[Ffmpeg-devel-irc] ffmpeg-devel.log.20150427

[burek]

[02:27:44 CEST] <cone-408> ffmpeg 03ZhouXiaoyong 07master:0ace686ae854: avcodec: optimize mathops for Loongson-3 v1
[03:43:06 CEST] <ScHlAuChi> hi, does this channel cover ffmpg2 too?
[03:57:40 CEST] <ScHlAuChi> So does this channel cover ffmpeg2 development too?
[03:59:20 CEST] <jamrial> unless you're talking about ffmpeg 2.x releases, i'm fairly sure no
[03:59:55 CEST] <ScHlAuChi> nope i do mean ffmpegsource2 - but guess not then :)
[04:02:21 CEST] <ScHlAuChi> thx anyway :)
[05:06:10 CEST] <prelude2004c> hey can anyone help me with a multiast problem?
[05:06:16 CEST] <prelude2004c> i can't seem to figure it out
[05:08:28 CEST] <cone-408> ffmpeg 03Michael Niedermayer 07master:46778ab212e2: avformat/aviobuf: Do not resize the buffer in ffio_ensure_seekback() if there is no read function
[07:00:44 CEST] <ubitux> wm4: yes& :(
[07:17:03 CEST] <fluke_> Hello. I am developing videotoolbox hwaccel for iOS. I want to know where can I find hwaccel architecture documents/slides. 
[10:22:36 CEST] <ffmpeg-devel285> Hello
[10:24:17 CEST] <ffmpeg-devel285> I tried convert HEVC Raw video to MP4 (with codec copy) on ffmpeg
[10:25:11 CEST] <ffmpeg-devel285> and convert MP4 to yuv (codec copy) on ffmpeg
[10:25:49 CEST] <ffmpeg-devel285> compared yuv file and original raw video
[10:26:01 CEST] <ffmpeg-devel285> I found they different
[10:26:17 CEST] <ffmpeg-devel285> different hex value
[10:27:10 CEST] <ffmpeg-devel285> Is it normal?
[10:29:32 CEST] <JEEBsv> uhhh
[10:29:37 CEST] <JEEBsv> > to yuv > codec copy
[10:29:42 CEST] <JEEBsv> u wot m8t
[10:30:10 CEST] <ffmpeg-devel285> I want purely copy stream when convert RAW to MP4
[10:30:22 CEST] <JEEBsv> yes, but see your line from mp4 to yuv :P
[10:30:31 CEST] <JEEBsv> anyways, just pastebin your command lines
[10:30:36 CEST] <JEEBsv> and link here
[10:30:47 CEST] <ffmpeg-devel285> ok plz wait..
[10:35:18 CEST] <ffmpeg-devel285> http://pastebin.com/n9HnhYWt
[10:36:35 CEST] <ffmpeg-devel285> My problem is some data change when convert RAW to MP4
[10:41:46 CEST] <BBB> JEEBsv: when he says raw yuv, Im assuming he means annexb hevc
[10:42:02 CEST] <nevcairiel> how is that raw yuv
[10:42:35 CEST] <nevcairiel> and some data needs to change when you convert raw hevc to mp4, since it uses a different syntax
[10:44:16 CEST] <wm4> ffmpeg-devel285: #ffmpeg
[10:53:58 CEST] <ffmpeg-devel285> hm.. so, that's normal?
[11:21:09 CEST] <mop> hi all,
[11:26:42 CEST] <mop> I'm trying to cross compile ffmpeg on ubuntu for wind with mingw , I'm also trying to use https://github.com/dlfcn-win32/dlfcn-win32 but the test seems not works also if I set all flags to specify the includes path, ideas ? 
