[FFmpeg-devel-irc] IRC log for 2010-09-27

irc at mansr.com irc at mansr.com
Tue Sep 28 02:00:26 CEST 2010


[00:31:10] <CIA-63> ffmpeg: stefano * r25216 /trunk/ (libavcodec/avcodec.h libavutil/avutil.h doc/APIchanges): Bump version and update APIchanges after r25210.
[01:18:35] <CIA-63> ffmpeg: stefano * r25217 /trunk/libavfilter/avfilter.c:
[01:18:35] <CIA-63> ffmpeg: Make avfilter_destroy() code less convoluted, introducing a temporary
[01:18:35] <CIA-63> ffmpeg: variable link.
[02:42:59] <bcoudurier> hey guys
[02:43:07] <bcoudurier> wtf is happening in LA now, are we having a very late summer
[02:43:33] <Dark_Shikari> Yeah, wtf
[02:43:36] <Dark_Shikari> it was SO FUCKING HOT today
[02:43:39] <Dark_Shikari> holy christ
[02:43:44] <bcoudurier> yeah
[02:43:47] <bcoudurier> it's still hot now
[02:44:20] <Dark_Shikari> I walk outside in the morning, at 11 AM, and the instant I move into the sun, it feels as if I moved to the Judean Desert
[02:44:57] <Dark_Shikari> Oh.  No wonder it did.  It was _actually_ near 110F.
[02:46:27] <Dark_Shikari> bcoudurier: it's odd. every year we seem to have exactly one week (no matter how late) in LA with temps above 100
[02:46:30] <Dark_Shikari> then it never happens again
[02:46:38] <Dark_Shikari> same with rain.  one week with rain.  then it never happens again.
[02:47:44] <Dark_Shikari> wow.  it's still 97.
[02:48:59] <bcoudurier> yeah
[02:49:02] <bcoudurier> it's freaking hot
[02:49:14] <bcoudurier> we had a _shitty_ summer and now it's hot
[03:05:16] <Compn> heh
[03:05:22] <Compn> its LA , what are you expecting ?
[03:06:55] * Compn realizes he is 15 mins late
[04:24:24] <Dark_Shikari> does anyone have anything that can open avc-intra?
[04:26:30] <Dark_Shikari> mru: haha oh wow, I found a chinese article series about porting x264 to blackfin
[04:26:53] <Dark_Shikari> there's some actually highly interesting results
[04:26:57] <Dark_Shikari> they compared write through with writeback caching
[04:27:07] <Dark_Shikari> writethrough is 23% slower
[04:27:22] <Dark_Shikari> (this is without asm, which means it's REALLY a lot slower)
[04:44:37] <CIA-63> ffmpeg: darkshikari * r25218 /trunk/libavcodec/h264.c:
[04:44:38] <CIA-63> ffmpeg: Improve error concealment of lost frames
[04:44:38] <CIA-63> ffmpeg: If a frame is lost, replace it with data from the previous valid frame.
[04:51:44] <Dark_Shikari> bcoudurier:
[04:51:45] <Dark_Shikari> [mxf @ 0x10463290] track 0: could not find essence container ul, codec not curre
[04:51:48] <Dark_Shikari> ntly supported in container
[04:51:51] <Dark_Shikari> do we not support h264 in MXF?!?!
[04:58:42] <thresh> moroning
[05:31:08] <bcoudurier> not yet
[05:32:08] <Dark_Shikari> well, I was able to confirm that if I mux to an mp4
[05:32:11] <Dark_Shikari> x264's avc-intra works in premiere
[05:32:42] <bcoudurier> that's great
[05:33:12] <Dark_Shikari> (with my patch)
[05:46:32] <KotH> hoi zäme
[06:02:15] <bcoudurier> Dark_Shikari, did you send the patch somewhere ?
[06:03:36] <Dark_Shikari> it's an x264 patch
[06:03:40] <Dark_Shikari> it'll go in the next push
[06:04:27] <bcoudurier> I fear the MXF mapping of AVCIntra in the P2 flavor is ugly
[06:04:39] <Dark_Shikari> what's the issue?
[06:06:55] <Dark_Shikari> do you need a test file or something?
[06:09:03] <bcoudurier> It's the clip wrapping method, like pack all the frames together in one packet
[06:09:23] <bcoudurier> and the audio is in a separate file
[07:29:18] <bcoudurier> night guys
[07:29:56] * kshishkov looks outside and disagrees with Baptiste and his leaving script
[07:34:18] <pJok> god morgon kshishkov
[07:36:07] <kshishkov> goda morgnar, pJok
[07:37:16] <pJok> im surprised to see swscale slowly getting fixed
[07:38:06] <kshishkov> what's the most surprising - that it's swscale or it's mere slowly instead of continental drift speed?
[07:46:37] <mru> Dark_Shikari: of course write-through is slow
[07:46:43] <mru> every store goes all the way to ram
[07:49:55] <Dark_Shikari> mru: so it saves load bandwidth, at the cost of store bandwidth
[07:50:00] <Dark_Shikari> probably crap unless you built your application around it
[07:50:03] <mru> no it doesn't
[07:50:14] <mru> loads are unaffected
[07:51:37] <Dark_Shikari> I thought the point of writeback is that you don't automatically load cacheline X if you store to cacheline X?
[07:51:47] <Dark_Shikari> er, write-through
[07:52:01] <mru> no, that's a different attribute
[07:52:16] <mru> you can either allocate on write or not
[07:52:31] <Dark_Shikari> ah k
[07:52:34] <mru> allocate on write means if a write misses, you fetch the line and dirty the cache
[07:52:42] <Dark_Shikari> so why would write-through even be an option?
