[FFmpeg-devel-irc] IRC log for 2010-05-20

[irc at mansr.com]

[00:53:21] * Terminating due to: TERM
[00:53:32] * /join #ffmpeg-devel ...
[00:53:34] *** TOPIC: Welcome to the FFmpeg development channel. That is, development of FFmpeg, not using FFmpeg, nor libav*. | Users should redirect their questions to #ffmpeg | FFmpeg 0.5.1 has been released! | this chat is now publicly logged.
[00:53:34] *** TOPICINFO: superdump!~rob at unaffiliated/superdump, 1267604523
[02:55:04] <ramiro> libvpx's configure is strangely similar to ffmpeg's...
[02:56:00] <astrange> learning from the best?
[02:56:54] <ramiro> =)
[02:57:37] <Yuvi> bcoudurier: I'm getting oggz-validate errors (timestamp out of order) with theora+flac, looking into it now
[02:57:50] <Yuvi> also eos sometimes not being set
[02:59:10] <peloverde> ramiro: ffmpeg's configure/makefile has a valid make install
[03:00:00] <ramiro> and it's great for cross-compilation
[03:00:24] <Yuvi> and it detects what arch you're on without additional arguments
[03:02:22] <ramiro> many hunks are copy&pasted. did mru have anything to do with this?
[03:02:23] <Compn> think of it this way
[03:02:37] <Compn> they already distribute and maintain some ffmpeg in chrome
[03:02:48] <Compn> would they chose autotools to use with vp8 ?
[03:03:06] <Compn> actually, i wonder if they use ffmpeg makefile in on2? nothing to do with google ?
[03:03:14] <Compn> on2 distributes mencoder hmm
[03:03:34] <Compn> have to diff it against svn makefile to see how old it is :D
[03:14:02] <bcoudurier> Yuvi, shit it's broken then
[03:21:22] <peloverde> I feel like this webm situation gives me even less motivation to merge psdec into trunk
[03:24:27] <bcoudurier> nah don't worry
[03:24:34] <bcoudurier> Yuvi, something is wrong with the last_pts check
[03:24:45] <bcoudurier> what if you remove it ?
[03:26:17] <ramiro> peloverde: heh, now I get the make install is configuring thing...
[03:26:41] <ramiro> imo make install shouldn't even try to build anything
[03:27:46] <astrange> you don't like typing 'make install clean' to install a bsd port?
[03:28:06] <peloverde> The big PS demotivator was losing funding for the OSS version, I mainly wrote it to sell to third parties (which I did successfully), webm is just icing on the cake
[03:32:19] <Yuvi> bcoudurier: yep, that fixes the ordering
[03:33:12] <Yuvi> looks like the eos too
[03:33:57] <Yuvi> peloverde: was the funding from google / have you talked to them about potential money?
[03:35:10] <peloverde> funding was osp.fi
[03:37:43] <peloverde> google knows who I am and what I do, I even pushed SBR as hard as possible to try to meet their timelines (as well as an earlier projected 0.6), but they largely seem disinterested
[03:40:08] <bcoudurier> Yuvi, got it
[03:43:21] <peloverde> I also sent them some thoughts on the general multimedia topics in December about multimedia including funding OSS work and hiring some MPEG thought leaders and got a form type letter back
[03:44:21] <Compn> peloverde : did you send a message to the head of google code ?
[03:44:26] <Compn> maybe he can get someone to work on sbr
[03:44:44] <peloverde> Compn: SBR is merged :)
[03:44:59] <Compn> er not sbr, but whatever needs doing :P
[03:45:03] <peloverde> PS
[03:45:11] <peloverde> (parametric stereo)
[03:45:13] <astrange> ps = he-aac v2
[03:45:18] <Compn> that then
[03:45:21] <Compn> did you email the guy?
[03:45:27] <astrange> it's used in broadcast but not on the web(?) so google might not care as much?
[03:45:32] <peloverde> you mean diBona?
[03:45:37] <Compn> yes dibona i think
[03:46:29] <peloverde> I talked to chrome people
[03:46:45] <Compn> forget those guys
[03:46:48] <peloverde> but just thinking about dibona makes me miss "Geeks in Space"
[03:47:25] <peloverde> that was a great podcast or "internet radio show" as it was known at the time
[03:47:32] <Compn> send him a mail :P
[03:47:38] <Compn> cant hurt anyhow
[03:47:58] <Compn> or post a note on ffmpeg.org saying it would be nice if someone would try it
[03:48:06] <peloverde> I suppose not :)
[03:48:14] <Compn> or continue to bitch on irc, see if anything comes of that
[03:48:37] <peloverde> I have actually gotten a few minor jobs on from ffmpeg.org
[03:48:50] <peloverde> there are a few googlers idling in here :)
[03:49:17] <Compn> are there ?
[03:49:19] <Compn> hmmm
[03:50:18] <peloverde> _skal_zZzZz_ scherkus
[03:51:00] <bcoudurier> Yuvi, please try updated patch
[03:51:09] <peloverde> off the top of me head, it wouldn't surprise me if there were a few more lurkers
[03:51:28] <Compn> this channel grew a bit over the past few months
[03:53:00] <bcoudurier> hiring some MPEG thought leaders ?
[03:53:33] <bcoudurier> what do you mean peloverde ?
[03:53:40] <Compn> bcoudurier : hyc is getting impatient about his ffserver patches , hehe
[03:53:47] <bcoudurier> ah right
[03:53:48] <peloverde> hiring some people from jvt-experts and mpeg-audio-call
[03:53:56] <bcoudurier> marting seems to be handling them well, though
[03:53:59] <bcoudurier> martin sorry
[03:54:02] <bcoudurier> ah
[03:54:05] <bcoudurier> why would they ?
[03:54:10] <peloverde> like a gary sullivan or schylur quckenbush
[03:54:16] <Compn> bcoudurier : well martin was waiting for an OK on the patches :)
[03:54:19] <bcoudurier> gary is microsoft
[03:54:33] <peloverde> gary is microsoft, he wasn't always microsoft
[03:54:40] <bcoudurier> and everybody knows microsoft has better 401k and health package
[03:55:14] <peloverde> ralph sperschneider or andreas scheneider from mepg audio are FhG and Dolby
[03:55:24] <peloverde> (respectively)
[03:55:26] <bcoudurier> google is for nerds < 30 years, sorry skal :)
[03:55:53] <bcoudurier> these are good company for benefits
[03:55:58] <bcoudurier> companies
[03:56:05] <bcoudurier> I guess
[03:56:35] <peloverde> Max Neuendorf from FhG would also be a potential good pick
[03:57:23] <bcoudurier> I'm not sure google cares about standardization though
[03:57:44] <bcoudurier> otherwise they wouldn't have chosen very "unstandard" codecs
[03:58:06] <peloverde> I think to a degree they need to be aprised of what is going on there since their major competitors are (M$ and apple)
[03:58:24] <bcoudurier> I agree, but that is not SMPTE nor ISO
[03:58:28] <peloverde> and regardless I think these people I named have good ideas that can be applied elsewhere
[03:58:46] <bcoudurier> of course
[03:58:55] <bcoudurier> but OSS is not their play fields
[03:59:30] <peloverde> only because mpeg and/or proprietary pays more
[04:00:23] <bcoudurier> maybe
[04:00:52] <peloverde> also what I know is when CT was independent and what I wanted aligned with what
[04:00:56] <bcoudurier> Yuvi, are you there ?
[04:01:09] <peloverde> CT wanted Andreas Schneider knew how to get things done at MPEG
[04:01:09] <bcoudurier> CT ?
[04:01:16] <peloverde> Coding Technologies
[04:01:20] <bcoudurier> oh
[04:01:31] <peloverde> now known as Dolby International
[04:02:03] <bcoudurier> dolby care about standards because of equipment
[04:02:52] <bcoudurier> anyway home time
[04:02:57] <peloverde> I've interviewed at Dolby more times than I care to admit, their business model is to make very good production tools and license the formats downward
[04:03:23] <peloverde> CT was even more dependent on licensing
[04:03:42] <Yuvi> bcoudurier: yeah
[04:03:53] <Yuvi> the new patch on the ML fixes the issues, looking at it more now
[05:35:45] <Yuvi> youtube's webm muxer is based off of lavf :)
[05:36:01] <astrange> isn't it still mencoder?
[05:36:19] <Yuvi> probably
[05:38:10] <Yuvi> but iirc they used gpac for mp4 for instance
[05:49:47] <astrange> yeah, vfr is still broken
[05:56:28] <astrange> http://astrange.ithinksw.net/ffmpeg/hiromoving_vfr.mp4 -> http://www.youtube.com/watch?v=TlxVrbkq5kw is all wrong ( <-- _skal_zZzZz_ )
[05:58:25] <pJok> goda morgonar kshishkov :)
[05:59:08] <av500> Guten Morgen Herr kshishkov
[05:59:24] <pJok> mornings av500 :)
[05:59:49] <pJok> i wonder when the patent trolls will star to appear
[05:59:52] <pJok> start*
[06:01:29] <av500> not for a while
[06:01:56] <av500> trolling early means killing your prey at a very yound age
[06:02:01] <pJok> true
[06:02:04] <av500> you want the sow to fatten
[06:02:34] <Tjoppen> I already see a lot of people touting it as non-encumbered. this worries me
[06:02:41] <av500> Tjoppen: dont be
[06:02:54] <av500> ppl are just repeating press releases
[06:03:52] <pJok> are people still touting vp8 as the best codec since sliced bread?
[06:04:43] <av500> gg is touting it as "free"
[06:35:40] <wbs> bcoudurier: regarding the ffserver patches, yes, please have a look at them, I only applied the really obvious ones... the rest of them requires quite a bit of familiarity of the intended architecture and stuff
[06:35:50] <KotH> moin people
[06:36:11] <wbs> didn't have time to dive in that deep on the matter just to approve/disapprove patches on something I don't maintain nor use :-)
[06:36:25] <wbs> but the latest extradata handling patches are at least much less hacky than the initial ones :-)
[06:40:04] <av500> KotH: moin, btw is gg street view in .ch as controversial as in .de?
[06:45:03] <KotH> av500: juup
[06:47:27] <Yuvi> "Yamka is designed around clear separation of responsibilities between an EBML Element and EBML Element Payload." that's a reason to develop a new library?
