[Ffmpeg-devel-irc] ffmpeg-devel.log.20131103

burek burek021 at gmail.com
Mon Nov 4 02:05:02 CET 2013


[00:42] <cone-155> ffmpeg.git 03James Zern 07master:e87043bb1fbc: MAINTAINERS: add myself for libvpx*
[00:43] <cone-155> ffmpeg.git 03James Zern 07master:517afd72c677: libvpxenc: add VP9 options
[01:01] <wm4> does af_volume not clamp results to [-1,1] range?
[01:01] <wm4> it doesn't seem to do so for float/double
[02:30] <cone-155> ffmpeg.git 03Timothy Gu 07master:624672fcce84: avutil/adler32: add Doxy group
[02:30] <cone-155> ffmpeg.git 03Timothy Gu 07master:50f52a5498ff: avutil/xtea: add Doxy @file and group
[02:30] <cone-155> ffmpeg.git 03James Zern 07master:23c03ac91eae: libvpxenc: allow qmax of 0
[02:30] <cone-155> ffmpeg.git 03Michael Niedermayer 07master:4fb1221e66db: h264: reduce whitespace differences to libav
[02:30] <cone-155> ffmpeg.git 03Michael Niedermayer 07master:6e87dfb58945: mpegvideo: reduce whitespace differences to libav
[02:30] <cone-155> ffmpeg.git 03Michael Niedermayer 07master:63a37d0e1ec5: avcodec/hevc_sei: check active_seq_parameter_set_id validity
[02:30] <cone-155> ffmpeg.git 03Michael Niedermayer 07master:8a701ef7ddbb: avcodec/hevc_sei: Fix null dereference in decode_pic_timing()
[10:04] <cbsrobot> ubitux: you could try the gdbinit from http://reverse.put.as/gdbinit/ to prettyprint the registers & not sure it works though
[10:49] <j-b> 'morning
[10:54] <michaelni> morning
[11:09] <saste> michaelni, still waiting on lavc: drop deprecated audio_convert API at the next major bump
[11:20] <michaelni> saste, reply written
[12:01] <cone-408> ffmpeg.git 03Rudolf Polzer 07master:34a0ac41a880: vf_aspect: support variables in evaluating the aspect ratios
[12:01] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:a75bf4e5c186: Merge commit '34a0ac41a880039c7fb4b3abce10d8e2ce48fac7'
[12:34] <cone-408> ffmpeg.git 03Rudolf Polzer 07master:3a14b160bd44: vf_scale: make the dar variable return the DAR.
[12:34] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:7edc9240a8f4: Merge commit '3a14b160bd442794c6a16a6a883adb547550e847'
[12:44] <ubitux> cbsrobot: will try, thx
[12:48] <cone-408> ffmpeg.git 03Vittorio Giovara 07master:5858a67f135a: attributes: disable flatten on llvm/clang
[12:48] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:08e7644b6853: Merge commit '5858a67f135a7395c548482c73cf3d39bcdd3022'
[12:53] <cone-408> ffmpeg.git 03Vittorio Giovara 07master:c0bba95c2363: wtv: fix variable sign in format
[12:53] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:b6413cdc02f3: Merge commit 'c0bba95c2363641d3297b3852b2ece1474cda295'
[13:07] <cone-408> ffmpeg.git 03Vittorio Giovara 07master:3ea5f64ffff0: ansi: fix possible use of uninitialized variables
[13:07] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:b568eeba7524: Merge commit '3ea5f64ffff0a51f62922efd2e2bc231b13b2179'
[13:07] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:cf49d5907aae: avcodec/ansi: set w/h correctly at the top
[13:14] <saste> we have several LIBAVCODEC_VERSION_MAJOR < 55 in libavcodec/version.h
[13:15] <saste> why the dead code was not removed?