[07:52:56] <mru> it's useful in a few situations
[07:53:43] <mru> if you do frequent (clustered) reads, not too many writes, and writes need to be visible by some other device
[07:53:54] <mru> otherwise you'd need a cache flush
[07:54:18] <mru> it's also useful for debugging purposes
[08:07:14] <pJok> kshishkov, its most surprising that its swscale
[08:17:38] <pJok> hrm
[08:17:44] <pJok> i wonder when OS/2 finally dies
[08:17:55] <pJok> even the banks here are slowly switching from it
[08:19:10] <mru> never underestimate the koreans
[08:19:12] <kshishkov> well, you need to talk to North Korean leader, so he does something to the biggest OS/2 user
[08:19:35] <kshishkov> and I heard that Windows 98 is still popular in Vietnam
[08:19:57] <cartman> kshishkov: [OT] http://www.meforum.org/2753/turkey-future
[08:20:56] <pJok> mru, can't you just let them fork their own version of ffmpeg with OS/2 support?
[08:20:56] <kshishkov> cartman: I wonder why they decided to compare Turkey to Russia?
[08:21:16] <cartman> kshishkov: should be compared to Iran
[08:21:34] <mru> pJok: I don't mind as long as supporting os/2 limited to a few lines configure
[08:21:42] <mru> and they actively maintain it
[08:21:44] * kshishkov has heard some bad things about last Swedish election winners
[08:21:51] <mru> winners?
[08:21:55] <mru> there were no winners
[08:22:13] <kshishkov> if you got to Riksdag you're winner
[08:22:29] <kshishkov> (population obviously doesn't count)
[08:27:25] <pJok> kshishkov, the only winners in the swedish election are the media
[08:27:55] <pJok> gives them something to write about
[08:28:42] * kshishkov thought that's not a problem for media anyway since majority of news nowadays are generated by themselves
[08:32:34] <lu_zero> ^^;
[08:39:14] <_av500_> gm
[08:39:23] <kshishkov> hej
[08:40:09] <mru> buon giorno
[10:02:33] <CIA-63> ffmpeg: stefano * r25219 /trunk/libavutil/opt.c: Cosmetics: apply misc style fixes.
[10:28:53] <asac> hi gents! what's the status of arm/neon optimizations on vp8 codec? who should i talk to about potential contributions on this etc.?
[10:29:24] <kshishkov> I suspect it's mru, he'll be here in a minute
[10:30:10] <asac> thanks kshishkov!
[10:30:19] <asac> mru: ^^ ;)
[10:30:21] <Dark_Shikari> yes, he probably has neon on highlight
[10:34:36] <asac> mru: we (linaro) are currently planning our work for the next cycle; and one topic that was selected by our steering committee on multimedia is to optimize important software codecs for ARM. VP8 is obvious target, but I would also be interested to hear if there are other common codecs that need some opts for ARM/NEON etc. and that you would like to see contributions for.
[10:35:09] <CIA-63> ffmpeg: stefano * r25220 /trunk/libavutil/opt.c:
[10:35:09] <CIA-63> ffmpeg: Simplify include, since opt.c is now in libavutil there is no need
[10:35:09] <CIA-63> ffmpeg: to use the library path prefix for eval.h.
[10:35:36] <mru> asac: there are vp8 neon patches in the queue already
[10:35:42] <mru> I just haven't had time to take care of them
[10:36:59] <spaam> mru: dont slack ;P
[10:37:09] <mru> that's exactly what I'm not doing
[10:37:27] <mru> I'm simply prioritising paying clients
[10:37:30] <Dark_Shikari> asac: decoders, encoders, or both?
[10:37:44] <mru> if linaro were to send some cash my way...
[10:43:06] <asac> Dark_Shikari: i think both ... decoders first, but encoders for camera are also important i guess
[10:43:32] <asac> mru: we dont have cash. only engineers
[10:43:39] <mru> who pays them?
[10:43:42] <Dark_Shikari> x264 has a bunch of NEON functions missing
[10:43:43] <asac> at least atm ;)
[10:43:53] <Dark_Shikari> x264 is the only video encoder that's optimized for ARM at all
[10:44:00] <asac> mru: we are an engineering foundation
[10:44:01] <Dark_Shikari> nobody else seems to have tried
[10:44:17] <Dark_Shikari> we're missing a few functions, as Yuvi wasn't able to write every last one during his GSOC
[10:44:28] <asac> thanks Dark_Shikari
[10:44:30] <Dark_Shikari> probably at least 20-30% speed could be gotten, plus we could do some larger-scale optimizations
[10:44:32] <mru> Dark_Shikari: I'm available for hire
[10:44:37] <Dark_Shikari> to compensate for things like the cortex a8's crappy memory
[10:44:38] <asac> so x264 encoding
[10:44:48] * asac notes down
[10:44:49] <Dark_Shikari> mru: We don't have boatloads of money, and if we did, we'd spend it on really important things first
[10:44:52] <Dark_Shikari> like 10-bit support in lavc
[10:44:56] <mru> asac: but these engineers of yours, someone surely pays them, right?
[10:45:01] <Dark_Shikari> (part of our evil plan to swap the entire world over to 10-bit h264)
[10:45:16] <mru> Dark_Shikari: I don't care who pays me
[10:45:23] <cartman> isn't h.264 decoding more important anyway?
[10:45:26] <asac> mru: of course ;). but they wont get fired. we want to help out with code first
[10:45:53] * mru is still struggling to understand what linaro is
[10:45:56] <asac> http://www.linaro.org/commercial-sponsors/
[10:46:04] <asac> http://www.linaro.org/about-linaro/
[10:46:09] <mru> heck, I haven't even figured out what they're pretending to be yet
[10:46:15] <Dark_Shikari> cartman: we already have h264 decoding asm
[10:46:16] <Dark_Shikari> shitloads of it
[10:46:17] <asac> https://wiki.linaro.org/Releases/1105/TechnicalTopics
[10:46:23] <cartman> Dark_Shikari: need moarrrr
[10:46:26] <mru> I've read the website
[10:46:32] <mru> that's not what I'm talking about
[10:46:34] <asac> mru: what do you want to know?