[06:47:59] <av500> yamka?
[06:48:09] <Yuvi> what sorenson developed to write webm
[06:48:21] <Yuvi> http://sourceforge.net/projects/yamka/
[06:48:39] <av500> they needed to write a new lib to write mkv files?
[06:49:38] <av500> urg, cpp
[06:49:57] <astrange> well, libmatroska isn't great but i didn't see any problem with using it for encoding
[06:51:12] <Yuvi> and libmkv is even simpler
[06:51:24] <Yuvi> I think yamka's headers are larger than all of libmkv's source
[06:51:37] <astrange> does it do audio?
[06:51:52] <Yuvi> yeah
[06:52:55] <KotH> av500: gg was already fined for not complying with local laws regarding street view
[06:53:17] <KotH> av500: got us a nice comment from gg saying that swiss people are privacy nazis
[06:54:18] <av500> I assumed swiss ppl would use their mandatory gun to shoot the street view cars
[06:55:07] <KotH> said rifle is for official use only ;)
[06:57:34] <av500> come on, you need some target practice
[07:15:18] <Tjoppen> has anyone used libMXF? it seems to insist on generating TimecodeComponents that don't properly link with with their Tracks
[07:17:42] <astrange> why does webm allow OutputSamplingFrequency if it can't contain sbr?
[07:18:15] <Yuvi> they're working on grafting sbr onto vorbis?
[07:18:45] <Yuvi> flix doesn't think audio packets deserve to have the keyframe flags set...
[07:21:31] <Yuvi> (it also decimated my 30 fps to 12 for no apparent reason)
[07:22:36] <Yuvi> is there an actual spec for webm other than "matroska with webm doctype"?
[07:23:21] <av500> hmm, vp8 allows several outframes per decoded frame, what is the temporal relationship?
[07:23:55] <Yuvi> are those the reference-only frames / not intended for display?
[07:23:58] <astrange> http://www.webmproject.org/code/specs/container/
[07:24:04] <astrange> those are the altref frames
[07:24:23] <av500> Yuvi: well, their simple_decoder example does that, but their lavc driver just uses 1....
[07:24:24] <astrange> er, altref means less output frames than input
[07:24:36] <Dark_Shikari> astrange: why?
[07:24:47] <astrange> because the altref is invisible
[07:24:49] <Dark_Shikari> altref is just another reference frame iirc
[07:24:55] <Dark_Shikari> at least that's what the spec said
[07:25:05] <astrange> When enabled, the VP8 encoder will at its discretion inject a new frame — the alternate reference (AR) — into the output prior to the frame that depends on it. There will be at MOST 1 frame added between I/P-frames. The dependent frame (D) will always be a P-frame. The AR will be marked with the invisible flag by the codec SDK. This frame must be decoded before D, but will produce no output on its own.
[07:25:12] <astrange> stuck in the container spec
[07:25:36] <Dark_Shikari> that's the encoder delay
[07:25:39] <Dark_Shikari> not an invisible frame
[07:26:31] <astrange> the frame is also supposed to have its timestamp set 1 tick less than the next frame, so it certainly wouldn't show up for very long anyway
[07:27:26] * Dark_Shikari never understood alt / golden
[07:27:47] <astrange> i think alt is new, golden is just long-term referencing the keyframe
[07:28:26] <Dark_Shikari> yeah I know
[07:28:55] <Dark_Shikari> the worst part is due to alt+golden, the quality is all over the place
[07:28:59] <Dark_Shikari> psnr bounces like a superball on crack
[07:29:01] <Yuvi> "Seeking will be disabled if the webm file does not have a key frame Cues element." :<
[07:29:56] <superdump> that description makes it sound like the encoder has the option of inserting a frame for reference purposes only and that will not be displayed
[07:30:15] <superdump> Yuvi: are you interested in implementing an ogg mapping for vp8? :)
[07:30:15] <astrange> give them a patent license for mbtree
[07:30:21] <Dark_Shikari> astrange: lol
[07:30:31] <Dark_Shikari> sounds like they sure as hell need it
[07:30:35] <Yuvi> superdump: writing it? no, someone else is doing that
[07:30:44] <Yuvi> once I've heard it's final I'll add it to the demuxer
[07:30:53] <superdump> i said implementing :)
[07:31:06] <Yuvi> well is it final?
[07:31:07] <superdump> the mapping documentation should hopefully go up soon
[07:31:19] <Yuvi> I saw a pre-1.0
[07:31:35] <Yuvi> and I don't really want a 4th _old mapping variation in the decoder
[07:31:38] <Yuvi> demxuer
[07:31:45] <superdump> :)
[07:31:51] <superdump> understandable
[07:32:23] <superdump> just checking you were aware of it, i figured you would be
[07:34:21] <superdump> Yuvi: http://people.collabora.co.uk/~slomo/webm/ogg-vp8.pdf <--- that's apparently going to be the final mapping, so slomo says
[07:34:26] <superdump> he told me to point you to that
[07:34:38] <superdump> they're going to sort out getting it up on xiph.org soon
[07:34:48] <superdump> he also said he's not going to change anything in that
[07:35:58] <av500> there is not much in there to change
[07:36:13] <superdump> indeed
[07:36:14] <astrange> ugh, i have the feeling that altref means perian will break
[07:36:40] <superdump> astrange: why? does perian have to have one frame in, one frame out or something?
[07:36:49] <astrange> yes
[07:36:53] <superdump> 'oop's
[07:36:58] <av500> so, just reoutput the last frame
[07:37:03] <astrange> or rather, quicktime has to have it, and perian's middleman handling isn't good enough
[07:37:13] <superdump> that's what it sounds like to me
[07:37:23] <Yuvi> xiph's theora decoder uses a flag for dropping that should mostly work
[07:37:48] <Yuvi> ignoring frame-accurate seeking/stepping
[07:38:26] <astrange> if it's a p-frame at the right place that should be ok (except for stepping sometimes displaying the same frame twice)
[07:39:01] <astrange> but perian's video decoder just crashes if the ffmpeg codec does anything at all unexpected with buffer orders. need to rewrite it
[07:39:45] <Yuvi> I have no clue how that works, I tried reworking ffmpeg's vp3's buffer model to what I thought should work and it still crashed
[07:41:19] <Yuvi> maybe I should write PrevSize in mkv, I didn't know that field existed
[08:05:16] <Yuvi> bcoudurier: another reason for the 1/1000 timebase in mkv: block timestamps are (cluster timestamp) + 16 bits signed
[08:05:53] <bcoudurier> humm, so how can mkv claim to support nano sec accuracy ?
[08:06:04] <Yuvi> one block per cluster
[08:06:21] <Yuvi> which has like 6% overhead at reasonable bitrates
[08:06:37] <Yuvi> it's one of those "in theory" things
[08:06:44] <bcoudurier> pfff
[08:06:53] <bcoudurier> the more I lean about mkv the more I hate it
[08:07:12] <KotH> mkv, the new ogg?
[08:07:29] <bcoudurier> btw did you check the updated ogg patch ?
[08:07:46] * av500 never liked mkv
[08:10:03] <hyc> av500: what format is better?
[08:10:23] <hyc> bcoudurier: any chance you'll have time to review the remaining ffserver patches I posted?
[08:10:41] <bcoudurier> yes, I'll try
[08:10:47] <bcoudurier> this weekend at worst
[08:10:48] <hyc> cool, thx
[08:11:03] <bcoudurier> is it me or we suddently have a big bump in patch submission ?
[08:11:13] <bcoudurier> which is great, don't get me wrong :)
[08:11:22] <hyc> well, I had 85 new emails on the ffmpeg list in the past 8 hours
[08:11:28] <hyc> looks like a bump :P
[08:11:36] <Yuvi> bcoudurier: I think it looks okay, though it means I now really have to support seeking within a page
[08:11:43] <Yuvi> in the demuxer
[08:12:13] <bcoudurier> ah, indeed :)
[08:12:15] <Yuvi> which in turn means porting oggz_auto which I've been trying to ignore ;P
[08:12:52] <Yuvi> unrelated: maybe vrev should be dropped and you just change the bitstream version to 3.2.1 for old files?
[08:15:05] <av500> hyc: the most sane container I encountered so far is asf
[08:15:09] * av500 hides
[08:16:52] * hyc throws brick at av500 anyway 
[08:29:26] <bcoudurier> Yuvi, can I do that ?
[08:29:54] <Yuvi> hm, maybe not if you care about pre-3.2.0
[08:30:07] <Yuvi> but I'm pretty sure the muxer can't do anything with those since they only have 1 header
[08:30:23] <Yuvi> the only difference 3.2.0 -> 3.2.1 is that pts change
[08:32:34] * av500 watches VP8 decode on omap3.... :)
[08:34:51] <av500> and it crashes in vpx_codec_destroy() doh
[08:35:04] <hyc> destroy() was a good name for it then
[08:37:51] <av500> yes, it destroys...
[08:38:45] <kshishkov> so nothing to complain about?
[08:39:14] <av500> as in?
[08:39:54] <kshishkov> as if it crashed in vpx_codec_finish_and_not_segfault()
[08:40:42] <elenril> is it me or interwebz are crashing?
[08:40:55] * elenril can't get on a bunch of unrelated sites
[08:41:35] <kshishkov> IIRC last time Internet crashed because of some Czech provider
[08:41:56] <elenril> does e.g. tvtropes.org work for you?
[08:42:40] <av500> it does
[08:43:31] <elenril> hmm, works again
[08:43:34] <elenril> with absurd pings
[08:43:51] <elenril> ~180ms
[08:43:59] <kshishkov> local ISP troubles, I think
[08:44:54] <av500> InternetSlowdownProliferator
[08:47:18] * elenril blames av500 
[08:48:26] * av500 takes blame
[08:49:40] * KotH blames av500 
[08:49:41] <av500> hmm, it crashes for 360p content, but does not crash for 720p...
[09:05:05] <av500> mru: no wonder the tegra ppl dont talk to you on the forum, you are calling them idiots
[09:05:27] <twnqx> friend of mine makes fun of the tegra demoboard
[09:05:35] <twnqx> "nonworking pos" and stuff like that
[09:24:04] <kshishkov> g'day, matey
[09:36:41] <KotH> twnqx: wood screws?
