[Ffmpeg-devel-irc] ffmpeg-devel.log.20150326

burek burek021 at gmail.com
Fri Mar 27 02:05:02 CET 2015


[00:06:00 CET] <jamrial> philipl: no problem
[00:50:34 CET] <cone-007> ffmpeg 03Himangi Saraogi 07master:1301aa55f645: avfilter/vf_telecine: Fix AV desync by using the first input timestamp
[02:19:05 CET] <cone-007> ffmpeg 03Michael Niedermayer 07master:df43d03731c1: avformat/rmdec: fix support for 0 sized mdpr
[05:09:34 CET] <cone-007> ffmpeg 03Michael Niedermayer 07master:2139e584391b: avformat/dv: fix used dimension variables
[05:11:21 CET] <klaxa> i hope this is not too off-topic, but what am i expected to write in my proposal for gsoc?
[06:47:02 CET] <electronrotoscop> hello. I've found a feature that's partially implemented in mencoder and I've found what I think is the code they use but I don't know really much about how to code myself. I think it would be useful for ffmpeg, is there a right place to suggest including it in ffmpeg's code?
[06:51:48 CET] <rcombs> electronrotoscop: maybe say what the feature is
[06:52:13 CET] <electronrotoscop> MPEG2 soft pulldown as sort of explained in https://trac.ffmpeg.org/ticket/2602
[06:53:37 CET] <electronrotoscop> basically mpeg2 ES streams can be encoded with a flag that shows up in mediainfo as Scan type : Progressive | Scan order : 2:3 Pulldown and DGPulldown has the ability to add that flag to MPEG PS or TS or ES files
[06:54:35 CET] <electronrotoscop> but DGPulldown requires rewriting the whole thing to a new file, and it's a windows-only binary. MEncoder can do it live, but it seems to only be included in the MPEG-PS muxer code so it doesn't work writing out to raw MPEG-ES m2v files
[06:56:37 CET] <electronrotoscop> If this were included in ffmpeg, then video streams that could be encoded to progressive DVDs could be created in one step, which would be pretty cool I think! The code seems to be lines 2468 - 2491 of muxer_mpeg.c in libmpdemux in the latest .tar i just downloaded from mencoder, but I know very little about C and how these things work
[06:57:10 CET] <electronrotoscop> anyways maybe that was more information than you needed
[07:34:40 CET] <akshay_> hello this is Akshay
[07:35:04 CET] <akshay_> I was working on implementing websockets for ffmpeg
[07:35:55 CET] <akshay_> I want to talk to someone experienced in websockets
[07:36:13 CET] <akshay_> If someone had any info of it, please ping
[07:48:43 CET] <akshay_> I have to write code that would help ffmpeg to send data over websocket protocol
[11:22:00 CET] <michaelni> klaxa, it doesnt really matter much, what you write in it but it should say what you will work on in summer, you could mention how many hours you will spend per  per week working on it in summer, what kind or prior experience you have with similar work, stuff like that but again its not that important
[12:53:27 CET] <kierank> electronrotoscop: you'd probably have to write a bitstream filter
[14:26:17 CET] <electronrotoscop> kierank: that's just the thing, I can't really write C myself. I've never suggested an ffmpeg feature before, I was wondering if there's a good place to do it...
[14:26:30 CET] <kierank> open a ticket
[14:26:35 CET] <kierank> as a feature request
[14:27:41 CET] <Daemon404> fwiw, dgpulldown is gpl
[14:27:48 CET] <Daemon404> and very small code
[14:27:56 CET] <Daemon404> should be easy to look at or port or w/e
[14:28:33 CET] <electronrotoscop> this is what I was thinking, especially since someone's already taken the important parts into mencoder's MPEG-PS muxer