[10:46:42] <mru> I want to understand
[10:46:46] <mru> there's a difference
[10:46:58] <asac> sure. i am here to answer questions then ;)
[10:47:35] <asac> in short: we want to bring the linux arm software stack to a shape that its ready to compete with intel ;)
[10:47:46] <Dark_Shikari> Then you _really_ have a lot to do with x264 ;)
[10:47:47] <Dark_Shikari> lol
[10:47:52] <mru> it's the "we" part that's a bit fuzzy
[10:48:02] * Dark_Shikari mutters about intel still not sending us a sandy bridge
[10:48:29] <asac> mru: ARM is basically the main drivers ... with all the partners seeing that this is importantt to do and hence sending engineers to work on this
[10:48:31] <mru> Dark_Shikari: people like to do other things than encode h264, you know
[10:49:02] <mru> asac: well, I'm currently discussing possible paid work with ARM
[10:49:03] <Dark_Shikari> well, there are 5 things people care about
[10:49:09] <Dark_Shikari> audio encoding, decoding, video encoding, decoding, and colorspace conversion
[10:49:13] <Dark_Shikari> (mostly)(
[10:49:19] <Dark_Shikari> video decoding (h264) is well-optimized, at least asm-wise
[10:49:32] <Dark_Shikari> audio decoding is pretty good iirc, with the exception of float/int conversion stuff which is bikeshed
[10:49:41] <Dark_Shikari> audio encoding isn't very optimized, and for encoding there's only x264
[10:49:44] <Dark_Shikari> *video encoding
[10:49:49] <mru> aac-he needs optimising
[10:49:58] <Dark_Shikari> aac-he is a gigantic mess
[10:50:06] <mru> that too
[10:50:49] <asac> aac-he -> encoding is a mess? or both?
[10:51:49] <mru> asac: so which company do you work for?
[10:52:22] <asac> mru: i am working for canonical ... we are there as all the partners are competitors, so we provide the balance and keep things rolling ;)
[10:52:37] <mru> yada yada yada
[10:53:27] <asac> but i am 100% linaro ;) ... in the long run it shouldnt matter much where you originate from ;)
[10:56:41] <mru> right now it matters a little
[10:57:02] <cartman> it'll cost a leg and an ARM :P
[10:57:11] <asac> haha
[10:57:59] <mru> all else aside, I do appreciate you coming directly to us
[10:58:05] <mru> instead of hoarding patches in ubuntu
[10:58:09] <asac> heh
[10:58:13] <asac> linaro != ubuntu
[10:58:27] <mru> only ~=
[10:58:30] <arpu> hello can someone help me i wnat split/cut a avi file with x264 codec (libavformat)
[10:58:43] <arpu> i get strange artefakts
[10:59:13] <asac> mru: linaro works upstream only ... unless upstream release cycle is really long ...which is when we backport and deliver to distributions like meego, ubuntu, etc.
[10:59:46] <asac> if you are working on ARM you should try our toolchain ;)
[10:59:54] <mru> I have
[10:59:57] <mru> it compiles ffmpeg
[11:00:11] <asac> cool. any feedback? compared to fsf/codesourcery ?
[11:00:12] <mru> and kudos for making it build without patching
[11:00:31] <cartman> asac: why do you need just another toolchain?
[11:00:41] <mru> cartman: it's yet another gcc fork
[11:02:24] <cartman> mru: gcc doesn't merge useful arm stuff?
[11:02:24] <asac> cartman: its not another toolchain. its fsf toolchain. we put stuff upstream and backport so the arm ecosystem gets a well optimized toolchain from stable branch
[11:02:25] <superdump> the messy part of he aac is the signalling
[11:02:25] <asac> so partly doing what CS did ... just that we bake the toolchain in full distros and hence we think our quality is quite high!
[11:02:26] <Dark_Shikari> "bake the toolchain in full distros"
[11:02:26] <superdump> plus it's a lot of code and complexity for the gains it gives
[11:02:46] <Dark_Shikari> superdump: I think he means decoding
[11:02:58] <superdump> i'm talking about decoding
[11:03:35] <Dark_Shikari> "the hard part is the signalling"
[11:03:43] <Dark_Shikari> but signalling is done by the encoder isn't it?
[11:04:05] <superdump> ok, writing code to handle all the different signalling cases
[11:04:12] <cartman> asac: I see, thanks
[11:04:46] <superdump> and you can't always know from the outset whether sbr is used or so
[11:04:52] <asac> cartman: welcome
[11:04:54] <superdump> which is a pita
[12:28:42] <CIA-63> ffmpeg: thardin * r25221 /trunk/libavformat/mpegtsenc.c: mpegtsenc: Write subtitle extradata if set
[12:28:43] <CIA-63> ffmpeg: thardin * r25222 /trunk/libavformat/mpegtsenc.c: mpegtsenc: Indent
[13:33:12] <kshishkov> pross-au: g'day, mate. One of our chaps has found 5.1 Bink audio, it seems
[13:34:00] <pross-au> Holy crap :D
[13:34:08] <DonDiego> astrange: i hear an approved patch of yours is waiting to be applied..
[13:34:24] <Dark_Shikari> _what_
[13:34:24] <Dark_Shikari> 5.1?