[09:36:50] <KotH> kshishkov: you scared him!
[10:03:51] <Kovensky> <+elenril> with absurd pings <+elenril> ~180ms <-- my avg ping is 1800ms :(
[10:04:40] <elenril> o_0
[10:06:43] <elenril> clearly it's time for you to move to a civilized country
[10:21:27] <mru> av500: where did I call then idiots?
[10:21:48] <mru> I certainly _implied_ as much, but I don't remember saying it explicitly
[10:27:33] <mru> I also made no references to today's dilbert strip
[10:31:18] <j-b> can I call someone idiot and not imply it?
[10:31:31] <mru> that's harder
[10:40:04] <mru> you have to do it rather indirectly
[10:44:40] <lu_zero> yaaawn
[10:44:46] <lu_zero> what's about tegra?
[10:46:46] * av500 is entering "idiot" in the tegra forum search
[10:58:16] <twnqx> if that fails, try baka
[12:05:02] <peloverde> happy thursday
[12:05:19] <mru> morning peloverde
[12:05:30] <kshishkov> what's so ahppy about thursdays?
[12:06:38] <mru> next day is friday
[12:06:42] <thresh> exacrlt
[12:06:51] <thresh> oh god. i wish friday evening came faster.
[12:06:59] <mru> and friday is beer day
[12:07:21] <peloverde> new comic books came out yesterday, NBC comedies to DVR tonight, and thursday is beer day here
[12:07:44] <mru> the wine club at the uni here is doing a belgian beer special tomorrow
[12:08:38] <peloverde> do we have the case against autotools writtendown somewhere?
[12:08:47] <mru> ffmpeg/configure
[12:11:50] <thresh> "thursday is beer day here", so you get dead drunk today and don't work tomorrow? niiiiiiice.
[12:11:50] <av500> wasnt drunk a prerequisite to work on AAC PS=
[12:11:50] <peloverde> gnome quacks are trying to get libvpx to switch to autotools and I'd like to give them a concise argument why autotools is retarded
[12:11:50] <mru> the best days are those when you can drink at work
[12:11:50] <av500> ?
[12:12:48] <mru> peloverde: rational arguments don't work on gnome quacks
[12:12:48] <peloverde> friday I usually just work hungover
[12:12:48] <peloverde> mru: I'm not trying to convince the gnome quacks just google
[12:14:27] <mru> do you ever play FPS games?
[12:15:55] <superdump> if someone proposes an autotools patch that is proven to work for all the cases needed and proves they will maintain it and arguments are presented against autotools but no solution implemented, which might google choose?
[12:16:11] <mru> noticed how all them have a type of monster that's harmless on its own, but often attacks in large swarms?
[12:16:22] <superdump> yup
[12:16:27] <superdump> zerg rush kekekeke
[12:16:44] <mru> they're usually spider-like
[12:16:47] * av500 only plays 24FPS movies
[12:17:07] <mru> shotgun tends to be the best defence
[12:17:08] <thresh> you hate all those fancy 29.97 newer ones?
[12:17:12] <Compn> av500: on a 24hz tv ?
[12:17:15] <mru> or flame thrower if you have one
[12:17:31] * TheFluff wonders where mru is going with this
[12:17:36] * superdump too
[12:17:41] <av500> violent
[12:17:43] <mru> freetards are exactly like that
[12:17:52] <kshishkov> mru: you're wrong. Try Gatling
[12:17:57] <Compn> hahaha
[12:18:02] <mru> gatling is too slow to start
[12:18:08] <superdump> run out of ammo too quickly as well
[12:18:09] <kshishkov> flamethrower?
[12:18:20] <av500> thresh: I dont like the visual effect on the missing 0.03Hz :)
[12:18:23] <mru> played resistance 2?
[12:18:51] <kshishkov> definitely not
[12:18:54] <mru> kshishkov: you have a ps3 in the office
[12:18:59] * kshishkov had no luck with 3D games
[12:19:06] <mru> I'd be surprised if that game isn't there
[12:19:20] <Compn> i get motion sickness playing portal sometimes :P
[12:19:26] <twnqx> did you play duke nukem forever?
[12:19:31] <j-b> peloverde: no autotools, so what do you suggest?
[12:19:55] <peloverde> a shell script works just fine for ffmpeg
[12:19:57] <thresh> go cmake go go go
[12:20:11] * kshishkov gives a bottle of methanol to thresh
[12:20:48] <mru> kshishkov: pointless... he's cheering on cmake so he must be blind already
[12:21:06] <kshishkov> that causes death too
[12:21:40] <superdump> a shell script does work fine for ffmpeg and mru does a great job of maintaining it so it works well
[12:21:40] <thresh> I'm not in a suicide mood today :-(
[12:21:53] <mru> use it to build a flame thrower instead
[12:22:04] * av500 votes for mru as lipvpx shell script maintainer
[12:22:08] <superdump> but if no other shell script ninja steps up to do the work and google don't do it (or don't do it well) either then...
[12:22:22] <j-b> anything would be better than their actual shit
[12:22:23] * mru votes for obsoleting libvpx with a native decoder
[12:22:43] <kshishkov> thresh: все равно, для людей, поддерживающих cmake единственный напиток - водка "Палёнушка"
[12:22:58] <peloverde> I just don't want the thing to check for a FORTRAN compiler every time I run it
[12:23:00] <kshishkov> mru: fine with me
[12:23:02] <av500> wodka what?
[12:23:04] <thresh> kshishkov: это факт :))
[12:23:30] <kshishkov> av500: local Russian and Ukrainian brand of the cheapest vodka
[12:23:46] <thresh> cheapest == made of methanol
[12:23:49] <av500> yes, what does the name mean?
[12:24:03] <kshishkov> 'counterfeited"
[12:24:10] <av500> ah, that brand
[12:24:34] <thresh> P + Alenushka
[12:27:42] <peloverde> I really enjoy the public review they had of the bitstream format before setting it in stone... oh wait
[12:28:32] <av500> peloverde: what, u missed it? :)
[12:28:40] <kshishkov> lucky guy!
[12:28:43] <mru> the problem is you can't do a public review _and_ release it to thunder and lightning
[12:28:53] <mru> obviously, thunder and lightning is more important than quality
[12:29:19] <kshishkov> Apple prefers glamour and sequins instead
[12:30:35] <peloverde> There is also no reason the FFmpeg patches couldn't have been pre-reviewed except for lack of interest on their part
[12:31:42] <mru> several of us even had NDAs in place
[12:32:32] <mru> not that I agree there was any need for one
[12:32:50] <av500> NonDamageAgreement was needed for Dark_Shikari
[12:34:13] <mru> only because vp8 is too weak to stand on its own
[12:34:20] <mru> without a mountain of hype to support it
[12:34:26] <kshishkov> so it will stand on xiph's shoulders
[12:34:41] <peloverde> maybe they shouldn't have called it 50% better than h.264
[12:34:50] <mru> xiph's shoulders are too slopey to stand on
[12:35:12] <kshishkov> not if you full of hot air
[12:35:59] <mru> but then xiph will function more as an anchor
[12:36:23] <kierank> mru: or a low water mark
[12:43:38] <Kovensky> <@peloverde> I just don't want the thing to check for a FORTRAN compiler every time I run it <-- bug fixed in libtool 2.2.7a (aka svn snapshots) =p
[12:44:05] <kshishkov> what about Ada compiler check?
[12:52:10] <janneg> Dark_Shikari's review is good for identifying and classifying freetards
[12:52:52] <pross-au> what's a freetard?
[12:52:59] <mru> free+retard
[12:53:20] <pross-au> I understand what a retard is. What does 'Free' have to do with it?
[12:53:43] <mru> not talked to debian folks much?
[12:53:58] <pross-au> Oh 'Free'
[12:54:07] <pross-au> U gotta capitalise the 'F'
[12:54:24] <kierank> and put GNU/Freetard
[12:54:35] <av500> gnutard?
[12:54:42] <mru> gnu turd
[12:54:51] <av500> is that a gcc coredump?
[12:55:00] <pross-au> Haha
[12:55:16] <av500> has "duck" a special meaning for on2?
[12:55:17] <kshishkov> "Free" as "a free beer", "free" as "a free speech" and "free" as "'free' in Stallman speeches"
[12:55:37] <kshishkov> av500: it was originally Duck Corporation
[12:55:39] <mru> freedom to do as I say
[12:55:52] <av500> kshishkov: ah, that explains: #define duck_memalign(X,Y,Z) vpx_memalign(X,Y)
[12:56:09] <pross-au> So we're talking about the extreme-socialist-left of software politics?
[12:56:33] <kshishkov> what's the difference with pirate parties?
[12:56:44] <av500> what is the extreme-conservative-right of software politics?
[12:56:52] <pross-au> Microsoft
[12:57:01] <mru> kshishkov: pirate parties want something that actually benefits them
[12:57:05] <mru> freetards don't
[12:57:45] <kshishkov> yes, it's all principles and higher good
[12:57:54] * av500 got a visit from his local pirate party candidate at home
[12:58:14] <hyc> well, *someone* has to have some principles around here
[12:58:15] <mru> did he bring some dvds?
[12:58:32] <av500> "she" did not and my wife described here as a "geek type"
[12:59:21] <kierank> do they have a serious campaign?
[12:59:29] <av500> "there was a woman here today who looked like she works with computers"
[12:59:38] <av500> kierank: here in .de, yes
[12:59:50] <kshishkov> s/.de/.se/
[12:59:57] <mru> both
[13:00:02] <mru> it's a total joke in the uk
[13:00:02] <kshishkov> .se is better
[13:00:09] <av500> the best! :)
[13:00:14] <kshishkov> all politics is a total joke in UK
[13:00:21] <peloverde> to better articulate my opposition against freetards, I think saying software patents can just be avoided expresses a certain naivety
[13:00:32] <av500> mru: come on, you are ruled by a house of lords...
[13:00:43] <mru> peloverde: you're being far too polite
[13:01:08] <peloverde> I think some of them are legitimately naive
[13:01:20] * pross-au puts fingers in ears and signs alalalallaala
[13:01:21] <mru> is there such a thing?
