixel format flags output
[23:28] <cone-769> ffmpeg.git 03Tobias Rapp 07master:7d327d624262: ffprobe: add pixel format component bit depth output
[23:28] <cone-769> ffmpeg.git 03Tobias Rapp 07master:1acc9189b4c2: ffprobe: add pixel format chroma sub-sampling output
[00:00] --- Wed Oct  8 2014


More information about the Ffmpeg-devel-irc mailing list