[13:34:24] <Dark_Shikari> lol
[13:34:33] <Dark_Shikari> that sounds like something bink invented on the request of a single customer
[13:34:34] <pross-au> Bink is proving its durability
[13:35:20] <pross-au> Dark_Shikari: ffmpeg needs to support that for completeness
[13:39:07] <kshishkov> pross-au: < kierank> kshishkov: have you seen 5.1 variant of bink audio?
[13:39:29] <kshishkov> pross-au: < kierank> header is 1FCB
[13:40:26] <pross-au> Must be recent titles
[13:40:48] <pross-au> Let me punch that into Bing.com
[13:40:53] <kshishkov> speaking of which, you still haven't completed BINKb decoder yet
[13:41:33] <pross-au> Yeah, i logged ontonight specifically to post the update wtv demuxer
[13:42:10] <pross-au> apparently bink does 5.1 and 7.1 now
[13:42:44] <kshishkov> test video decoder while you're on it
[13:43:25] <pross-au> Wilco
[13:43:50] <pross-au> Im liking the wmvp work too
[13:43:55] <pross-au> Nice job
[13:44:01] <kshishkov> it's just a hack
[13:44:34] <kshishkov> real decoder should decode those pictures to separate planes and perform all that zoom/rotation/fading effects
[13:45:13] <pross-au> id rather see a hack then a standalone 'wmvp2ffmpeg' type project
[13:46:53] <kshishkov> you can tease Dark_Shikari with wvp2 decoder though
[14:04:05] <pross-au> 'ello kierank
[14:10:52] <kierank> hello pross-au
[14:12:01] <kierank> looks like steam uses libavutil...
[14:12:08] <Dark_Shikari> _what_?
[14:12:27] <Dark_Shikari> whoa.  you're right
[14:12:28] <Dark_Shikari> steam uses avcodec
[14:12:33] <Dark_Shikari> wtf does steam use avcodec for?
[14:12:35] <_av500_> and vapor?
[14:12:43] <kierank> they just use the utils it seems
[14:12:49] <Dark_Shikari> do they abide by lgpl?
[14:13:12] <Dark_Shikari> "third party legal notices" I'm guessing this has the LGPL in it
[14:13:15] <Dark_Shikari> and informationa bout sources
[14:13:31] <_av500_> they might want to switch to libavcore, no?
[14:13:36] <Dark_Shikari> amusing they use it only for utils
[14:13:54] <kierank> avcodec etc is in the /bin directory
[14:14:09] <kierank> it's because they use chromium as a backend
[14:14:15] <Dark_Shikari> aha
[14:14:19] <Dark_Shikari> >chromium embedded framework
[14:14:32] <mru> sounds scary
[14:14:48] <kierank> pross-au: I had a new bink variant from Civ V but I deleted it by accident.
[14:14:56] <Dark_Shikari> "For FFMPeg, part of Chromium, part of CEF:"
[14:15:01] <Dark_Shikari> lol they mispelled ffmpeg
[14:15:08] <Dark_Shikari> and spelled it differently in different parts of the document
[14:15:30] <Dark_Shikari> but looks fine otherwise =p
[14:15:44] <Dark_Shikari> Interesting that they're now using chromium for the web browser.
[14:15:45] <kshishkov> pross-au: and Dark_Shikari claimed to play Civ V, so extort Bink sample from him
[14:15:56] <kierank> what about the reverse engineering clauses?
[14:16:34] <Dark_Shikari> picking nits is stupid, go after the real violators
[14:17:34] <DonDiego> kierank: do they have them?
[14:17:45] <Dark_Shikari> >civ5_opening_movie_en_US.wmv
[14:17:49] <Dark_Shikari> bink
[14:17:50] <Dark_Shikari> lol right
[14:17:55] <Dark_Shikari> bink my ass
[14:18:04] <kierank> there's a bink audio file in there somewhere
[14:18:17] <kierank> the file extension is .binka
[14:18:19] * kshishkov claims supporting Bink in WMV then
[14:18:37] <Dark_Shikari> confirmed, found it
[14:18:46] <Dark_Shikari> OpeningMenu-51mix.binka
[14:18:59] <_av500_> .binka sounds like some swedish non-alcohol drink that kshishkov might like
[14:19:15] <_av500_> i had a nice bottle of Binka...
[14:19:31] <Dark_Shikari> OpeningMenu-51mix.binka: Invalid data found when processing input
[14:19:33] <kierank> kshishkov: you're missing out by not drinking alcohol
[14:20:43] <mru> we found bink beer in brussels
[14:21:50] <kshishkov> _av500_: just don't ask who likes Trocadero on this channel
[14:22:13] * pJok should actually go buy a trocadero
[14:22:26] <kshishkov> pJok: why?
[14:23:33] <pJok> kshishkov, never tried it
[14:24:26] <Dark_Shikari> kshishkov: http://www.mediafire.com/?6endqr45j3nbox5
[14:24:37] <Dark_Shikari> now I'm going to sleep, when I wake up, I expect my ffmpeg to be able to play this back!
[14:24:43] <Dark_Shikari> get moving ;)
[14:25:22] <kshishkov> Dark_Shikari: it's for pross-au then, not me
[14:25:28] <Dark_Shikari> lol
[14:25:30] <merbzt> ahh trocadero
[14:25:30] * Dark_Shikari sleeps
[14:26:15] <pross-au> miles\win32\binkawin.asi
[14:26:36] <pross-au> there are no other bik/bink files in civ5
[14:27:27] <BBB___> j-b: you are our biggest isv, how can you claim not to be part of our community?