[13:17] <cone-408> ffmpeg.git 03Stefano Sabatini 07master:5a6e6dcaa4d2: lavc: drop deprecated audio_convert API at the next major bump
[13:17] <cone-408> ffmpeg.git 03Stefano Sabatini 07master:90c329cc251f: lavc/version.h: remove empty line
[13:33] <ubitux> ok i have a 8x8 almost working
[13:33] <ubitux> it looks ok but fails on fate :(
[13:35] <nevcairiel> saste: if its currently disabled, just submit a patch to clear it out?
[13:37] <iive> ubitux: are you working on sse optimizations for vp9?
[13:37] <ubitux> ssse3
[13:37] <ubitux> yes
[13:46] <cone-408> ffmpeg.git 03Vittorio Giovara 07master:d7639785832c: rtmpproto: Reorder conditions to help dead code elimination
[13:46] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:91a697576169: Merge remote-tracking branch 'qatar/master'
[13:51] <ubitux> http://b.pkh.me/8x8ref.pnghttp://b.pkh.me/8x8out.png almost there..
[13:53] <michaelni> hi superdump
[13:53] <michaelni> do you have time to take a quick look at 2 bugs?
[13:53] <michaelni> actually iam nit even sure they are valid/real
[13:53] <michaelni> https://trac.ffmpeg.org/ticket/1848 and https://trac.ffmpeg.org/ticket/1849
[13:54] <cbsrobot> ubitux: did it help ?
[13:55] <ubitux> cbsrobot: didn't look yet
[13:55] <ubitux> but i'll try
[13:56] <cbsrobot> you get all additional commands with "help user"
[13:58] <ubitux> ok
[14:03] <cone-408> ffmpeg.git 03Lukasz Marek 07master:6ac9afd16e38: lavd/alsa: fix timestamp calculation
[14:03] <cone-408> ffmpeg.git 03Lukasz Marek 07master:e56d1a120324: lavd/alsa: add stream validation
[14:03] <cone-408> ffmpeg.git 03Nicolas George 07master:eeb975f5cda3: lavf/wavenc: check for a single stream.
[14:03] <cone-408> ffmpeg.git 03Nicolas George 07master:863fb11f63f7: lavd/lavfi: support unknown channel layouts.
[14:03] <cone-408> ffmpeg.git 03Nicolas George 07master:f775eb3fb4c7: lavfi/avfiltergraph: suggest a solution when format selection fails.
[14:03] <cone-408> ffmpeg.git 03Nicolas George 07master:d300f5f6f570: lavfi/avfiltergraph: do not reduce incompatible lists.
[14:03] <cone-408> ffmpeg.git 03Nicolas George 07master:6e2473edfda2: lavfi: parsing helper for unknown channel layouts.
[14:03] <cone-408> ffmpeg.git 03Nicolas George 07master:4a640a6ac890: lswr: fix assert failure on unknown layouts.
[14:03] <cone-408> ffmpeg.git 03Nicolas George 07master:4e9adc9b7363: lavfi/af_pan: support unknown layouts on output.
[14:03] <cone-408> ffmpeg.git 03Nicolas George 07master:7b0a587393e0: lavfi/af_pan: support unknown layouts on input.
[14:03] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:2ed562d5fa05: Merge remote-tracking branch 'cigaes/master'
[14:35] <michaelni> BBB, ubitux vp9 is missing in the MAINTAINERs file
[14:36] <BBB> uhm, maybe add him and me?
[14:36] <ubitux> or just BBB ;)
[14:51] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:73e7d8f5f028: MAINTAINERS: add vp9 maintainers
[14:51] <ubitux> :(
[14:52] Action: ubitux just got cursed of maintainership
[14:53] <Daemon404> can i has the codez?