[11:27:29 CEST] <nevcairiel> why would you need that for ffmpeg
[11:28:25 CEST] <wm4> mop: I wouldn't use dlfcn for anything
[11:28:32 CEST] <wm4> I mean
[11:28:34 CEST] <wm4> dlfcn-win32
[11:29:14 CEST] <ubitux> fluke_: i don't think you'll find much doc about them but you can look at the already existing implementations
[11:29:33 CEST] <ubitux> i'm very much interested in the outcome of this btw
[11:29:40 CEST] <mop> I opened a pull request to be able to load filters dynamically so I need it (or an equivalent) to accomplish this task
[11:29:53 CEST] <mop> it works fine with linux
[11:30:05 CEST] <wm4> you can't load filters dynamically because that'd require use of private APIs
[11:30:09 CEST] <wm4> so there's no ABI
[11:30:13 CEST] <mop> https://github.com/FFmpeg/FFmpeg/pull/134
[11:30:24 CEST] <wm4> and dlfcn-win32 is not thread-safe
[11:30:25 CEST] <nevcairiel> windows has dll loading functions too
[11:31:02 CEST] <mop> wm4, my official intent is to be more quick to develop new filters to be proposed/integrated into ffmpeg
[11:31:25 CEST] <mop> so I just changed the policy in case of 
[11:31:32 CEST] <wm4> that's a very bad reason
[11:31:37 CEST] <wm4> just use the ffmpeg build system correctly
[11:31:52 CEST] <nevcairiel> its not exactly hard to just build avfilter with your new filter integrated, if thats your only reason
[11:31:57 CEST] <wm4> I wouldn't mind user-addable filters, but IMO the time is not ready yet
[11:32:15 CEST] <ubitux> yes i think the api needs more changes before opening it again
[11:32:16 CEST] <wm4> because of API and ABI issues
[11:32:34 CEST] <wm4> right, it was once open
[11:32:35 CEST] <ubitux> they are more important shortcomings to fix before opening it
[11:32:43 CEST] <ubitux> and opening it will prevent fixing them anytime soon
[11:33:01 CEST] <ubitux> or at best slow it down by a factor of several years
[11:33:31 CEST] <mop> wm4, this is the reason because I changed my original plans to add this feature as "users feature" and move to a more strict developers usage purpose 
[11:33:39 CEST] Action: wm4 is still dreaming of a light-weight ffmpeg filter framework
[11:33:59 CEST] <wm4> mop: what's the problem with developing with the current situation?
[11:34:45 CEST] <mop> wm4, I cannot have my own filters as separated repo
[11:35:06 CEST] <mop> (for example)
[11:35:07 CEST] <wm4> and?
[11:35:27 CEST] <wm4> currently there are no user filters, so that's a given
[11:35:52 CEST] <mop> I'm developing my own
[11:36:00 CEST] <wm4> and dynamic loading doesn't change anything about the API and ABI issues
[11:36:07 CEST] <wm4> mop: but you can?
[11:36:14 CEST] <mop> with my fix yes
[11:36:30 CEST] <wm4> you don't need your "fix"
[11:36:58 CEST] <wm4> what's the difference between using a separate repo and a branch in your ffmpeg git checkout?
[11:37:04 CEST] <mop> and how can I use my own filter without it  ?
[11:37:38 CEST] <mop> wm4, the weight and the time to compile it, for example
[11:37:57 CEST] <saste> wm4: the general idea is this, you enable a configure flag allowing to export the symbols
[11:38:07 CEST] <saste> this would be only enabled explicitly
[11:38:28 CEST] <saste> of course in case the internal ABI changes, the plugin filters will become useless
[11:38:33 CEST] <wm4> mop: ffmpeg has an incremental build system, so changing your filter doesn't require full recompilation
[11:39:28 CEST] <mop> wm4, I already know it, but I strongly prefer have a repo with just file needed by my functionalities and not the whole project
[11:39:45 CEST] <mop> is thousands time more simple to read and understand it
[11:40:11 CEST] <wm4> uh, you just don't understand how git works
[11:40:14 CEST] <mop> in any case, it is a feature turned off as default and just for developers
[11:40:22 CEST] <wm4> it's useless
[11:40:29 CEST] <wm4> it doesn't make anything easier
[11:40:43 CEST] <wm4> and it creates the false impression that libavfilter actually can do plugins
[11:41:31 CEST] <mop> it can, with instability of interfaces, but it can
[11:42:19 CEST] <wm4> which in turn means it absolutely has no value
[11:43:09 CEST] <mop> I respect your opinion but I disagree and I found more simple develop my filter in this way
[11:43:50 CEST] <mop> in any case, this will be the steps that you must performs when the interface will be stable enough
[11:44:20 CEST] <wm4> I doubt it
[11:44:27 CEST] <nevcairiel> just know that you are on your own then
[11:44:32 CEST] <wm4> for one, it shouldn't just load such things globally
[11:44:57 CEST] <wm4> so the final result will look differently anyway
[11:45:33 CEST] <mop> wm4, what you mean with load such things globally ? 