[14:28:10] <j-b> BBB___: technically, I am part of the ffmpeg community, not ffmpeg-dev community
[14:28:33] <j-b> and as I was answering on ffmpeg-dev@ not ffmpeg-user@ :D
[14:29:07] <BBB___> I'll buy you beer anyway :)
[14:29:21] <BBB___> you'll be our honorable ffmpeg-user attendee
[14:29:28] <j-b> :)
[14:39:03] <cartman> BBB___: got a guestion for you
[14:39:07] <cartman> hmmmpf what was that
[14:39:35] <astrange> DonDiego: i'll apply it today
[14:39:41] <cartman> ah yes
[14:39:48] <cartman> BBB___: did you ever figure out a solution for http://blogs.gnome.org/rbultje/2009/03/04/firefox-gmail-and-content-type-of-attachments/ ?
[14:40:25] <cartman> BBB___: in my case Chrome/Safari misbehaves too
[14:40:48] <DonDiego> astrange: good :)
[14:43:02] <BBB___> cartman: no :(
[14:43:06] <cartman> BBB___: :-/
[14:45:08] <kshishkov> BBB: where is xvp8?
[14:45:22] <BBB> kshishkov: I knew you'd beat me to it! congrats btw
[14:45:25] <BBB> where's wvp2?
[14:45:30] <BBB> and non-i-frames?
[14:52:59] <kshishkov> BBB: where is I-only xvp8 at least?
[15:17:08] <CIA-63> ffmpeg: michael * r25223 /trunk/libavcodec/flicvideo.c:
[15:17:08] <CIA-63> ffmpeg: Fix several security issues in flicvideo.c
[15:17:08] <CIA-63> ffmpeg: This fixes CVE-2010-3429
[16:11:22] <CIA-63> ffmpeg: mstorsjo * r25224 /trunk/libavformat/rtpproto.c: rtpproto: Use a sockaddr_storage instead of a sockaddr_in with recvfrom
[16:16:55] <BBB> mru: see? I can write inline asm
[16:17:05] <BBB> (kind of a waste of my time, but ohwell)
[16:17:48] <BBB> the reason it's faster is likely because gcc knows better which registers to use and how many (x86-64 has many more regs available then x86-32, and I don't take advantage of that to make the code more similar)
[16:24:40] <CIA-63> ffmpeg: stefano * r25225 /trunk/ (8 files in 2 dirs): Move av_get_token() from libavfilter to libavutil.
[16:25:23] <mru> how often is lack of core registers a problem?
[16:27:35] <mru> wow, what an ungrateful itard
[16:28:21] <KotH> itard?
[16:28:34] <mru> iTard
[16:28:38] <KotH> oh..
[16:28:42] <KotH> -devel?
[16:28:46] <mru> !
[16:28:59] <CIA-63> ffmpeg: stefano * r25226 /trunk/doc/APIchanges: Add APIchanges entry after r25225.
[16:30:28] * KotH wonders, whether people will understand those one character communications when they discover the irc logs in 1000 years, burried under a pile of uncompiling code, written in an ancient programming language nobody has ever heard of
[16:35:08] * cartman trolls mru 
[16:35:14] <cartman> you unhelpful!!111!!
[16:36:21] <DonDiego> ah
[16:36:27] * DonDiego hadn't flamed in a while
[16:40:40] <KotH> if that is a flame, then a burned down candle is a fire storm ;)
[16:40:57] <cartman> KotH: you were just too damn nice
[16:41:12] <KotH> DonDiego: you should definitly learn to insult people by being polite :)
[16:41:51] <KotH> cartman: read again until you spot the insult ;)
[16:41:58] <cartman> :D
[16:42:42] <KotH> mhmmm.. prinzenrolle...
[16:47:26] <BBB> "now go away" is nothing more but a kind request to follow advice
[16:47:29] <BBB> it's not a flame
[16:47:37] <BBB> DonDiego: and you can do so much better than that :-p
[16:49:38] <KotH> hmm... this guy is ...offensive
[16:50:09] * KotH would kick him from the ml for that, if it wasnt against his ethos
[16:53:24] <BBB> JFDI
[16:53:40] <BBB> \o/ >30 cycles off another h264 function
[16:53:48] <BBB> I might at some point feel like I'm doing something useful
[16:55:01] <KotH> BBB: no, it wouldnt be ok morally to kick someone for so little offense
[16:55:37] <KotH> it might not look like it, but even a BofH has moral
[16:59:50] <CIA-63> ffmpeg: stefano * r25227 /trunk/libavfilter/ (avfilter.c avfilter.h):
[16:59:50] <CIA-63> ffmpeg: Make AVFilterLink store the pointers to the source and destination
[16:59:50] <CIA-63> ffmpeg: pads, rather than their index.
[16:59:57] <twnqx> BBB: come back once my atom can play 1080p. in software.
[17:00:27] <cartman> twnqx: you can do that with dxva
[17:00:42] <twnqx> not the "in software" part.
[17:01:01] <twnqx> i can do it with vdpau as well :P
[17:01:07] <cartman> :D
[17:01:37] * cartman wonder if its even possible
[17:03:50] <CIA-63> ffmpeg: stefano * r25228 /trunk/doc/APIchanges: Add APIchanges entry after r25227.
[17:05:00] <DonDiego> KotH, BBB, lol
[17:05:28] <KotH> DonDiego: =
[17:05:30] <KotH> DonDiego: ?
[17:05:44] <DonDiego> lol @ your comments, nothing mor
[17:05:45] <DonDiego> e
[17:14:32] <Compn> btw
[17:14:46] <Compn> i think we spend more time telling people they are on the wrong list than just answering said dumb questions :P
[17:17:06] <KotH> Compn: you know, give a man fire and he has warm for a day, set him on fire and he has warm for the rest of his life
[17:19:00] <KotH> Compn: you're cheap!