[14:53] Action: Daemon404 emails ubitux from india
[14:53] <ubitux> :)
[14:59] <michaelni> ubitux, feel free to remove yourself if you dont want to maintain it
[14:59] <ubitux> i was half joking :)
[15:01] <michaelni> :)
[16:50] <wm4> hm avio opens a certain unseekable http url as seekable, with size==2000000000
[16:50] <wm4> this is the url: http://8283.live.streamtheworld.com:80/CBC_R1_VCR_H_SC
[16:51] <wm4> this is very bad, because it's an aac stream, and lavf tries to seek to the "end" of the file to read the id3
[16:51] <nevcairiel> it gets the seekable info from the http headers, blame the server for sending a range header then
[16:51] <nevcairiel> my guess, anyway
[16:55] <wm4> where is the AVIOContext seekable flag set? not in http.c...
[16:56] <nevcairiel> i guess seekable is never set, but it has its own seek function which checks this on the is_streamed flag
[16:56] <nevcairiel> i suppose it might be smart if this goes through to seekable
[16:58] <Compn> isnt there an option to ignore seeking ?
[16:58] <Compn> -noidx or something
[16:58] <wm4> and of course useful things like dumping headers is hidden with av_dlog...
[16:58] <wm4> Compn: doesn't help shit
[16:59] <wm4> you expect the user to guess lavf is taking so long because it ties to read an id3 at the "end" of a web radio stream?
[17:00] <Compn> lavf likes 5mb probe anyways
[17:00] <wm4> yeah I'm already preventing that for aac via a hack
[17:00] <Compn> lol
[17:00] <wm4> so horrible
[17:00] <Compn> back to mime-types then? 
[17:00] <wm4> it uses the mime type for the hack
[17:00] <Compn> yes!
[17:01] <Compn> btw this works pretty quickly
[17:01] <Compn> mplayer ffmpeg://http://8283.live.streamtheworld.com:80/CBC_R1_VCR_H_SC -nocache -v -v -v -lavfdopts format=aac
[17:02] <Compn> add -demuxer lavf
[17:02] <Compn> any by pretty quickly i mean instantly
[17:02] <Compn> at least with my old mplayer
[17:02] <Compn> dont need -v -v -v either
[17:02] <wm4> here it hangs when printing "Checking for Nullsoft Streaming Video"
[17:02] <Compn> mplayer ffmpeg://http://8283.live.streamtheworld.com:80/CBC_R1_VCR_H_SC -nocache -lavfdopts format=aac -demuxer lavf
[17:02] <wm4> MPlayer 1.1-4.8 (C) 2000-2012 MPlayer Team
[17:02] <Compn> you forgot -demuxer lavf
[17:03] <wm4> mp_seek(0x930a398, 1999999872, 0)
[17:03] <wm4> it does the same crap
[17:03] <Compn> did you get the -nocache ?
[17:03] <wm4> yes
[17:03] <Compn> hmm
[17:03] <Compn> maybe my mplayer is too old
[17:03] <wm4> or mine is?
[17:03] <wm4> see 2012 copyright
[17:03] <Compn> could be
[17:03] <Compn> well
[17:04] <Compn> MPlayer sherpya-r36349+gb24f15c-4.6 (C) 2000-2013 MPlayer Team
[17:04] <Compn> MPlayer Redxii-SVN-r35679-4.6.3 (C) 2000-2012 MPlayer Team
[17:04] <Compn> so there :P
[17:04] <Compn> works in both of those for me
[17:04] <Compn> ffplay http://8283.live.streamtheworld.com:80/CBC_R1_VCR_H_SC also works instantly
[17:05] <Compn> ffplay version N-52523-g0fb64da Copyright (c) 2003-2013 the FFmpeg developers
[17:05] <wm4> what is instant?
[17:05] <wm4> it takes at least 10 seconds here
[17:05] <Compn> 1 second, tops
[17:06] <Compn> maybe 2-3 seconds  :P
[17:06] <Compn> hehe
[17:06] <Compn> but not 10
[17:08] <Compn> do you have ipv6 stuff ?