[11:45:57 CEST] <mop> my patch search for specific symbol and load it as filter, no other things
[11:46:14 CEST] <wm4> loading and adding them to a global list
[11:47:39 CEST] <mop> ah , so, also if the interface will be stable, third party filters will be loaded in a different manner ? 
[11:48:54 CEST] <wm4> at least I'd hope so
[11:50:53 CEST] <mop> what is the schema that you have in mind ? 
[11:52:28 CEST] <wm4> plugins should be registered to some kind of context, not globally
[11:59:45 CEST] <mop> in this way you will manage filters in different ways depending if are built-in or dynamic, what is the plus in doing it ? 
[12:00:01 CEST] <fluke_> @ubitux thank you. I have already implemented a videotoolbox hwaccel/decoder follow the vda implementation. I still want to explain the details to people, so I need to figure out more details for myself first. I don't understand the whole architecture much.  
[12:00:35 CEST] <ubitux> oh it's already done? is this public somewhere?
[12:00:41 CEST] <ubitux> ok bye.
[12:01:24 CEST] <ubitux> fluke_: 12:00 <@ubitux> oh it's already done? is this public somewhere?
[12:01:50 CEST] <ubitux> you can add something like doc/writing-hwaccel.txt or something
[12:02:07 CEST] <ubitux> feel free to ask question on the mailing list if you don't get any here
[12:02:23 CEST] <ubitux> currently, appart from the doxy and the code itself, you won't have much
[12:02:40 CEST] <ubitux> (any [answer])
[12:03:05 CEST] <fluke_> It's not public for now. I work for company. But I plan to open it in monthes.
[12:03:58 CEST] <ubitux> other companies are interested in such thing; if you don't release it soon enough, you might face competiting implementations
[12:04:07 CEST] <ubitux> and your work might not be upstreamed
[12:04:11 CEST] <ubitux> damn.
[12:04:28 CEST] <ubitux> 12:03 <@ubitux> other companies are interested in such thing; if you don't release it soon enough, you might face competiting implementations
[12:04:30 CEST] <ubitux> 12:03 <-- fluke_ (~fluke at 209.9.128.43) has quit (Read error: Connection reset by peer)
[12:04:32 CEST] <ubitux> 12:04 <@ubitux> and your work might not be upstreamed
[12:04:34 CEST] <ubitux> 12:04 <@ubitux> damn.    
[12:04:36 CEST] <ubitux> fluke_ ^
[12:04:38 CEST] <ubitux> (you should fix your connection...)
[12:04:57 CEST] <fluke_> What is hwaccel v2?
[12:05:29 CEST] <fluke_> freenode connection is not stable from my place.
[12:05:52 CEST] <mop> wm4, I read above, please
[12:06:47 CEST] <wm4> ubitux: you know there's an old videotoolbox patch somewhere on the ML?
[12:07:11 CEST] <ubitux> wm4: ah? i missed that
[12:07:39 CEST] <wm4> <mop> in this way you will manage filters in different ways depending if are built-in or dynamic, what is the plus in doing it ?  <- different API users in the same process shouldn't clash... consider different plugins with the same name
[12:08:24 CEST] <ubitux> wm4: oh indeed... (http://ffmpeg.org/pipermail/ffmpeg-devel/2012-September/130716.html)
[12:10:28 CEST] <ubitux> and nothing since then? :(
[12:11:39 CEST] <mop> scoping the name not  implies change the context necessarly
[13:14:53 CEST] <cone-653> ffmpeg 03Rong Yan 07master:603c839398f8: swscale/ppc/swscale_altivec.c: POWER LE support in yuv2planeX_8() delete macro GET_VF() it was wrong
[13:14:54 CEST] <cone-653> ffmpeg 03Rong Yan 07master:8a03d428f175: avcodec/ppc/fdctdsp: POWER LE support in ff_fdct_altivec() delete macros VEC_FMERGEH() VEC_FMERGEL(), they where wrong
[13:23:44 CEST] <fffan> I have a problem  with ffmpeg mp4 muxing, Can anybody help me.
[13:38:27 CEST] <fffan> hi, j-b, are you online?
[13:39:32 CEST] <j-b> maybe
[13:39:56 CEST] <fffan> I have a ffmpeg build(with fdkaac, without faac)
[13:40:42 CEST] <fffan> and use x264 and faac to genrate media data, and using ffmpeg to mux them to a mp4 file.