[13:01:32] <KotH> peloverde: allah? ;)
[13:01:41] <pross-au> to admit software patents are a problem, is to admit defeat.
[13:02:00] <mru> ALLAH AKBAR!
[13:02:01] <kshishkov> avoided - no, ignored - hell yes!
[13:02:14] <peloverde> the first step of fixing a problem is admitting you have one... or something like that
[13:02:23] <mru> now let's draw mohammed!
[13:02:44] <mru> is there an official smiley for a beard?
[13:02:51] <hyc> :}
[13:02:53] <pross-au> :}
[13:02:53] <kshishkov> :)>
[13:02:57] <mru> and turban?
[13:03:00] <kshishkov> {
[13:03:03] <pross-au> [:}
[13:03:04] <hyc> hm, ya got me on that
[13:03:10] <janneg> =:)>
[13:03:43] <hyc> dunno about avoiding s/w patents. better to just abolish them
[13:04:50] <hyc> keep patents purely in the physical realm
[13:05:09] <kshishkov> and in very strict limits
[13:05:11] <mru> kill those too
[13:05:12] <janneg> there are far too many and they are too vagely described to even find most of the ones you possibly violate
[13:05:21] <mru> patents have outlived their purpose
[13:05:26] <hyc> and if you develop a hardware process that someone is able to reproduce in software, tough luck. software is not hardware, so patent doesn't cover.
[13:05:27] <peloverde> agreed, now if the freetards dumped as much money is lobbying as in funding freetard codecs maybe we'd have a shot at it
[13:06:01] <kshishkov> mru: the UK once had it right - they granted each patent with Parliament act
[13:06:23] <hyc> that would certainly slow things down ;)
[13:06:49] <mru> patents only worked when the invention was still useful much longer than the patent validity
[13:07:06] <hyc> I dunno, there's so many people entrenched in keeping the patent system alive, who would have to look for legitimate work after that
[13:07:06] <mru> and when the system wasn't subject to constant abuse
[13:07:11] <av500> patents were meant to allow others to do what you do, in exchange for a license, if I am able to do pinch zoom without reading the patent - in dont need to pay - so easy...
[13:07:20] <kshishkov> with current situation it's not the inventor who benefits from patent (much) which defies its original purpose
[13:07:31] <hyc> hard to believe any one movement would have enough lobbying power to get rid of them completely
[13:07:39] <kshishkov> hyc: because it's $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ or so
[13:07:42] <mru> so let's start shooting lawyers
[13:07:48] <twnqx> and lobbyists.
[13:07:50] <av500> kshishkov: you forgot one $
[13:07:50] <hyc> that might work
[13:07:52] <mru> same thing
[13:08:07] <kshishkov> av500: I forgot many of them
[13:08:10] <twnqx> lobbyists are more often retired politicians
[13:08:23] <kshishkov> who keep their source of bribes
[13:08:28] <mru> and many politicians are lawyers
[13:08:32] <hyc> right, patents were meant to allow you to license something from someone else that you couldn't figure out on your own
[13:08:40] <av500> hyc: right
[13:08:50] <hyc> but now, if you independently invent the same thing, you're in breach
[13:08:52] <hyc> that's BS
[13:09:06] <mru> that was always the case
[13:09:13] <hyc> ok,true
[13:09:17] <hyc> but it's still BS
[13:09:23] <mru> but the idea was that patents would only be granted for things very unlikely to be invented by several people
[13:09:25] <hyc> but ... it was much less of a problem before
[13:09:44] <hyc> because it was seldom possible to independently invent something that was actually patent worthy
[13:09:44] <av500> there is that nice article about the steam engine patent
[13:09:52] <mru> hence the requirement for the invention being non-obvious to someone skilled in the relevant art
[13:09:52] <twnqx> <hyc> right, patents were meant to allow you to license something from someone else that you couldn't figure out on your own <- no, patents were to protect the small inventor from the large company who would just have him carry the burden of development, then reap the rewards
[13:10:15] <mru> the idea of patents was good
[13:10:41] <hyc> twnqx: ok.
[13:10:43] <av500> somebody should have patented it...
[13:10:53] <kshishkov> meta-patented
[13:11:31] <hyc> one of my previous ideas was to prevent corporations from owning IP. if you work at a company and create something that gets patented, you keep it, not the corp.
[13:11:49] <kshishkov> and did it work out?
[13:11:53] <av500> hyc: you are free to negotiate that with the corp
[13:12:01] <hyc> if they want to be able to use it without a license, they need to keep you happily employed
[13:12:03] <mru> av500: free to try
[13:12:08] <av500> sure
[13:12:13] <kierank> well it would be part of your employment contract
[13:12:39] <av500> corp might remind you that it pays you for your time btw... :)
[13:12:48] <hyc> kierank: to an extent. the idea is that your employment contract can't make you sign anything over.
[13:13:01] <av500> hyc: why not=
[13:13:03] <av500> hyc: why not?
[13:13:28] <hyc> av500: because I don't think corporations are natural persons, and should not have the right to own things.
[13:13:47] * av500 goes out to re-own the local bank
[13:14:21] <hyc> IP should belong to the individual creators, that's all
[13:14:21] <av500> damn, the security gaurd said "hyc? never heard of him"
[13:14:27] <hyc> :P
[13:14:30] <pross-au> nothing wtong with a 'limited' monopoly on inventions and artwork
[13:15:07] <hyc> right now, corporations have so much vested in preserving their huge patent portfolios
[13:15:07] <mru> yes, limited
[13:15:26] <hyc> and they wield them like large clubs, bludgeoning everyone else around
[13:15:45] <pross-au> with the pace of technology, that limit should be 10 years or something.
[13:15:58] <hyc> that would happen less often if the patents belonged to the individuals, not the corps
[13:16:17] <mru> pross-au: http://a-hackers-craic.blogspot.com/2010/05/australian-bananas-suck.html
[13:16:41] <pross-au> on the other hand, patents make the research/work public
[13:17:01] <hyc> true, it's better for the world than everything staying trade secret
[13:17:10] <mru> not really
[13:17:20] <av500> I'd prefer the 1click to have stayed a trade "secret"
[13:17:23] <mru> they can refuse to license it
[13:17:36] <av500> ppl would have puzzled for years how it works
[13:17:37] <mru> a plain old secret can be legally reverse engineered
[13:18:43] <pross-au> yep
[13:19:08] <av500> but it would have been hard to have e.g. mpeg standardisation with trade secrets
[13:19:28] <mru> they could still make money selling implementations
[13:19:34] <pross-au> exactly
[13:19:40] <mru> but then they'd have to make actually good encoders
[13:19:42] <pross-au> make money producing things people want
[13:19:45] <mru> and that's :effort:
[13:19:51] <hyc> sigh...
[13:19:51] <pross-au> a free maket?
[13:19:54] <pross-au> *market
[13:19:55] <av500> and designing h264 is not?
[13:20:19] <hyc> make money by producing things that are actually worth paying for
[13:20:57] <hyc> when the EULA for a product disclaims fitness for any purpose, you're paying money for something that wasn't fit to be sold.
[13:20:58] <mru> patents are work once, get paid forever
[13:21:40] <hyc> mru: not quite forever. copyrights are forever. :P
[13:21:47] <mru> and diamonds
[13:22:05] <hyc> all of it is a broken system, and it was broken by the large corps who own all of the IP
[13:22:20] <janneg> and if forever arrives copyrights are extended for Disney
[13:22:23] <hyc> take away their ownership, and it removes their ability to profit from these stupid games.
[13:22:29] <janneg> s/if/when/
[13:22:42] <pross-au> copyrights are not forever
[13:22:52] <mru> some are
[13:22:54] <hyc> 70+ years now, close enough
[13:22:57] <TheFluff> in practice they are because the terms just keep getting longer and longer
[13:23:01] <KotH> pross-au: do you know the mikey mouse law?
[13:23:01] <hyc> yep
[13:23:25] <TheFluff> nothing produced after ~1930 will ever get into the public domain at this rate
[13:23:26] <pross-au> Yeah, that mickey mouse steamboat video is Public Domain here
[13:23:36] <TheFluff> (in the US, that is)
[13:23:57] <pross-au> Change your government
[13:23:57] <KotH> pross-au: not for long
[13:24:04] <mru> and nowadays many copyrights belong to companies
[13:24:04] <KotH> pross-au: just wait until ACTA gets passed
[13:24:05] <peloverde> in practice no new works will enter the public domain except for a handful of things that copyright law explicitly excludes
[13:24:38] <mru> do the 70 years start counting when the company is bankrupt or what?
[13:24:56] <hyc> and then there's government funded research, where the results are given to a private corporation to monetize
[13:24:59] <peloverde> the allegedly "originalist" justices on the us supreme court have no problem with this
[13:26:16] <hyc> yeah, most of the current govt needs to just be thrown out. and a lot of constitutional amendments, and most new laws written since the 1930s
[13:26:48] <mru> it's not just govt that's a problem
[13:27:01] <mru> is the way businesses are locked into these bizarre ways
[13:27:18] <mru> everybody knows everything they say is bullshit
[13:27:30] <mru> but they have to say it because everybody else is saying it
[13:27:39] <mru> and if you don't say it you'll look bad
[13:28:12] <mru> biggest cargo cult operation ever
[13:28:41] <KotH> mru: afaik, the 70 years start counting from the moment on, when the _original_ author dies
[13:28:46] <KotH> mru: but IANAL
[13:29:27] <TheFluff> isn't it 90 years in the us now
[13:30:15] <peloverde> it's
[13:30:23] <peloverde> FFMAX(90, life+70)
[13:30:39] <TheFluff> 95 years from publication or 120 years from creation whichever is shorter (anonymous works, pseudonymous works, or works made for hire, published since 1978)[298]
[13:30:43] <TheFluff> 95 years from publication for works published 1964-1977; 28 (if copyright not renewed) or 95 years from publication for works published 1923-1963 (Copyrights prior to 1923 have expired.)[299]
[13:30:47] <TheFluff> says wikipedia
[13:30:50] <TheFluff> 70 years for works published after 1978
[13:31:13] <TheFluff> 70 years after death of creator, that is
[13:31:29] <av500> the creator is dead? omg
[13:32:07] <TheFluff> lorf
[13:36:15] <av500> opensuse supports vp8?