[17:08] <Compn> that always caused trouble in mplayer for me
[17:08] <Compn> trying to dns ipv6 , so lets stop trying ipv4...
[17:08] <Compn> -prefer-ipv4 in my config i think
[17:08] <wm4> no ipv6 here
[17:08] <wm4> anyway, the server indeed sends crap
[17:08] <wm4> content-range: bytes 0-1999999999/2000000000
[17:08] <wm4> yeah no
[17:08] <Compn> lol
[17:09] <nevcairiel> i told you :)
[17:10] <nevcairiel> otherwise it would have no idea about the file size and where to seek to
[17:10] <wm4> so what do, shall I add a funny hack?
[17:10] <wm4> there's already one:
[17:10] <wm4> if (s->seekable == -1 && (!s->is_akamai || s->filesize != 2147483647))
[17:10] <wm4>                 h->is_streamed = 0; /* we _can_ in fact seek */
[17:11] <Compn> "is_akamai"
[17:11] <wm4>         } else if (!av_strcasecmp (tag, "Server") && !av_strcasecmp (p, "AkamaiGHost")) {
[17:11] <wm4>             s->is_akamai = 1;
[17:11] <wm4> and this particular stream sends "Server: MediaGateway 3.5.2-001"
[17:11] <Compn> why i dont like id3 nonsense...
[17:12] <wm4> maybe http.c should just export the server so applications can decide whether they do ridiculous server specific hacks
[17:13] <nevcairiel> too bad protocols dont have access to the metadata dicts in the format context
[17:13] <wm4> you could add one to AVIOContext
[17:13] <Compn> so a user has to know what hacks to add because they get a server name ?
[17:13] <Compn> thats what you are saying here
[17:14] <wm4> yes, because it would be a ridiculous hack
[17:14] <wm4> just like the akamai one btw.
[17:14] <Compn> akamai is ridiculous :D
[17:14] <nevcairiel> open a ticket, i know a dev who likes contributing suck hacks
[17:14] <nevcairiel> such*
[17:14] <wm4> "'Expires: Thu, 01 Jan 1970 00:00:00 GMT"
[17:14] <wm4> lol this server
[17:15] <nevcairiel> personally i would just tell the user that complained to go complain at the host
[17:15] <Compn> lets seek back to 1970
[17:15] <Compn> does it work in vlc ?
[17:15] <wm4> there's no user
[17:15] <wm4> I just care about it because mplayer can handle it fine
[17:15] <wm4> (with its native http)
[17:15] <Compn> so which software is causing problem with it ?
[17:15] <nevcairiel> because it doesnt read id3? :d
[17:15] <wm4> maybe merely because it doesn't read id3 tags from aac streams
[17:15] <Compn> oh mplayer with lavf
[17:16] <nevcairiel> i used to get compared to ffdshow all the time, but ffdshow can do this, and it can do that, but i simply refused to implement all these ugly hacks to support broken things, and it helped keep me sane :p
[17:17] <Compn> wm4 : btw if you do mplayer ffmpeg://http://jasonsloan.bandcamp.com/album/live-on-sadayatana -demuxer aac -nocache it works 
[17:17] <Compn> how you know its id3 problem and not other probe ?
[17:18] <Compn> nevcairiel : did they start 'rewriting' ffdshow ?
[17:18] <Compn> yet
[17:18] <Compn> ffdshowtryouts was going to do it
[17:18] <nevcairiel> thats usually what people refer to when they say ffdshow
[17:18] <nevcairiel> the original died in 2006 or something
[17:18] <Compn> yeah :P
[17:18] <wm4> Compn: because I'm watching what it does
[17:19] <nevcairiel> but it was never "rewritten", just more hacks on hacks
[17:19] <Compn> wm4 : because the nullsoft stream probe isnt the last demuxer before lavf ?