[13:40:51 CEST] <kierank> so when does outreachy get announced
[13:41:03 CEST] <fffan> but my ffplay can't play audio from the mp4
[13:41:36 CEST] <fffan> my question is : can a ffmpeg(without faac) muxing faac data into a mp4 file?
[13:41:43 CEST] <fffan> vlc is just ok.
[13:42:11 CEST] <fffan> my ffplay warns " aacDecoder_DecodeFrame failed"
[13:42:29 CEST] <fffan> my ffplay contains fdkaac, contains not faac.
[13:42:55 CEST] <fffan> since my old ffmpeg build(with faac) is just ok.
[13:43:03 CEST] <fffan> it bothers me.
[13:44:31 CEST] <fffan> could you give me some tip?
[13:44:55 CEST] <sfan5> fffan: ffmpeg should be able to mux and decode any aac whether encoded by faac, ffmpeg aac, libvo_aacenc or any other lib
[13:45:11 CEST] <sfan5> it could also help if you could pastebin the complete output somewhere
[13:45:43 CEST] <fffan> so, why? why fdkaac tell me aacDecoder_DecodeFrame failed? but vlc is just ok.
[13:46:02 CEST] <fffan> ok
[13:46:08 CEST] <fffan> I'll paste.
[13:49:03 CEST] <fffan> https://gist.github.com/rtmpnewbie/a7e45fa145942a4ee2e8
[13:49:26 CEST] <fffan> sorry, I can't access pastebin, I just use github.
[13:51:23 CEST] <fffan> sfan5, is github gist ok?
[13:52:04 CEST] <sfan5> looks like you're using libav
[13:52:13 CEST] <sfan5> the /topic says "uestions about using FFmpeg or developing with libav* libs should be asked in  #ffmpeg"
[13:52:19 CEST] <sfan5> Questions*
[13:52:45 CEST] <fffan> ok
[13:59:59 CEST] <Compn> fffan : depends if the aac parser was enabled at build :p
[14:01:43 CEST] <fffan> ?
[14:02:19 CEST] <fffan> my ffplay(without faac) can play any media file except my own.
[14:03:00 CEST] <fffan> my code works ok under my old ffmpeg build.
[14:03:11 CEST] <Compn> hang on lets talk in #ffmpeg
[14:03:16 CEST] <fffan> ok
[14:22:13 CEST] <cone-653> ffmpeg 03ZhouXiaoyong 07master:4a7b50679db3: makefile: add Loongson-3 support
[14:43:51 CEST] <ubitux> so mmh audio frames have no timestamps?
[14:49:19 CEST] <Daemon404> ubitux, define frame
[14:49:33 CEST] <ubitux> i must be doing something wrong
[14:49:43 CEST] <Daemon404> audio packets sure do..
[14:50:10 CEST] <ubitux> yeah that seems to be the case with ffmpeg, so i'm probably doing something wrong
[14:50:55 CEST] <vibr> kierank, to mimic the behavior of ffmpeg -f framecrc, i dont' have to reencode the decoded frames to rawvideo first, do i?
[14:52:43 CEST] <vibr> kierank, is it enough to decode the frames and then write them with the AVOutputformat set to framecrc?
[14:54:00 CEST] <ubitux> mmh ok, best_effort_timestamp doesn't work with audio
[14:54:06 CEST] <ubitux> is this on purpose?
[14:55:12 CEST] <wm4> ubitux: how does it not work?
[14:55:20 CEST] <wm4> oh let me guess, partial packet consumption?
[14:55:57 CEST] <ubitux> no idea, i get nopts with best_effort_ts, but i actually have a (probably) good value with f->pts
[14:57:03 CEST] <vibr> as kierank isn't her
[14:57:11 CEST] <vibr> e does anyone else know?
[15:01:37 CEST] <vibr> lglinskih, do you know when kierank is going to be here?
[15:06:00 CEST] <lglinskih_> vibr: no =(
[15:07:09 CEST] <ubitux> vibr: decoded frame means raw
[15:10:28 CEST] <wm4> ubitux: broken decoder? what decoder?
[15:10:52 CEST] <ubitux> flac
[15:10:54 CEST] <vibr> ubitux, so can i write the decoded frame with av_write_frame directly?