[14:28:53 CET] <nevcairiel> typical mencoder, putting it into the muxer =p
[14:29:59 CET] <electronrotoscop> ha ha ha I tell you I'm almost relieved, it means i don't have to learn a whole new system for writing batch commands
[14:33:29 CET] <electronrotoscop> kierank should I add the information to the ticket https://trac.ffmpeg.org/ticket/2602 or start a new one? sorry I'm new to all this
[14:53:48 CET] <electronrotoscop> I went ahead and added it to the ticket. Hope that was okay.
[15:55:58 CET] Action: Daemon404 watches selinux be painfully slow to update on arm
[16:17:49 CET] <cone-433> ffmpeg 03Michael Niedermayer 07master:71258e5d87cb: avcodec/ac3dec: Put CONFIG* first in if()
[16:17:50 CET] <cone-433> ffmpeg 03Michael Niedermayer 07master:47aee6f047d0: avcodec/arm/flacdsp_init_arm: Put CONFIG_FLAC_DECODER first in if()
[16:17:51 CET] <cone-433> ffmpeg 03Michael Niedermayer 07master:cd5a48e54984: avcodec/h264_slice: Put CONFIG_GRAY first in if()
[16:17:52 CET] <cone-433> ffmpeg 03Michael Niedermayer 07master:0245abc7c1a5: avcodec/x86/hpeldsp_init: Put CONFIG_* first in if()
[19:04:03 CET] <klaxa> michaelni: okay thanks that cleared it up for me
[21:51:55 CET] <jamrial> nevcairiel: your samplerate "hack" fixes the above
[21:52:45 CET] <nevcairiel> i know
[21:52:52 CET] <nevcairiel> but i dont know if it has any side effects
[21:54:50 CET] <Daemon404> shoulda used haskell
[21:55:24 CET] <jamrial> [libdcadec @ 000000000034c5c0] dcadec_context_parse() failed: 2 (Invalid bitstream format)
[21:55:43 CET] <jamrial> with or without your patch, so guess it doesn't really fix it :p
[22:02:28 CET] <wm4_> Daemon404: apparently all real haskell code is full of side effects
[22:09:27 CET] <nevcairiel> jamrial: an incomplete frame from a bad cut could always cause that
[22:11:46 CET] <jamrial> i thought it could be that, but oddly enough dcadec (the tool) doesn't complain even in strict mode
[22:12:26 CET] <nevcairiel> maybe its dts stream reader filters out bad frames at the start
[22:12:40 CET] <nevcairiel> do you know if thats the first frame?
[22:13:22 CET] <jamrial> no
[22:29:26 CET] <jamrial> i think it's because dcadec_stream_read() (which we don't use in the wrapper since it's used to read from a file) returns 0 when a frame is incomplete, and dcadec simply stops gracefully in that case
[22:51:06 CET] <j-b> wm4_: why do you want clean code? I mean, I can get 0.0564% improvement on a p3!
[22:52:39 CET] <wm4_> if only all unclean code at least had a performance benefit on some ancient architecture
[22:53:50 CET] <wm4_> but in ffmpeg we keep unclean code because... fuck I don't even know why
[22:54:02 CET] <j-b> wm4_: hey, don't touch my xvmc acceleration!
[22:54:15 CET] <j-b> soo valuable, and fast
[22:54:49 CET] <wm4_> why does ffmpoeg have a svp3 encoder? (or whatever it was)
[22:56:02 CET] <wm4_>  DEV.L. svq1                 Sorenson Vector Quantizer 1 / Sorenson Video 1 / SVQ1
[22:56:12 CET] <wm4_> no svq3 encoder unfortunately, only decoder
[22:56:19 CET] <wm4_> WE NEED A SCQ3 ENCODER
[23:19:30 CET] <wm4_> it's funny how Libav wanted to cleanup shit, but they also have the svq1 encoder
[23:55:28 CET] <shanlar-> is it possible to set audio as option for a ffserver stream?
[23:55:33 CET] <shanlar-> sometimes the stream has audio and sometimes it doesnt. when it doesnt have audio ffmpeg errors out with "Missing audio stream which is required by this ffm"
[00:00:00 CET] --- Fri Mar 27 2015


More information about the Ffmpeg-devel-irc mailing list