[17:19] <Compn> actuall nsv is right after lavf, which means mplayer skipped over -demuxer 30 (aac) :\
[17:20] <Compn> unless im confused about demuxer order lol
[17:20] <nevcairiel> does it actually stop when it probes a demuxer, i thought it continues to see if one has a higher score
[17:20] <wm4> Compn: I was testing this with my mplayer fork
[17:20] <Compn> er aac is 39 , nsv is 36
[17:20] <Compn> wm4 : i'm saying if you disable nsv demuxer , i think it will work :P
[17:21] <Compn> it gets stuck there because nsv demuxer probe broken
[17:21] <wm4> so on my fork it actually reads only 2 KB
[17:21] <wm4> and then lavf seeks to the "end" of the stream
[17:21] <wm4> so doing all these nonsense reconnects makes it awfully slow
[17:21] <nevcairiel> 2kb is the usual probe size, iirc
[17:21] <Compn> nevcairiel : dont think so, why probe 40 demuxers if demuxer 2 works ?
[17:21] <nevcairiel> must be enough to figure out its aac
[17:22] <wm4> nevcairiel: aac can probe much more in some situations
[17:22] <nevcairiel> Compn: some demuxers only respond with a low score because of false positives, especially such raw-format demuxers
[17:22] <nevcairiel> the whole score system would be useless if it stopped anyway
[17:23] <Compn> wm4 : with enough -v , mplayer http demuxer outputs http header :P
[17:23] <Compn> Fields:
[17:23] <Compn>  0 - Expires: Thu, 01 Dec 2003 16:00:00 GMT
[17:24] <ubitux> 8x8 working \o/
[17:24] <Compn> nevcairiel : well then mplayer http must be detecting mime type to skip probing 
[17:25] <Compn> nevcairiel : and mplayer doesnt report all demuxer probes when playing files, so maybe its just all hidden ?
[17:25] <Compn> well it also does fileextension cheats :D
[17:25] <Compn> so many hacks
[17:25] <nevcairiel> mplayer is a pile of hacks, i'm not going to start guessing how it works
[17:25] <Compn> lol
[17:26] <wm4> indeed, mplayer is full of dumb crap
[17:26] <wm4> which makes it amazing that some things work so much better than in ffmpeg
[17:26] <durandal_1707> michaelni: you added stuff into vf_aspect that are not needed
[17:27] <durandal_1707> saste: ^
[17:28] <Compn> wm4 : anyways if you want to fix this particular problem, have ffmpeg:// pass the mime type to mplayer, and it will do the rest
[17:28] <Compn> no need for hack.
[17:28] <wm4> man the worst is that the mediagate server pretends seeking was a success
[17:29] <Compn> see, when a server sends so many broken things, it sounds like they stole some gpl code and didnt know what to do with it
[17:29] <Compn> so strings their server and see what gpl project they are breaking :D
[17:32] <Compn> is it weird i cant find 'media gateway' server ?
[17:35] <Compn> anyone want to rescue the bayer rgb patches so we can get r3d camera files working ?
[17:36] <saste> durandal_1707, like what?
[17:36] <ubitux> ok 8x8 ~6x faster
[17:36] <ubitux> maybe i can do better with 2x2
[17:42] <wm4> is there a mechanism that tells lavf that seeking doesn't work, but that it can seek forward by skipping large amounts of data by reading over it?
[17:42] <wm4> you'd want that for a local file (maybe), but definitely not with a slow source
[17:44] <cone-408> ffmpeg.git 03Paul B Mahol 07master:3d9fc5c6ad31: avfilter/vf_aspect: remove dupe constants setup
[17:46] <Compn> sint there an is_slow flag ?
[17:47] <wm4> no
[17:47] <wm4> maybe what I want is AVIO_SEEKABLE_NORMAL
[17:48] <wm4> the way it's used seems to be useless though
[17:49] <wm4> in fact that's the only flag that can be set on aviocontext->seekable
[17:59] <{V}> ubitux, in vp9?