[15:11:25 CEST] <vibr> i thought i had to re-wrap it into a packet somehow
[15:11:34 CEST] <ubitux> vibr: idk what you are trying to do
[15:11:49 CEST] <vibr> ubitux: trying to mimic ffmpeg -f framecrc
[15:12:22 CEST] <ubitux> just run a CRC over the frame->data[0..N] planes of that decoded frame?
[15:12:48 CEST] <ubitux> ah mmh yeah format, sorry mmh
[15:12:59 CEST] <ubitux> lemme check what's going on here
[15:14:11 CEST] <vibr> i found out that setting "framecrc" as the third argument to avformat_alloc_output_context2 results in using that muxer (framecrc)
[15:14:41 CEST] <vibr> don't know if that's the correct way to set it
[15:14:52 CEST] <ubitux> yeah encode it with rawvideo to get a "packet"; you can look how it's done in lavc/rawenc.c
[15:15:14 CEST] <vibr> thanks, i'll have a look
[15:15:23 CEST] <ubitux> which is basically doing a call to av_image_copy_to_buffer() in the end
[15:15:35 CEST] <ubitux> so... crc(av_image_copy_to_buffer(frame))
[15:15:39 CEST] <ubitux> i would guess
[15:15:55 CEST] <ubitux> or you can pass it through the muxer to get packets if you prefer
[15:16:13 CEST] <ubitux> i don't exactly know what's your goal (if it's testing the whole chain or actually having the crc itself)
[15:17:17 CEST] <vibr> the goal is reproducing the crcs files found in http://git.videolan.org/?p=ffmpeg.git;a=tree;f=tests/ref/fate;h=3c25ec16d76bb1d0a0d67378212d5cee8aa6fd83;hb=HEAD
[15:18:40 CEST] <vibr> so for an input file from fate-suite, say SVA_Base_B.264, i want to get the corresponding crc file (seq-3341-7_seq-3342-5-24bit.flac in this example)
[15:19:44 CEST] <ubitux> full chain: encode to rawvideo codec, mux to rawvideo format, run crc on packet->data
[15:19:52 CEST] <ubitux> short path: crc(av_image_copy_to_buffer(frame))
[15:20:10 CEST] <ubitux> afaict.
[15:20:17 CEST] <ubitux> i let you find the necessary glue for that
[15:20:48 CEST] <vibr> hm, but muxing to framecrc format seemed to do most of it
[15:21:15 CEST] <vibr> i'd only have to decode first, i suppose
[15:22:32 CEST] <ubitux> mux to framecrc sorry
[15:23:21 CEST] <vibr> when muxing to framecrc i don't need to run crc on packet->data, do i?
[15:23:31 CEST] <ubitux> yes, i don't know what you want
[15:23:44 CEST] <ubitux> if you want the crc itself, you should just take the short path
[15:23:51 CEST] <ubitux> if you want the same output, use framecrc, so long path
[15:24:05 CEST] <vibr> yes, the later (long path)
[15:24:56 CEST] <vibr> i need the complete file as in test/ref/fate
[15:25:49 CEST] <vibr> ok, and i definitely have to encode to rawvideo after decoding, i can't leave out that step?
[15:28:46 CEST] <vibr> ubitux,thanks so far, i'll try and see
[15:30:14 CEST] <ubitux> encoding is mostly free
[15:30:22 CEST] <ubitux> it's calling av_image_copy_to_buffer to create a packet
[15:30:41 CEST] <ubitux> the muxer only takes packet, so you can emulate that part if you want, but that's probably saner to just call the encoder
[15:31:14 CEST] <vibr> ok, thanks a lot
[16:02:22 CEST] <prelude2004c> hey, anyone around this morning?
[17:51:34 CEST] <ubitux> anyone has any plan to support MediaCodec in ffmpeg? (the android stuff)
[17:59:39 CEST] <BtbN> Doesn't MediaCodec use ffmpeg? Or is that something else?
[18:01:08 CEST] <nevcairiel> how its implemented depends on the device
[18:01:14 CEST] <nevcairiel> but it also has OMX hardware codecs behind it
[18:01:57 CEST] <av500> depends
[18:01:59 CEST] <av500> some HW, some SW
[18:04:50 CEST] <prelude2004c> hey guys.. can anyone give me a hand.. i am not sure where the problem is... i am using pimd to get the multicast stream from one ethernet to another.. then i am using openvpn to create a basic tunnel with no encryption. The servers are talking and the client server can see the multicas stream. It seems when i pull one stream it starts ok but looses packets sometimes.. then i pull more streams ( program id's ) and it starts to colla
[18:05:03 CEST] <av500> pse
[18:06:16 CEST] <ubitux> prelude2004c: you message is truncated
[18:06:19 CEST] <ubitux> +r
[18:06:32 CEST] <ubitux> "starts to colla<truncated>"
[18:06:44 CEST] <cone-653> ffmpeg 03Alexis Ballier 07master:6fe2afaa7127: Adapt libavcodec/tableprint_vlc.h to the av_dlog -> ff_dlog change.