[17:19:22] * KotH doesnt get up for anything less than 100CHF/h.. and that's for simple and fun work
[17:22:42] <BBB> this guy is hilarious
[17:24:16] * cartman is too cheap it seems
[17:28:59] <DonDiego> sshhh
[17:29:15] <DonDiego> tread carefully now, mozilla is watching :)
[17:29:45] <CIA-63> ffmpeg: stefano * r25229 /trunk/libavfilter/avfilter.c:
[17:29:45] <CIA-63> ffmpeg: Make avfilter_insert_filter() propagate an error code in case the
[17:29:45] <CIA-63> ffmpeg: called avfilter_link() fails.
[17:31:35] <kierank> DonDiego: why are you scared of mozilla?
[17:31:52] <kierank> I am not scared of mozilla
[17:31:54] <mru> kierank: mozilla is scared of us
[17:32:37] <DonDiego> some mozilla person entered the channel and i wished to extend a greeting :)
[17:33:01] <DonDiego> jrmuizel: welcome
[17:34:55] <cartman> I eat mozerella in the morning
[17:35:28] <spaam> good for you
[17:37:23] <jrmuizel> DonDiego: thanks
[17:39:05] <mru> hmm, he did stop top-posting
[17:39:20] <mru> now if he'd also break his long lines
[17:43:04] <Compn> diego getting trolled by the lawrence guy
[17:44:14] <Compn> jrmuizel : anything in particular bring you here, or are you just along for the ride?
[17:44:20] <mru> I find it amusing when people ask a question only to demand a different answer than the they're given
[17:47:05] <jrmuizel> Compn: nope, nothing in particular
[17:49:08] <BBB> is icc payware?
[17:49:17] <BBB> I can't find a free icc download on Intel's website
[17:50:22] <cartman> BBB: try google
[17:51:03] * Compn googles around
[17:51:07] <cartman> BBB: http://software.intel.com/en-us/articles/non-commercial-software-download/
[17:51:27] <Compn> BBB : http://software.intel.com/en-us/articles/intel-software-evaluation-center/
[17:51:34] <Compn> damn, cartman beats me
[17:51:35] <cartman> ;)
[17:52:29] <BBB> I don't use linux guys, how often do I have to keep fanboying apple?
[17:52:51] <cartman> BBB: payware
[17:52:52] <mru> no icc for you then
[17:52:57] <cartman> torrentware rather
[17:53:00] <BBB> nah
[17:53:02] <mru> catfree for non-commercial
[17:53:03] <BBB> I just won't test it
[17:53:06] * cartman got it all
[17:53:16] <BBB> it's not like I care
[17:53:16] <mru> *cartman: free ..
[17:53:20] <BBB> I was just trying to be nice
[17:53:27] <kierank> BBB: virtualbox
[17:53:31] <cartman> free as in torrentware
[17:53:44] <kierank> "you wouldn't download a car"
[17:53:44] <mru> no
[17:53:53] <cartman> kierank: I would if I could damn it
[17:53:59] <iive> I would, if I could.
[17:54:07] <mru> intel will give you a non-commercial license for free
[17:54:39] <Compn> i would download two, just to have a backup :P
[17:54:54] <BBB> download a car
[17:54:57] <BBB> that's like dreamland
[17:55:03] <iive> Compn: you can download one, then make as many copies as you want :P
[17:55:23] <Compn> oh right!
[17:55:32] <iive> BBB: give the 3d printer some time to improve :)
[17:55:38] <mru> BBB: for some cars you can download firmware to boost the horsepowers
[17:55:45] <mru> and void warranty
[17:56:21] <Compn> dont you like your car crippled mru ?
[17:56:26] <BBB> bsod
[17:56:52] <mru> Compn: I don't have a car at all in fact
[17:57:03] * elenril wonders if rms will stop using cars now
[17:57:16] <mru> here's hoping
[17:57:20] * Compn wishes he could go without a car
[17:57:21] <kierank> he'll find some piddly car like a reliant robin
[17:57:30] <Compn> driving = dangerous :(
[17:57:40] <kierank> not as dangerous as riding a motorcycle
[17:57:54] <mru> I ride a bike in traffic
[17:57:58] <mru> not a motorbike
[17:58:35] <kierank> I hate cycling when there are massive lorries about
[17:58:39] <Compn> dangerous too, since people who drive cars have some animosity towards bicycles
[17:59:04] <mru> there is a simple trick to cycling in traffic
[17:59:17] <Compn> swing a crowbar around ?
[17:59:21] <mru> keep in the centre of the lane
[17:59:33] <mru> if you start creeping along the edges, it gets dangerous
[18:00:16] <kierank> and hire a car to carry your belongings behind you like MPs do
[18:01:20] <cartman> mru: intel gives free non-commercial licenses for !linux version ?
[18:01:39] <mru> cartman: what's wrong with the linux version?
[18:01:52] <cartman> mru: nothing
[18:02:01] <cartman> if you can get it to run Mac
[18:02:51] <Compn> why is intel trying to sell a compiler ?
[18:03:02] <Compn> i guess it has optimizations that gcc doesnt have hmm
[18:04:47] <BBB> most performance-relevant parts of ffmpeg are written in asm already
[18:04:53] <BBB> so we wouldn't really care?
[18:04:57] <BBB> (I think)
[18:33:46] <pJok> i dont think a compiler would be able to compete with hand optimized assembly code
[18:34:20] <mru> pJok: I _know_ it can't
[18:34:25] <mru> and so do my clients
[18:35:50] <pJok> mru, well, you do optimization for a living, don't you?
[18:37:06] <mru> pJok: yes
[18:37:28] <pJok> last time i used assembler, it was on a 6510... and it really wasn't that much
[18:40:09] <pJok> but at least if you know what you are doing, in any care, you are better than gcc's random code generator
[18:43:01] <BBB> I would hope so...