[17:59] <ubitux> yes
[17:59] <{V}> congrats
[17:59] <ubitux> i think i can do better
[18:18] <cone-408> ffmpeg.git 03Stefano Sabatini 07master:d3397c3e829b: lavc: remove dead FF_API_ALLOC_CONTEXT API
[18:18] <cone-408> ffmpeg.git 03Stefano Sabatini 07master:a68956042818: lavc: remove dead FF_API_AVCODEC_OPEN API
[18:18] <cone-408> ffmpeg.git 03Stefano Sabatini 07master:cfe7185a92da: lavc: remove dead FF_API_OLD_TIMECODE API
[18:19] <cone-408> ffmpeg.git 03Stefano Sabatini 07master:535d58959dec: ffmpeg: add verbose consistency checks in case of filtering inconsistent options
[19:50] <cone-408> ffmpeg.git 03Mikulas Patocka 07master:074155360d28: avcodec/x86/hpeldsp: fix crash on AMD K6-3+
[21:10] <ubitux> cbsrobot: it's buggy with my gdb unfortunately
[21:14] <ubitux> it would be nice if it was supporting the xmm and mmx registers
[21:15] <ubitux> ...as well as coloring registers only when they change from previously
[21:54] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:8b73a3f6f659: avformat/utils: never decrease has_b_frames in compute_pkt_fields()
[22:40] <ubitux> 7776 decicycles in idct_idct_8x8_add_c, 4192821 runs, 1483 skips
[22:40] <ubitux> 1120 decicycles in ff_vp9_idct_idct_8x8_add_ssse3, 4193860 runs, 444 skips
[22:40] <nevcairiel> how does overall decode speed benefit from all the idcts?
[22:41] <ubitux> didn't bench yet
[22:41] <ubitux> i'm assuming it will start to be relevant when the iadst variants will be done as well
[22:41] <ubitux> i think i can remove 2 mov with the current code, can be very slightly faster
[22:43] <cone-408> ffmpeg.git 03Michael Niedermayer 07master:336a1902d6be: avcodec/ffv1enc: Check high bpp RGB against coder type too
[22:49] <ubitux> nevcairiel: http://pastie.org/8453081
[22:49] <nevcairiel> just for 8x8?
[22:49] <ubitux> first 3: current, last 3 4x4+8x8
[22:49] <nevcairiel> ok
[22:50] <ubitux> 40 seconds decode / 1080p
[23:11] <cone-408> ffmpeg.git 03wm4 07master:3220a894f794: http: add hack to make streams served by MediaGateway not seekable
[23:11] <ubitux> https://github.com/ubitux/FFmpeg/compare/vp9-asm
[23:11] <ubitux> BBB ^
[23:12] <wm4> eh that was quick
[23:12] <ubitux> BBB: i'll try to honor what you said in previous review and resubmit
[23:26] <wm4> so what's the deal with AVPicture?
[23:28] <juanmabc> i'm gonna need advice to use 'your' api ;) , the cause, my project current tarball, https://openalextensions.googlecode.com/files/ale-0.4.18.tar.bz2
[23:29] <juanmabc> here's the meat decoding module: http://ur1.ca/fzcc8 -> http://paste.fedoraproject.org/51352/51765213
[23:30] <juanmabc> before 1.0 it worked (worked?, sounds so bad that english) , now it runs bad as you can see by a fast ./configure; make; cd demos/; ./alestream somefile.mp3 (or whatever ffmpeg decodeable file)
[23:32] <juanmabc> i'm idle, and i'd be glad to promote this code to a sample decoding implementation of ffmpeg that you can/could refer to anyone :)
[23:36] <juanmabc> this could promote ffmpeg to games, btw
[23:36] <juanmabc> the trend is to use such specific libs like libvorbis, libogg, ... but you give a lot of power that ffmpeg has
[23:37] <BBB> ubitux: \o/
[23:37] <ubitux> actually i'm too lazy for the xmm in 4x4 now :)
[23:37] <ubitux> i'd better have the iadst combinations done
[23:39] <ubitux> mmh next week end is 3-days one again in my country
[23:39] <ubitux> maybe i'll be able to get it done next week then
[23:41] <BBB> what is topleftcorner?