[18:12:19 CEST] <BtbN> And please don't ask the same question in both channels at the same time. Specialy when it's not entirely ffmpeg related and even less ffmpeg dev related.
[18:26:37 CEST] <cone-653> ffmpeg 03Clément BSsch 07master:11d29277d648: tests/subtitles: fix microdvd crlf in ref
[18:26:38 CEST] <cone-653> ffmpeg 03Clément BSsch 07master:2879a4ecb9c3: tests: make sure subtitles tests are run with a rawdiff
[18:35:53 CEST] <anshul_mahe> I am trying to make ref file for fate
[18:36:18 CEST] <anshul_mahe> I am having this error http://pastebin.com/hML6khFa
[18:36:28 CEST] <anshul_mahe> is this related to ffmpeg-devel
[18:36:55 CEST] <anshul_mahe> basically my sample have lavc version in it
[18:37:33 CEST] <anshul_mahe> while when I do it from fate it does not have lavc version, how to produce sample without lavc version
[18:49:48 CEST] <anshul_mahe> I got it thanks, cmd was in log file there
[18:53:10 CEST] <anshul_mahe> for making it work, I wrote 2 lines.http://pastebin.com/x49UZqLv
[18:53:54 CEST] <anshul_mahe> but first line look wrong to me
[18:54:22 CEST] <anshul_mahe> can any past expirence with writing fate test review it
[18:59:52 CEST] <anshul_mahe> ok I will find the meaing of 3 words, no worry. sorry for noise
[19:29:05 CEST] <Compn> lol
[19:29:52 CEST] <nevcairiel> that guy doesnt seem to be very smart
[19:37:09 CEST] <kierank> send a dmca
[20:14:06 CEST] <vibr> kierank, can i ask a question?
[20:17:59 CEST] <kierank> vibr: yes
[20:21:41 CEST] <vibr> avcodec_open2 returns something < 0 and i don't know where to look for the reason
[20:22:01 CEST] <vibr> i call avcodec to set up the encoder using rawvideo codec
[20:22:57 CEST] <vibr> do i have to initialise something else before calling avcodec_open2?
[20:23:32 CEST] <vibr> for the rawvideo encoder, that is?
[20:25:15 CEST] <vibr> greater picture: i set up a new formatcontext (for output) with avformat_alloc_output_context2, create a stream with avformat_new_stream, look for the codec (rawvideo) with avcodec_find_encoder
[20:25:25 CEST] <vibr> and so far everything's fine
[20:25:39 CEST] <vibr> then i call avcodec_open2, and it returns something < 0
[20:26:19 CEST] <vibr> so my guess is, that i have to initialise/setup something before calling call avcodec_open2, but i don't know what and how?
[20:29:57 CEST] <BtbN> What exactly does it return?
[20:34:22 CEST] <vibr> BtbN, it returns 1
[20:34:29 CEST] <vibr> erm, no
[20:37:54 CEST] <vibr> BtbN, it returns -22
[20:39:21 CEST] <BtbN> av_make_error_string can turn that into a usefull error message.
[20:40:13 CEST] <BtbN> Or av_strerror
[20:41:06 CEST] <BtbN> 22 is "Invalid Argument" iirc, so some argument you passed, or forgot to pass, is invalid.