[13:37:54] <kshishkov> yep, opensuse is very evil
[14:00:21] <dezodorant> hi all
[14:00:41] <dezodorant> got nice bug/feature in probing
[14:01:50] <dezodorant> libav detecting wrong bit rate, if first packet 0-length
[14:03:56] <dezodorant> in "FFmpeg version SVN-r0.5.1-4:0.5.1-1ubuntu1" detection was fine
[14:05:36] <BBB> might want to report that on roundup.ffmpeg.org
[14:05:41] <BBB> as a regression
[14:05:50] <BBB> lu_zero: I like it, if you can make the html that'd be great
[14:05:56] <jai> against trunk revisions
[14:15:33] <lu_zero> ping me this night
[14:17:30] <lu_zero> hopefully I'll have the split images by that time
[14:17:37] <lu_zero> otherwise I'll split them mysel
[14:19:55] <lu_zero> f
[14:20:23] <av500> dont split yourself
[14:20:48] <kshishkov> yes, fork yourself!
[14:21:38] <KotH> no! dont! a child process is very time consuming!
[14:21:49] <BBB> what if you have twins?
[14:22:34] <mru> what if fork() occasionally created two new processes?
[14:22:38] <janneg> forked long time ago if identical
[14:22:44] <kshishkov> BBB: thrice as worse - synergy!
[14:23:08] <BBB> awesome
[14:23:13] <BBB> kshishkov: how's your vp8 decoder going?
[14:23:27] <mru> kshishkov is working on a vp8 decoder?
[14:23:36] <KotH> native vp8 decoder?
[14:23:41] <av500> I doubt it, it has specs, so no fun
[14:23:49] <mru> we already have a naive one, only add t
[14:23:50] <kshishkov> no, I just expressed agreement with mru on idea to have native one for FFmpeg
[14:24:20] <kshishkov> I should be able to start working again soon
[14:25:02] <av500> you dont work now?
[14:25:04] <BBB> looking through the sources is fun
[14:25:07] <dezodorant> https://roundup.ffmpeg.org/issue1954 , but i`m wondering - is it normal to place such packets at the begining
[14:25:09] <BBB> much easier than RE'ing
[14:27:55] <kshishkov> I tend to disagree
[14:28:05] <kshishkov> in both cases you need to understand their logic
[14:28:07] <mru> c++ is easier to RE
[14:29:11] <kshishkov> RV4 decoder was C++
[14:31:38] <BBB> the logic is easier if you know where to start
[14:31:51] <BBB> having a haeder file telling you which function is init/decode/etc. makes things easier
[14:32:22] <kierank> that is true
[14:33:01] <mru> if you have a dll there are usually named entry points
[14:33:42] <kshishkov> yes, good luck with COM
[14:35:45] <BBB> the named entry points are usually called "DllServer"
[14:35:49] <BBB> that isn't terribly helpful
[14:38:27] <kshishkov> what is not really helpful is that loads of functions are called via dynamic function pointer tables
[14:39:02] <BBB> right
[14:39:16] <BBB> DllServer in my case tends to load a function ptr table, call into it and that's all I know
[14:39:17] <BBB> it sucks ass
[14:39:23] <BBB> I prefer documented sources
[14:39:27] <BBB> no matter how illogically documented
[14:39:30] <BBB> they are documented
[14:39:43] <kshishkov> well, VfW codecs are usually easy to RE
[14:41:14] <BBB> the sources are funny
[14:41:25] <BBB> "next 3 bytes are width and height, 12 bits each"
[14:41:28] <BBB> then the code does:
[14:41:45] <BBB> width = read_be16(data+3) & 0x3fff;
[14:41:51] <BBB> height = read_be16(data+5) & 0x3fff);
[14:42:01] <BBB> something is confusing there
[14:42:12] <mru> comments are evil
[14:42:17] <BBB> heh :)
[14:42:29] <BBB> I think the code is wrong
[14:42:35] <BBB> there is a comment and a check of width height
[14:42:38] <BBB> and they're all commented out
[14:42:44] <BBB> and there's a second function that re-reads them
[14:42:47] <mru> the comment is commented out?
[14:42:58] <BBB> sorry, a printf
[14:43:16] <kshishkov> is it read the same way second time?
[14:43:56] <kshishkov> also I have some experience understanding TrueMotion VP2 source
[14:44:14] <BBB> the function is called "peek_si"
[14:44:20] <BBB> suggesting there is a second "get_si" or so
[14:44:25] <BBB> but doesn't appear to be that way
[14:44:28] <BBB> maybe the comment is wrong
[14:44:30] <BBB> dunno
[14:44:35] * BBB goes find a webm movie
[14:44:43] <kshishkov> www.youtube.com?
[14:45:45] <av500> BBB: http://lachy.id.au/lib/media/elephantsdream/Elephants_Dream-360p-Stereo.webm
[14:45:59] <BBB> fantastic, thank you
[14:46:04] <av500> http://lachy.id.au/lib/media/elephantsdream/Elephants_Dream-720p-Stereo.webm
[14:46:10] <BBB> 1 is enough :)
[14:48:23] <j0sh_> is webm streamable?
[14:48:33] <mru> it's mkv
[14:48:59] <j0sh_> mkv restricted to vp8 and vorbis basically?
[14:49:29] <twnqx> if i read it correctly only stripped by some rarely used features
[14:49:39] <twnqx> though Dark_Shikari didn't go into details :P
[15:03:51] <BBB> twnqx: mkv isn't that bad, so little to complain there
[15:04:09] <BBB> j0sh_: hey there
[15:18:22] <dezodorant> any audio gurus here ?
[15:18:38] <kshishkov> what specific area of audio?
[15:18:41] <BBB> ask :)
[15:18:55] <hyc> the answer is 42
[15:18:59] <av500> the MOD guru is not in today...
[15:19:26] <dezodorant> i got a lot of audio helpers and classes for processing
[15:19:47] <Tjoppen> BBB: did you see my example test program based on libmicrohttpd?
[15:19:56] <dezodorant> and planing to start project for common audio tasks
[15:20:18] <dezodorant> only one part is missing - decoding
[15:20:37] <dezodorant> now i`m using libav for this purposes
[15:21:56] <dezodorant> but i do not need lot of stuff from libav and want to fork some audio related code from libav
[15:22:20] <dezodorant> so question: is it ok ?
[15:22:24] <twnqx> why fork? why not contribute back and improve?
[15:22:34] <dezodorant> c++
[15:22:39] <twnqx> yuck
[15:22:40] <mru> that's not ok
[15:23:01] <dezodorant> ;-(
[15:23:46] <mru> calling libavcodec from c++ works fine
[15:23:49] <BBB> dezodorant: I don't see why you wouldn't just use ffmpeg without encoders
[15:23:57] <mru> for values of fine compatible with c++
[15:24:00] <BBB> forking code is bad practice and will most likely just frustrate both you and us
[15:28:21] <dezodorant> most part of my framework already compleated and written in c++, so hardly they will be usefull for ffmpeg
[15:28:43] <dezodorant> probing and formats implemented in c++ too
[15:28:59] <dezodorant> only codecs i want to take
[15:29:46] <mru> so call libavcodec from your code
[15:29:48] <mru> it's not hard
[15:33:07] <dezodorant> as a variant
[15:34:51] <av500> dezodorant: I do the same, I use ffmpeg only for the codecs
[15:34:56] <av500> works fine
[15:35:50] <j-b> is James Zern on IRC?
[15:42:15] <peloverde> i guess not?
[16:04:42] <hyc> rtsp over http? why not just use plain http then???
[16:05:37] <siretart> a usable C++ wrapper library that would encapsulate (most) important data structure of lavc and lavf would be awesome, I guess.
[16:06:04] * mru fails to see why
[16:06:26] <Compn> hyc : rtsp (and i think rtmp does this too) does http tunneling for people behind corporate firewalls etc
[16:06:33] <elenril> a python wrapper would be nice
[16:07:22] <dezodorant> python is slooooowly
[16:07:34] <elenril> nah, not really
[16:07:43] <dezodorant> yes it is yes it
[16:07:51] <Compn> av500 : what do you play on the videowall ? i have some ideas like 'tokyo scanner' or some imax movies :P
[16:07:57] <dezodorant> and no threading )
[16:07:59] <av500> BBB!
[16:08:10] * BBB kicks av500
[16:08:12] <Compn> big buck bunny eh
[16:08:28] <av500> yes, rendered in really huge
[16:08:40] <kierank> I also have a suggestion but it's a little nsfw
[16:08:47] <kierank> but it was filmed in 4k
[16:08:57] <av500> kierank: I was asking for 4k all the time :)
[16:09:00] <av500> gimme
[16:09:03] <Compn> also there are a bunch of software freedom documentaries, altho those are mostly boring interviews and dont show off much
[16:09:27] <peloverde> Why does gcc assume I don't know order of operations?
[16:10:13] <mru> gcc assumes you're stupid
[16:11:20] <hyc> most people *are* stupid so it's a safe assumption
[16:11:24] <mru> last year TI got slapped for showing quake3
[16:11:30] <mru> nsfw might not be a good idea
[16:11:37] <peloverde> Multiplication comes before addition, they taught us that in elementary school
[16:12:24] <mru> and and is like mul and or is like add
[16:12:26] <janneg> FPS shooter is more problematic than nudity
[16:12:31] <av500> mru: huge BBB and the fosdem loop shold be ok
[16:12:36] <av500> should
[16:12:49] <mru> janneg: in europe yes
[16:12:58] <mru> in the US it's probably the opposite
[16:13:24] <Compn> tokyo scanner is just a bunch of zooms + pans around tokyo via glider or airplane, and most of those imax movies are rated G too :P
[16:13:55] <Compn> er, its a helocopter i remember now
[16:13:56] <Compn> heh
[16:14:18] <mru> but is it available in 4k?
[16:14:40] <BBB> \o/
[16:14:49] * BBB can print the size of a vp8 keyframe
[16:15:07] <mru> INT_MAX?
[16:15:40] <Compn> no, i was only able to find mpeg2 version and its full of blocks :\
[16:15:43] <ramiro> off topic, is dlfcn.h supposed to be #include'able by c++ without extern "c"?