[23:41] <BBB> is that a 2x2 sub8x8?
[23:41] <BBB> it seems it is, yes
[23:41] <ubitux> not really 2x2
[23:41] <ubitux> DC, right and bottom
[23:41] <BBB> does it require that 1,1==0?
[23:41] <ubitux> it's not a square like in 4x4 in the zigzag
[23:41] <BBB> right I know
[23:42] <ubitux> BBB: require no
[23:42] <BBB> but you (I) could change that to be subblock driven
[23:42] <ubitux> but it will be always 0
[23:42] <BBB> i.e. not just go by eob, but go by coeffzeromasking
[23:42] <BBB> and then we could call this if anything outside 2x2 is 0
[23:42] <BBB> that's why I'm asking
[23:42] <ubitux> i don't get it
[23:42] <wm4> <juanmabc> the trend is to use such specific libs like libvorbis, libogg, ... but you give a lot of power that ffmpeg has <- actually, using ffmpeg in games is the pure horror
[23:43] <ubitux> ah you mean another flag for that 2x2 area?
[23:43] <wm4> juanmabc: I know because 1. I've done it myself, and 2. have others seen doing it
[23:43] <juanmabc> wm4: reasons?
[23:43] <juanmabc> is good for video on textures too btw
[23:43] <wm4> juanmabc: complex, low level, ever changing API
[23:43] <juanmabc> ever changing API, i'm there bro
[23:44] <juanmabc> i think at least i'm gonna freeze with an internal copy, and update when i want
[23:44] <wm4> lol
[23:44] <juanmabc> hehe, it's a real valid solution
[23:44] <wm4> having a copy of libvorbis is really much better than having a copy of ffmpeg
[23:45] <juanmabc> my main goal is power, and multiformat is much superior
[23:45] <juanmabc> plus i'd reuse that for video on texture as i said :)
[23:45] <juanmabc> ala Blink Video
[23:45] <juanmabc> ;)
[23:47] <juanmabc> i did it already (on an old api ;)) https://www.youtube.com/watch?v=BHPzjpojlz0
[23:48] <juanmabc> ogre3d people do it, blender people do it, crystal space people do it, mine is just that, not tied to an engine
[23:48] <juanmabc> so you glue that lib to any engine, and is more reusable
[23:50] <juanmabc> plus newcomers loves my aleBufferData(buffer, "audio.ogg"); :)
[23:52] <juanmabc> about the code, i'd keep around for more detailed help :)
[23:52] <wm4> so you're providing the high level API ffmpeg really should provide
[23:53] <wm4> like decoding a file into a memory buffer
[23:53] <juanmabc> nope, i do that plus pass that memory to openal
[23:53] <wm4> (fair enough)
[23:53] <wm4> yeah, getting support for openal is only a small step further
[23:55] <juanmabc> that's why i said that can be referred as a sample usage implementatino :)
[23:55] <juanmabc> main stuff is in aleStreamBufferChunk
[23:55] <juanmabc> rest is bolierplate
[23:56] <juanmabc> i keep researching, but api is so new so yet not so spread
[23:57] <juanmabc> i mainly avcodec_decode_audio4 and pass it
[23:57] <juanmabc> i think i fail at the math at 'passing it', samples or channels
[23:57] <juanmabc> again 'it worked for good pre 1.0'
[23:59] <juanmabc> without Convert i get the smurfs
[23:59] <juanmabc> with convert i get left well, right is loud
[23:59] <juanmabc> :S
[00:00] --- Mon Nov  4 2013


More information about the Ffmpeg-devel-irc mailing list