[18:43:07] <BBB> gcc is a little funny
[19:24:19] <bcoudurier> hey guys\
[19:24:47] <saintdev> hey bcoudurier's onjoin
[19:58:17] <Dark_Shikari> first open source AVC-Intra 50 encoder going in today
[19:58:20] <Dark_Shikari> now we need a decoder.
[19:59:37] <kshishkov> mru: http://c0389161.cdn.cloudfiles.rackspacecloud.com/dyn/str_strip/337170.full.gif
[19:59:41] <Dark_Shikari> I think we should sponsor a 10-bit project for ffh264
[20:00:28] <CIA-63> ffmpeg: janne * r25230 /trunk/ffmpeg.c:
[20:00:28] <CIA-63> ffmpeg: use strict_std_complience from avcodec_opts[AVMEDIA_TYPE_*]
[20:00:28] <CIA-63> ffmpeg: Options are not yet set for video encoders and for codec specific
[20:00:28] <CIA-63> ffmpeg: defaults it is desireable to set the codec options only after the codec
[20:00:28] <CIA-63> ffmpeg: is known.
[20:08:36] <cartman> is janne on irc?
[20:08:56] <mru> jannau: ^^
[20:09:02] <cartman> jannau: Did you mean: compliance  ?
[20:09:07] <cartman> typo there :-)
[20:09:18] <mru> cartman: tab is your friend
[20:09:29] <cartman> mru: /me hugs tab
[20:10:09] * mru hopes for cartman's sake it's not a bar tab he's hugging
[20:10:30] <cartman> :P
[20:12:10] <jannau> cartman: fortunately only in the commit message
[20:12:27] <cartman> jannau: oh thats not a big damage then :-)
[20:53:53] <bcoudurier> shit so hot in LA today
[21:00:08] <j-b> shit so cold in Paris today
[21:01:51] <cartman> Istanbul was rainy
[21:02:27] <mru> fair weather here today
[21:02:48] <mru> autumn chill definitely setting in though
[21:03:58] <Tjoppen> frost nights here
[21:04:11] <mru> but you're in the fucking arctic
[21:04:25] <Tjoppen> no water at work today. temporary water hose froze
[21:04:40] <mru> where do you live again?
[21:04:44] <Tjoppen> umeå
[21:05:14] <Tjoppen> they're digging up half the roads and messed up our water pipes. no welder on a week
[21:05:22] <Tjoppen> *in
[21:08:56] <lu_zero> brr
[21:10:17] <lu_zero> here is chilly as well
[21:10:36] * mru prefers chilli
[21:10:46] <spaam> Tjoppen: any snow yet? D:
[21:12:25] <Tjoppen> spaam: not yet
[21:14:39] <Tjoppen> time for dexter
[21:25:04] <j0sh> it is HOT here
[21:25:27] * j0sh just spent 2 hours basking by the pool in 105 degree weather
[21:25:40] <mru> where's "here" for you?
[21:25:57] <j0sh> LA
[21:26:10] <j0sh> well, a bit south (irvine)
[21:26:17] <mru> know that place
[21:26:21] <j0sh> cool
[21:26:32] <mru> I find that area mostly boring
[21:26:48] <j0sh> haha irvine is the least exciting place on earth
[21:27:05] <mru> costa mesa is no better
[21:27:15] <j0sh> s/irvine/orange county
[21:27:24] <mru> quite
[21:27:27] <j0sh> but its a nice area for the most part
[21:27:31] <mru> so why do you live there?
[21:27:33] <j0sh> and you cant beat the weather
[21:27:43] <j0sh> because i live with my parents
[21:27:49] <mru> fair enough
[21:28:07] <spaam> j0sh: aww
[21:28:13] * mru can't quite imagine living with parents anymore
[21:28:40] <j0sh> i just graduated, spent 6 years at school in NY... it is nice to be back home :)
[21:28:40] <spaam> mru: better to stay with one of your sisters? ;)
[21:29:10] <mru> haven't done that either for a very long time
[21:29:24] <mru> I left the damn country 7 years ago
[21:29:25] <jannau> is swscale covered by make fate at all? make fate-list is not conclusive
[21:29:38] <spaam> mru: damn? pfff  its nice ;)
[21:29:44] <KotH> eh.. i got an appology out of that itard
[21:30:12] <mru> spaam: if you like nanny states, sure
[21:30:38] <mru> though that wasn't the reason I left
[21:30:48] <mru> it was just how things turned out
[21:30:51] <mru> KotH: ?
[21:31:41] <KotH> mru: i wrote him a lengthy, not very friendly mail explaining him what he just did
[21:31:43] <CIA-63> ffmpeg: bcoudurier * r25231 /trunk/libavcodec/dvdata.c:
[21:31:43] <CIA-63> ffmpeg: In dv decoder, use the expected aspect ratio logic, like other decoders do,
[21:31:43] <CIA-63> ffmpeg: following width/height = par/dar.
[21:31:45] <KotH> mru: and he appologized
[21:32:08] <mru> KotH: and you kept the fun to yourself, shame on you
[21:32:27] <CIA-63> ffmpeg: bcoudurier * r25232 /trunk/libavcodec/dv.c: In dv decoder, set sample aspect aspect ratio, fix issue #1612
[21:32:28] <mru> jannau: some tests should invoke swscale
[21:32:52] <bcoudurier> j0sh, indeed
[21:33:00] <mru> e.g. dnxhd tests
[21:33:01] <bcoudurier> I'm in santa monica, so hot here too
[21:33:29] <KotH> mru: sometimes, it's better to write a pmail than to trampel on other peoples soul in public
[21:33:46] <bcoudurier> mru, yeah same here, I've left home 10 years ago now
[21:34:16] <mru> KotH: he deserved a public flogging
[21:34:53] <jannau> ok, the port to AV_CPU_FLAG_* seems to be not totally broken
[21:34:55] <spaam> KotH: tell us about it
[21:34:57] <KotH> mru: sure, he deserved it. but wouldnt have acheived what i was aiming at
[21:35:06] <BBB> lol @ itard
[21:35:06] <KotH> spaam: why should i tell you?