[16:15:56] <ramiro> I don't know what spec to refer to to find this
[16:15:59] <jai> BBB: working on a native decoder?
[16:16:05] <mru> the C++ spec doesn't mention that header afaik
[16:16:26] <mru> ramiro: http://www.opengroup.org/onlinepubs/009695399/basedefs/dlfcn.h.html
[16:16:35] <mru> the spec is defined in C
[16:16:55] <mru> I suspect using those interfaces from other languages is undefined
[16:17:17] <ramiro> mru: thanks, so I don't have to bother with c++
[16:17:51] <mru> as far as posix is concerned, c++ doesn't exist
[16:18:06] <hyc> ah, if only the rest of the world were so sane
[16:18:31] <ramiro> mru: this is the reason I asked http://code.google.com/p/dlfcn-win32/issues/detail?id=7&can=1
[16:18:44] <ramiro> it had appeared in the past and I had answered http://code.google.com/p/dlfcn-win32/issues/detail?id=1&can=1
[16:19:37] <TheFluff> dezodorant: if you only want audio decoding and nothing else http://code.google.com/p/ffmpegsource/ could be useful to you I guess
[16:19:49] <mru> the glibc header has the c++ guards but that doesn't mean anything
[16:19:54] <TheFluff> audio support is kinda unreliable though
[16:21:28] <dezodorant> TheFluff: i already have wrapper )
[16:21:42] <TheFluff> what are you asking for then
[16:21:45] <av500> dezodorant: add another one
[16:21:55] <av500> you need at least 5
[16:22:25] <dezodorant> av500: http://s.lurkmore.ru/images/1/1b/Dont_drink_1.jpg
[16:22:47] <TheFluff> use of ffms2 is also kinda debatable depending on what you want to do, in most scenarios that don't require sample-accurate seeking it's usually easier, faster and less buggy to just use ffmpeg itself directly
[16:23:07] <av500> dezodorant: a russian that does not drink does not exist
[16:23:28] <dezodorant> av500: lolwut
[16:24:50] <dezodorant> i`ll go to feed my bear
[16:25:12] <dezodorant> and play balalaika )
[16:27:13] <hyc> hmm. It seems if I specify -s XxY for the output from ffmpeg, and the X:Y aspect ration doesn't match the input, then ffmpeg will scale to the new size and lose the original aspect ratio
[16:27:34] <hyc> is there a way to force it to preserve the original input aspect ratio?
[16:28:22] <janneg> afaik only with manual padding or specifying a pixel aspect ratio
[16:28:50] <hyc> that's what i thought. ok. was hoping I wouldn't have to calculate the padding myself.
[16:29:02] <janneg> there was recently a related ml post
[16:30:11] <hyc> how recent, this month? I'll check the archive
[16:30:38] <janneg> 2-3 months ago
[16:30:43] <hyc> ok
[16:30:47] <janneg> i'm looking for it
[16:32:31] <hyc> March, Video size to fit in screen ?
[16:36:06] <janneg> not finding that one
[16:36:34] <hyc> this is what I found https://lists.mplayerhq.hu/pipermail/ffmpeg-user/2010-March/024548.html
[16:38:23] <janneg> no, I'm not reading -user. It was a patch micheal rejected for not using avfilter iirc
[16:39:08] <hyc> ah ok
[16:46:38] <janneg> there's "Mar 03 VolodyA! V Anarh  [FFmpeg-devel] [PATCH] pixel aspect ratio" which was subsequencely implemented as avfilter but I don't think I meant that one
[16:50:45] <hyc> hmm, says it's already committed
[16:53:19] <janneg> hyc: libavfilter/vf_scale.c has "-1 = keep original aspect"
[16:54:17] <janneg> so -s 640x-1 should preserve the aspect
[16:54:25] <hyc> hm, cool.
[16:54:45] <hyc> I'll try that, thanks!
[16:57:11] <hyc> nope, got "Incorrect frame size"
[17:06:05] <janneg> hyc: -vf scale=640x-1 doesn't work either
[17:07:18] <hyc> bummer
[17:07:37] <janneg> but -vf scale=640:-1 does
[17:09:07] <hyc> yeah, cool
[17:09:52] <hyc> hm. not sure I can configure that in ffserver.conf
[17:24:23] <superdump> is anyone here familiar with 'tc' from iproute?
[17:24:37] * mru has heard of it
[17:25:04] <mru> <cv>I have 10 years experience</cv>
[17:25:05] <superdump> i'm trying to rate limit all traffic on a network device
[18:16:05] <mru> t-shirt idea: what is your favourite colour?
[18:20:26] <janneg> bikeshed
[18:20:56] <mru> too obscure?
[18:24:27] <BBB> peloverde: holy mother 142kb?
[18:24:35] <mru> mostly tables
[18:26:28] <janneg> mru: probably. something vp8/webm related would be good
[18:26:45] <mru> I'm sick and tired of that already
[18:27:00] <mru> and we already did html5
[18:27:27] <janneg> "webm, faster, ffmpeg"
[18:27:42] <BBB> too much google hype
[18:27:50] <Dark_Shikari> s/google//
[18:27:51] <mru> I'd rather not give that abomination more publicity than necessary
[18:28:18] <hyc> too late ;)
[18:28:29] <Dark_Shikari> notice how all the freetards hate google
[18:28:33] <Dark_Shikari> until google releases vp8
[18:28:37] <Dark_Shikari> then for a short period, they love google
[18:28:41] <Dark_Shikari> then they'll go back to hating google again
[18:28:43] <mru> I did briefly reflect over that
[18:29:01] <mru> how google become evil though?
[18:29:05] <mru> they used to be cool
[18:29:11] <BBB> privacy bla bla
[18:29:20] <Dark_Shikari> the big guy is always evil
[18:29:21] <BBB> as long as lefty aclu organizations start hating you
[18:29:25] <BBB> you get bad press
[18:29:25] <Dark_Shikari> when the little guy becomes the big guy
[18:29:27] <BBB> and everyone hates you
[18:29:30] <Dark_Shikari> they become evil
[18:29:38] <BBB> everyone being freetards and other incompetent, lazy and useless lefties
[18:29:39] <mru> rule #1: what happens on the internet, stays on the internet, _forever_
[18:29:48] <hyc> well, power corrupts
[18:29:56] <BBB> hehe, see facebook?
[18:30:00] <hyc> so naturally, when the little guy becomes the big guy, they also become evil
[18:30:18] <mru> I don't have a problem with facebook
[18:30:19] <Dark_Shikari> and absolute power corrupts absolutely
[18:30:36] <mru> I wouldn't put anything on there that I'd mind if anyone saw
[18:31:23] <hyc> yeah, anyone putting private stuff on an online forum is just asking for trouble
[18:31:33] <hyc> doesn't matter if it's facebook or anything else
[18:31:43] <hyc> if it's supposed to be private, keep it off the net
[18:34:54] <hyc> hmmm... ffm transfers height/width as 16 bit numbers
[18:35:06] <hyc> I guess you won't be realtime-streaming to any video walls
[18:35:50] <mru> ours is only 3840 wide
[18:35:53] <CIA-7> ffmpeg: michael * r23202 /trunk/libavformat/avidec.c:
[18:35:53] <CIA-7> ffmpeg: Disable non interleaved avi code when there is no index available.
[18:35:53] <CIA-7> ffmpeg: Fixes issue1956.
[18:36:08] <hyc> I was actually looking to see if I can pass a -1 in there
[18:36:16] <hyc> but most likely it would be treated as 65535
[18:36:20] <kshishkov> mru: can I get 3XL T-shirt?
[18:36:39] <mru> kshishkov: if they have that size
[18:36:54] <kshishkov> and dark gray?
[18:37:08] <ramiro> kshishkov: did you manage to leave your country? or are you just going out for linuxtag?
[18:37:16] <mru> standard ffmpeg t-shirt is black
[18:37:40] <kshishkov> mru: only before first wash :P But I shan't mind
[18:37:40] <hyc> yah, bummer, get_be16 returns unsigned int
[18:37:49] <janneg> mru: "troll driven development" for the t-shirt?
[18:38:00] <mru> not bad
[18:38:03] <_av500_> +1
[18:38:34] <kshishkov> ramiro: no, I just happen to live in Germany right now
[18:38:48] <hyc> heh heh... "boy OMAP3 is really slow at decoding ____"
[18:38:48] <kshishkov> ramiro: and I don't consider Ukraine to be my country
[18:39:01] <mru> hyc: ?
[18:39:02] <_av500_> hyc: where?
[18:39:07] <hyc> just trolling :P
[18:39:55] <hyc> I think you've got a winner for a t-shirt
[18:40:14] <ramiro> kshishkov: oh, I didn't know you had managed to move to germany...
[18:40:43] <mru> kshishkov: website says they do 3xl
[18:41:15] <kshishkov> ramiro: well, I had some troubles with Internet access away from work
[18:42:44] * kshishkov still misses his hemland
[18:57:13] <pasteeater> is -devel ML having issues or is it my fault?  I haven't recieved anything today.
[18:57:32] <peloverde__> I have not had any problems
[18:57:34] <mru> seems fine here
[18:57:39] <pasteeater> ah, ok. thanks.
[18:58:40] <mru> pasteeater: what's your email? I can check logs for anything unusual
[19:00:08] <pasteeater> mru: lou at fakeoutdoorsman.com
[19:01:05] <mru> pasteeater: connection refused on your mx
[19:01:30] <mru> works now though
[19:02:02] <mru> hmm, IP address has changed
[19:02:12] <pasteeater> damn it!  stupid cloud-thing. i know what to do. thanks for looking.
[19:12:04] <Compn> heh
[19:25:05] <elenril> so PS is done?
[19:25:34] <mru> posted to ML != done
[19:25:47] <mru> you have a + so you should know that
[19:25:57] <elenril> hm i don't
[19:26:03] * elenril wonders where did it go
[19:26:26] <elenril> nvm, /me just fails at reading
[19:28:14] <elenril> anyways, the end of faad is near!
[19:31:40] <BBB> let's kill libvpx
[19:32:22] <mru> KILL KILL
[19:32:48] <mru> anyone working on a proper decoder yet?