[20:43:56 CEST] <BtbN> Did you look at some examples, like http://git.videolan.org/?p=ffmpeg.git;a=blob;f=doc/examples/decoding_encoding.c
[20:44:25 CEST] <BtbN> Make sure you call avcodec_register_all first
[20:50:44 CEST] <vibr> BtbN, yes,  i looked at the examples
[20:50:59 CEST] <vibr> the only one which is applicable for my case is transcoding.c
[20:51:19 CEST] <vibr> i called avcodec_register_all
[20:51:46 CEST] <vibr> so in transcoding.c there's some initialisation before avcodec_open2 is called
[20:52:22 CEST] <BtbN> yeah, selecting the codec and parameters
[20:52:35 CEST] <vibr> i mean these bits: enc_ctx->height = dec_ctx->height 
[20:52:37 CEST] <vibr> and so on
[20:52:39 CEST] <vibr> 5 lines
[20:53:24 CEST] <vibr> line 128-134
[20:54:12 CEST] <vibr> but the comment for this part says: "we transcode to same properties"
[20:54:22 CEST] <vibr> and i'm not sure when this applies
[20:54:37 CEST] <BtbN> What do you want to do?
[20:54:52 CEST] <vibr> in my case a transcode from some video codec to rawvideo
[20:54:56 CEST] <vibr> -a i
[20:55:23 CEST] <BtbN> so basicaly just decoding?
[20:55:48 CEST] <vibr> the chain is: decode, encode (as rawvideo), mux with framecrc
[20:55:49 CEST] <BtbN> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=doc/examples/transcoding.c#l49 that function does most of the stuff needed for that
[20:56:53 CEST] <vibr> line 149?? my source is 3 days old
[20:57:11 CEST] <BtbN> hm?
[20:57:32 CEST] <vibr> line 149 only contains a bracket ;)
[20:57:41 CEST] <BtbN> l49
[20:57:46 CEST] <BtbN> not 149
[20:57:59 CEST] <BtbN> just click the link, will take you right to the function
[20:58:35 CEST] <vibr> i see, the link
[20:58:47 CEST] <vibr> your link is open_input_file
[20:58:59 CEST] <vibr> that's not my problem, this part works fine
[20:59:26 CEST] <vibr> open_input_file initialises the decoder
[20:59:35 CEST] <vibr> i have problems with the encoder
[20:59:54 CEST] <vibr> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=doc/examples/transcoding.c#l87
[21:00:10 CEST] <vibr> <- corresponding part in transcoding.c
[21:00:49 CEST] <vibr> as i said, they set up these bits: http://git.videolan.org/?p=ffmpeg.git;a=blob;f=doc/examples/transcoding.c#l128
[21:00:57 CEST] <vibr> five lines from that
[21:01:05 CEST] <vibr> but i'm not sure it applies to my case
[21:01:29 CEST] <vibr> because of this comment http://git.videolan.org/?p=ffmpeg.git;a=blob;f=doc/examples/transcoding.c#l124
[21:16:22 CEST] <vibr> how can i find out what kind of initialisation has to be done for the rawvideo encoder before calling avcodec_open2?
[21:16:30 CEST] <vibr> i don't know where to look
[21:18:48 CEST] <vibr> kierank, different topic: i'm late for the outreachy application, as you can see
[21:19:14 CEST] <vibr> but i'm still trying to finish that task
[21:19:15 CEST] <Compn> stick around , devs will be back later :)
[21:19:31 CEST] <vibr> ok :)
[21:19:32 CEST] <Compn> or ask on the list
[21:20:45 CEST] <vibr> you mean ffmpeg-devel?
[21:31:08 CEST] <wm4> <vibr> how can i find out what kind of initialisation has to be done for the rawvideo encoder before calling avcodec_open2?
[21:31:15 CEST] <wm4> hm, at least width, height, and pixel format
[21:31:24 CEST] <wm4> pixel format or codec tag
[21:31:33 CEST] <wm4> but I don't remember exactly
[21:35:29 CEST] <vibr> wm4, thanks
[21:35:46 CEST] <vibr> and how do i know the correct settings?
[21:36:08 CEST] <vibr> i the example code here
[21:36:27 CEST] <vibr> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=doc/examples/transcoding.c#l128
[21:37:21 CEST] <vibr> they just re-use the values from input-formatcontext->stream->codeccontext
[21:37:35 CEST] <vibr> but i'm not sure when i can re-use these values
[22:09:41 CEST] <wm4> vibr: it generally depends on the codec
[22:09:49 CEST] <wm4> normally, there are not a lot of fields
[22:10:27 CEST] <wm4> here's what mpv does (it manually copies fields from the demuxer codec context to the decoder codec context): http://git.srsfckn.biz/mpv/tree/common/av_common.c#n48
[22:13:20 CEST] <vibr> wm4, thanks, i'll check this tomorrow - too tired now :)
[00:00:00 CEST] --- Tue Apr 28 2015