[19:36:26] <BBB> I'm reading libvpx and have a skeleton vp8dec
[19:36:30] <BBB> that's about it
[19:36:37] <Dark_Shikari> lol
[19:36:50] <BBB> I want to see how it works before doing anything stupid
[19:36:57] <BBB> the bitstream doesn't seem particularly difficult
[19:37:55] <mru> comparing against vp[56].c might be useful
[19:38:03] <mru> there's bound to be something similar in there
[19:38:24] <Dark_Shikari> yes
[19:38:46] <Dark_Shikari> you'll have to write vp8 versions of the intra pred code
[19:38:55] <mru> companies like to reuse their oldest, most pungent code for the longest time
[19:39:12] <mru> probably because whoever wrote it is a director now
[19:40:03] <Yuvi> the fun vp3 (i)dct that had a huge amount of leakage and only used (16x16 bit)>>16 multiplies was used until vp7
[19:40:17] <Dark_Shikari> vp8 still uses 16x16>>16
[19:40:22] <Yuvi> :/
[19:40:51] <Yuvi> signed x unsigned still too?
[19:41:20] <Dark_Shikari> not sure
[19:41:21] <Dark_Shikari> check the code
[19:41:24] <Dark_Shikari> they had big constants
[19:41:26] <Dark_Shikari> e.g. bigger than 32k
[19:42:06] <mru> ugh
[19:44:03] <mru> BBB: any word on that licence yet?
[19:44:11] <BBB> trying to get a word from them
[19:44:14] <BBB> nothing final yet
[19:45:31] <Yuvi> yep, 16 bit unsigned x signed
[19:46:00] <Yuvi> which requires two instructions on like every simd set ever
[19:47:12] <Yuvi> and they have a constant rounding of 0, why is that still in the code :/
[19:50:16] <Yuvi> at least it wasn't copy pasted into the spec
[19:51:53] <_av500_> mru: btw, I sent nvidia the vp8 question today...
[19:52:10] <mru> let's see if they answer you
[19:52:24] <_av500_> they will
[19:52:57] <mru> my trolling on the forums hasn't been very successful
[20:00:39] <Vitor1001> Dark_Shikari: Just wondering, how much of code of x264 could one reuse to a vp8 encoder?
[20:00:52] <Dark_Shikari> motion search: 100%
[20:01:02] <_av500_> golden frames: 0%
[20:01:03] <Dark_Shikari> scenecut decision and ratecontrol: everything but the b-frames
[20:01:15] <Dark_Shikari> mb-tree: everything
[20:01:27] <Vitor1001> does x264 support disabling b-frames already?
[20:01:29] <Dark_Shikari> yes
[20:01:37] <Dark_Shikari> intra: everything, with small changes
[20:01:40] <Dark_Shikari> entropy coder: rewrite
[20:01:43] <Dark_Shikari> quant/macroblock encode: rewrite
[20:01:45] <Dark_Shikari> dct: rewrite
[20:01:49] <Dark_Shikari> mv pred: rewrite
[20:01:54] <Dark_Shikari> you could reuse most of mode decision
[20:02:41] <Vitor1001> And how better would a franksteinish x264/libvpx mixture be than libvpx alone?
[20:02:55] <Dark_Shikari> well, suppose we did an x264 -> vpx conversion without using much of vpx
[20:02:59] <Dark_Shikari> things we would NOT have:
[20:03:02] <Dark_Shikari> golden/alt ref frame
[20:03:08] <Dark_Shikari> sub8x8 modes (easier to omit them)
[20:03:20] <Dark_Shikari> and a few other minor things I think
[20:03:20] <BBB> what is a goldenframe?
[20:03:29] <Dark_Shikari> It would still beat the living fuck out of libvpx
[20:03:30] <BBB> it's a p-frame that is used as reference?
[20:03:43] <Dark_Shikari> It's a second reference frame.
[20:03:45] <Dark_Shikari> With a special name.
[20:04:31] <_av500_> Dark_Shikari: but do you see any actual point in it? or just on2 cargo cult?
[20:04:32] <BBB> do goldframes have the keyframe flag set?
[20:04:43] <BBB> if no, then the parsing likely resembles more a p-frame style than a i-frame style
[20:04:49] <Dark_Shikari> BBB: any frame can be a golden frame
[20:05:00] <BBB> ok
[20:05:06] <BBB> and how's that different from altref?
[20:05:12] <Dark_Shikari> altref frame can be used to emulate b-frames sorta weirdly
[20:05:17] <_av500_> altref is not shown
[20:05:32] <Compn> Dark_Shikari : are you saying something like how h263 can be formed into mpeg4asp ? like vpx can be formed into h264base ?
[20:05:38] <BBB> notshown is a flag
[20:05:44] <Dark_Shikari> Compn: it's a bit more rewriting than that.
[20:05:47] <BBB> altref is not the same as notshown, I think
[20:06:15] <_av500_> BBB: what sense does it make to decode a frame and not show it if it is not a ref?
[20:06:43] <BBB> altref is ref ?
[20:06:51] <janneg> killing decoding time
[20:07:04] <_av500_> janneg: because_we_can :)
[20:11:09] <janneg> maybe an eloberate way of maintaining constant bitrate
[20:20:57] <mru> or maybe it's just a bug that made it into the spec
[20:22:41] <_av500_> unthinkable
[20:23:23] <mru> Dark_Shikari: steve jobs or someone impersonating him reads your blog
[20:23:26] <mru> http://www.theregister.co.uk/2010/05/20/jobs_on_vp8/
[20:24:01] <Dark_Shikari> what, fakesteve?
[20:24:09] <Dark_Shikari> WAIT WHAT?
[20:24:25] <JEEBsv> haha
[20:24:29] <Dark_Shikari> WHAT WHAT WHAT?
[20:24:33] <_av500_> realsteve
[20:26:12] <drv> apparently steve reads your blog :P
[20:28:02] * _av500_ blames Dark_Shikari for aapl not supporting the open web standards
[20:28:38] <BBB> Dark_Shikari: see, now this is why you should shut up about ffmpeg's h264dec being slow and just make it faster, or claim it is faster
[20:28:46] <mru> jobs has a nice way with brevity sometimes
[20:28:49] <Dark_Shikari> should I thank him for referencing my blog post?
[20:28:51] <BBB> now the whole world will read that our h264dec is the slowest in the world
[20:29:12] <BBB> btw, lol@"third-year college student"
[20:29:23] <mru> it's true
[20:29:48] <BBB> that's not the point
[20:30:28] <mru> I though we were in the business of writing code, not selling lies
[20:34:05] <BBB> fair enough
[20:34:12] <BBB> Dark_Shikari: go write some code you third-year college student!
[20:34:32] <BBB> did anyone notice that their vpx_read_bit() is really big?
[20:35:01] <mru> what is its purpose?
[20:35:06] <mru> get_bits1()?
[20:35:49] <BBB> I think so, but it's incredibly complex
[20:35:57] <BBB> for a get_bits1(), I mean
[20:44:04] * Kovensky thinks someone should add "<@BBB> Dark_Shikari: see, now this is why you should shut up about ffmpeg's h264dec being slow and just make it faster, or claim it is faster <@BBB> now the whole world will read that our h264dec is the slowest in the world" to the quotes page
[20:44:41] * Kovensky also thinks the quotes page should be in the topic :>
[20:44:49] <BBB> that would be my deathbed
[20:45:38] <Dark_Shikari> all michael has to do is quote it in a quote of his own on the ML
[20:45:45] <Dark_Shikari> and then get some other guy to add it to the niedermayer quotes page
[20:46:01] <_av500_> nm: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[20:46:10] <_av500_> êrr, mn: ^^^^^^^^^^^^^^^^
[20:46:48] <mru> Kovensky: add it to the quotes page yourself
[20:46:59] * Kovensky doesn't know the URL :(
[20:47:18] <mru> http://wiki.multimedia.cx/index.php?title=Quotes
[20:47:30] <Kovensky> mru: add to the /topic :>
[20:47:41] <Kovensky> meh, requires account
[20:48:06] <mru> ask mike for one
[20:48:17] <Kovensky> how2request mike for one
[20:48:39] <mru> mailto:mike at multimedia cx
[20:49:18] <mru> Dark_Shikari: http://my.opera.com/haavard/blog/2010/05/20/webm-analysis
[20:50:48] <Dark_Shikari> seen it
[20:50:50] <mru> and what is it with people who don't think companies will spend $100M on junk?
[20:50:54] <Dark_Shikari> lol
[20:51:00] <Dark_Shikari> yahoo did it for years
[20:51:25] <mru> I know many companies whose entire existence depends on others buying their junk
[20:51:48] <_av500_> what a load of bs
[20:51:54] <_av500_> the article
[20:51:59] <Dark_Shikari> "raw"
[20:52:02] <Dark_Shikari> "been around for 6+ years"
[20:52:08] <mru> av500: of course it's bs
[20:52:21] <_av500_> this one: http://my.opera.com/haavard/blog/2010/05/20/webm-analysis
[20:52:22] <Kovensky> "Despite claims to the contrary, The WebM format is not set in stone. The WebM FAQ points out that there is a separate development branch for possible future improvements. " <-- didn't you say google said that the *spec* is set in stone
[20:52:41] <mru> it's an article by an opera troll relating what a xiph told a freetard mailing list
[20:52:42] <_av500_> it is practically set in stone
[20:52:53] <_av500_> as it is released with fanfare
[20:52:59] <Dark_Shikari> the good thing is that nobody cares about opera.
[20:53:01] <Kovensky> <@mru> it's an article by an opera troll <-- I also remember that last name being associated with Trolltech's founding
[20:53:40] <Kovensky> Dark_Shikari: Opera is (was?) strong in the non-apple non-microsoft mobile field
[20:53:45] <astrange> maybe you should change "similar to H.264 Baseline" to "similar to H.264 Baseline without multiple reference frames"
[20:53:53] <Dark_Shikari> astrange: but it has multiple refs
[20:54:01] <astrange> not as many
[20:54:04] <Dark_Shikari> ok, true
[20:54:08] <Dark_Shikari> but most people rarely use more than a few
[20:54:13] <Dark_Shikari> and most devices limit to only a few
[20:54:19] <_av500_> Kovensky: opera is strong on the iphone :)
[20:55:13] <mru> with that thing that renders the pages on their servers and sends only a low-res image?
[20:55:34] <mru> and still doesn't get it right
[20:55:44] <Kovensky> _av500_: it is?
[20:55:59] <Kovensky> though I never checked, people tend to stay with stock stuff if it doesn't suck too much =p
[20:56:21] <Kovensky> though people still stay with ie6, so I guess the "if it doesn't suck too much" isn't really relevant
[20:57:24] <_av500_> Kovensky: it is #1 download in itunes app store
[20:57:53] <Dark_Shikari> does safari on iphone suck that much?
[20:57:59] <_av500_> no
[20:58:04] <_av500_> having 128mb sucks
[20:58:09] <Dark_Shikari> ?
[20:58:13] <_av500_> so safari has to drop loaded pages all the time
[20:58:23] <_av500_> ipad is even worse
[20:58:25] <Dark_Shikari> and opera ... adds more memory to your phone?
[20:58:32] <mru> or uses less
[20:58:36] <_av500_> no, but seems to keep a smaller page footprint
[20:59:04] <_av500_> ppl try to load several pages to read on a flight, then check the gate with another app - pages lost...
[20:59:25] <Kovensky> <_av500_> Kovensky: it is #1 download in itunes app store <-- o, they approved it?
[20:59:32] <_av500_> Kovensky: indeed they did
[20:59:42] <Kovensky> I haven't read anything on the subject ever since they started that site with the "countup" until approval
[20:59:53] <_av500_> it got approved fairly quikcly
[20:59:54] * Kovensky has over 20 thousand unread items on google reader
[21:00:00] <ohsix> suprised they didn't technical it cuz it can run ecmascript
[21:00:21] <_av500_> ohsix: i think they feared operas eu leverage
[21:00:32] <Kovensky> IIRC opera renders pages on their server and only sends the result?
[21:00:37] <astrange> opera mini doesn't run js
[21:00:56] <_av500_> Kovensky: yep
[21:01:40] * Kovensky never had any cellphone fancier than "takes pics with a shitty 320x240 camera"
[21:01:43] <BBB> Dark_Shikari: yes safari sucks
[21:01:47] <BBB> Dark_Shikari: it's dog slow
[21:01:56] <Kovensky> and the last one I had didn't even have a camera lol
[21:02:06] <twnqx> lucky you
[21:02:12] <twnqx> those are hard to come by :X
[21:02:16] <_av500_> ...Starting today, Google is using WebM to encode all YouTube videos larger than 720p...
[21:02:18] <Kovensky> twnqx: lol
[21:02:23] <twnqx> you laugh
[21:02:27] <_av500_> >720p only?
[21:02:28] <j-b> Opera Mini is fast
[21:02:28] <mru> twnqx: the last phone I bought has no camera
[21:02:45] <twnqx> you don't work with companies who forbid cells with camera
[21:02:45] <mru> I went to a shop and asked for the cheapest phone they had
[21:03:20] <Kovensky> twnqx: o_O
[21:03:49] <Kovensky> what do you work with, chemical weapons? nuclear research? reverse engineering?
[21:04:09] <mru> they make camera phones
[21:04:12] <Kovensky> ("reverse engineering" as in taken to the level of that ben affleck movie)
[21:04:20] <mru> paycheck?
[21:04:21] <twnqx> currently the company i work for would like me to have the most modern mobile around (mobile network operator)
[21:04:24] <Kovensky> yea
[21:04:47] <mru> uma wasn't worth enduring ben in that film
[21:05:00] <Kovensky> lol
[21:05:11] <Kovensky> the movie wasn't very good but was enjoyable
[21:05:37] <twnqx> but i've worked with government agencies and ministries and a car manufacturer, as well as comapnies delivering to that one
[21:05:59] <twnqx> those are pretty sensitive. some agencies don't allow any cells in :S
[21:06:14] <bcoudurier> hi folks
[21:06:15] <twnqx> neither laptops or portable storage
[21:06:23] <twnqx> annoying working environment
[21:06:25] <mru> twnqx: some companies don't allow any cells, not even brain cells
[21:08:05] <janneg> Dark_Shikari: including psnr numbers was an error. it would have spared use the "WebM (sic!) is clearly better than H264 baseline"
[21:08:22] <Dark_Shikari> But it is
[21:08:26] <Dark_Shikari> vp8 is better than h264 baseline
[21:08:31] <Dark_Shikari> It's just that their encoder is shit
[21:11:23] * Kovensky wonders if it can be made to approach main's efficiency
[21:11:31] <Kovensky> but I guess no because of no bframes
[21:11:42] <pJok> Kovensky, ffvp8enc will tell... ;)
[21:11:45] <janneg> as long as their encoder is the only one it's not better for practical purposes
[21:11:55] <Kovensky> ffmpeg has good video encoders?
[21:12:00] <pJok> dunnno
[21:12:08] <pJok> mpeg2 seems to be working
[21:12:09] <Kovensky> afaik the only good one is -vcodec libx264 :>
[21:12:10] <pJok> ;)
[21:12:42] <mru> so we need libxvp8
[21:12:58] <mru> or maybe that's xp8
[21:12:59] <mru> not sure
[21:14:26] <BBB> call it libx268
[21:15:45] <pJok> it has an 8 in it, it must be better than 4
[21:15:50] <Kovensky> but... think of 40 years in the future!
[21:16:09] * Kovensky wonders what will ITU-T do when they run out of H.26d
[21:16:26] * pJok wonders if those who thought of 720i and 1080i were thinking of the future or just their wallets right now...
[21:16:44] <mru> 720i doesn't exist
[21:16:55] <pJok> mru, oh
[21:16:58] <pJok> good at least
[21:17:13] <BBB> Kovensky: h270 is available
[21:17:19] <BBB> or in hex, h26A
[21:17:27] <BBB> endless possibilities
[21:17:37] <Kovensky> well, h.26A could happen =p
[21:17:37] <mru> anyone remember h26L?
[21:17:43] <Kovensky> H.270 is allocated to sth else IIRC
[21:17:51] * pJok mumbles something about 60fps 1080p should have been the lowest HD broadcast resolution
[21:18:04] <_av500_> in 40 years, the survivors will not care for codecs, but for braaaains
[21:18:06] <mru> pJok: then you'd still be waiting for it
[21:18:18] <mru> _av500_: those are not the survivors
[21:18:20] <Kovensky> pJok: they'd just have progressively-encoded interlacing with frame doubling
[21:18:38] <Kovensky> LOL
[21:18:40] <_av500_> mru: yes, the survivors from in 39.9 years
[21:18:48] <pJok> Kovensky, might be, but it'd still be 60fps for when they would eventually stop doubling the frames ;)
[21:19:37] <pJok> mru, considering i have a very low def tv at home, im still waiting for HD ;)
[21:19:48] * Kovensky sees a brighter future with a working moon base
[21:20:00] <pJok> its not like tv gets any better in HD anyways
[21:20:03] <_av500_> i found base 16 ok so far
[21:20:04] * Kovensky has an analog tv at home, digital broadcast first started last week
[21:20:37] <pJok> i think there are only a few bbc programmes that deserve HD broadcast... and possibly a few movies as well
[21:20:49] <pJok> but for the rest... give me content instead of HD
[21:20:56] <_av500_> +1
[21:21:07] * pJok watches dvd's in ½ resolution anyways
[21:21:20] * _av500_ wathces stuff on 5" in the tram :)
[21:21:20] <mru> http://xkcd.com/732/
[21:21:35] * Kovensky wishes he could watch anything on commute without doing it for the last time
[21:22:26] <pJok> mru, thats exactly 100% true
[21:23:13] <mru> that said, a 1080p picture _is_ far better than SD
[21:23:17] <Kovensky> _av500_: re: survivors: <+hatefulcunt> the survivors will care for loli butts <+hatefulcunt> same as always <+hatefulcunt> Strike Witches season 36 wheeen
[21:23:39] <Kovensky> mru: if only for the "smaller" macroblocking
[21:23:54] <mru> no, I can't see the scanlines in the display from across the room
[21:23:59] <mru> that really does help
[21:24:15] <mru> even when watching SD stuff
[21:24:45] <_av500_> dpi ftw!
[21:25:22] <Kovensky> also re zombie apocalypse: http://xkcd.com/734/
[21:27:52] <CIA-7> ffmpeg: stefano * r23203 /trunk/libavcodec/eval.h: Fix doxy reference to unexisting function.
[21:28:07] <bcoudurier> <Kovensky> ffmpeg has good video encoders?
[21:28:14] <bcoudurier> is that a joke ? if it is, it is a very bad one
[21:31:56] <mru> the video encoders are certainly better than the audio ones
[21:32:00] <mru> at least some of them
[21:35:17] <_av500_> so, ffmpeg has some encoders that are better than some others...
[21:35:23] <Kovensky> lol @ "M-x butterfly" actually "working" on emacs
[21:35:48] <mru> new version?
[21:38:17] <Kovensky> http://xkcd.com/742/ hah
[21:38:19] <Kovensky> mru: 23.2.1 here
[21:43:16] <Yuvi> vp8 has a start code... 3 bytes after the beginning of the frame
[21:43:33] <_av500_> late start
[21:44:43] <janneg> anything important before the "startcode"?
[21:45:11] <Yuvi> yes, the frame type, version, and size
[21:46:35] <_av500_> so we need ungetbits(24);
[21:50:02] <mru> skip_bits(-24) ?
[21:50:20] <Yuvi> in the parser though?
[21:50:25] <_av500_> mru: good that we made the parameter int :)
[21:51:09] <mru> DonDiego: are you coming to linuxtag?
[21:51:29] <_av500_> DonDiego: and do you bring yv along?
[21:52:15] <DonDiego> i'd like to come to linuxtag
[21:52:21] <DonDiego> dunno if yvonne will join
[21:52:25] <DonDiego> will ask her tomorrow
[21:52:33] <mru> tell her we'd like her to come along
[21:52:38] <DonDiego> i will
[21:53:00] <mru> I could tell her myself I suppose...
[21:53:06] <mru> when whe comes online
[21:53:17] <DonDiego> do you have her jabber?
[21:53:23] <mru> yeah
[21:53:24] <DonDiego> yeah, invite her..
[21:54:15] <mru> so are _you_ coming?
[21:54:23] <_av500_> he must