[21:35:20] <mru> you already did
[21:35:21] <KotH> spaam: you're not even a contributor
[21:35:45] <mru> contributory trolling...
[21:35:50] <KotH> lol
[21:36:20] <spaam> KotH: we are best friends , you and i <3
[21:36:44] <KotH> mru: one kick for calling me best friend please
[21:37:18] <j0sh> bcoudurier: is it as hot there as it is here
[21:37:25] <KotH> thanks!
[21:37:37] <mru> ok, now let's keep this channel sane
[21:37:50] <mru> save the kick wars for that other place
[21:37:52] <KotH> mru: was this channel ever sane?
[21:38:00] <mru> compared to come others, very
[21:38:04] <mru> *some
[21:38:08] <KotH> hehe :)
[21:38:09] <spaam> KotH: <3
[21:38:24] <KotH> anways.. i should be already sleeping for 2h...
[21:38:28] <jannau> bah, vsynth1-flashsv is broken and uses swscale. it wasn't earlier
[21:38:29] <KotH> good night boys
[21:38:36] <spaam> KotH: good night <3
[21:38:39] <jannau> goodnight KotH
[21:38:58] <mru> jannau: that uses rgb conversion
[21:39:15] <mru> how does it fail?
[21:42:39] <jannau> +stddev:   61.85 PSNR: 12.30 MAXDIFF:  156 bytes:  7603200/  7603200
[21:45:18] <mru> nothing else?
[21:46:01] <mru> md5 unchanged?
[21:48:13] <jannau> http://pastie.org/1185213
[21:48:37] <mru> so it all changed
[21:48:41] <mru> means encoding broke
[21:48:57] <CIA-63> ffmpeg: aurel * r25233 /trunk/libavfilter/vf_yadif.c: merge #if with if()
[21:49:12] <jannau> it's lukily only the last commit which implements runtime cpu detection with --eanble-runtime-cpudetect
[21:49:46] <mru> flashsv encodes rgb, so it seems the yuv to rgb conversion is broken
[21:51:21] <CIA-63> ffmpeg: aurel * r25234 /trunk/libavfilter/ (Makefile x86/Makefile x86/yadif.c): use a Makefile in x86 subdir
[21:54:20] <jannau> it doesn't like auto detected cpu flags in sws_setColorspaceDetails()
[21:55:10] <Dark_Shikari> Los Angeles, California,  has broken the all-time record high temperature of  112F with a temperature of 113F at 12:15pm PDT.  Their temperature could rise more through the afternoon.
[21:55:15] <Dark_Shikari> ffffffffffffffff
[21:56:03] <lu_zero> in celsius?
[21:56:05] <Dark_Shikari> 45
[21:57:18] <lu_zero> warm enough to kill somebody weak
[21:57:27] <mru> solution: don't be weak
[21:58:03] <lu_zero> or don't be stupid
[21:58:12] <CIA-63> ffmpeg: stefano * r25235 /trunk/libavfilter/ (avfilter.c avfilter.h):
[21:58:12] <CIA-63> ffmpeg: Use more expressive names for the avfilter_insert_filter() in and out
[21:58:12] <CIA-63> ffmpeg: parameters.
[22:01:30] <jannau> found it, hopefully just inconsistant cpu flags between init and the conversion
[22:10:47] <CIA-63> ffmpeg: stefano * r25236 /trunk/ (6 files in 2 dirs): Move av_set_options_string() from libavfilter to libavutil.
[22:11:32] <CIA-63> ffmpeg: mru * r25237 /trunk/configure: Fix out of tree builds with vf_yadif and mmx
[22:14:07] <CIA-63> ffmpeg: stefano * r25238 /trunk/doc/APIchanges: Update APIchanges after r25236.
[22:18:49] <CIA-63> ffmpeg: cehoyos * r25239 /trunk/ (libavcodec/avcodec.h libavformat/utils.c libavcodec/parser.c):
[22:18:49] <CIA-63> ffmpeg: Fix index_entries pos:
[22:18:49] <CIA-63> ffmpeg: It was being set wrong for files with data_offset > 0
[22:18:49] <CIA-63> ffmpeg: Patch by Michael Chinen, mchinen gmail
[22:25:06] <twnqx> does ffmpeg support ADX (playstation2 game audio)?
[22:25:51] <mru> it supports something it calls adpcm_adx
[22:26:05] <mru> described as "SEGA CRI ADX ADPCM"
[22:26:06] <twnqx> mh
[22:26:18] <mru> ffmpeg -codecs | grep adx
[22:26:46] <twnqx> but this 2MB fragment shouldn't really play pretty long, even with 24khz stereo in pcm...
[22:29:42] <twnqx> sound_00000_00008.adx: Invalid data found when processing input
[22:29:43] <twnqx> hm
[22:30:41] <mru> saintdev: ping
[22:30:48] <saintdev> mru: sup?
[22:30:58] <mru> you've been poking at the aac encoder, right?
[22:31:16] <saintdev> very carefully, it bites
[22:31:23] <mru> yeah...
[22:31:47] <mru> how much work would you say remains to make it usable?
[22:33:28] <saintdev> there's a few bugs in channel handling somewhere that need fixed
[22:35:05] <saintdev> quantization needs some work, right now on the samples i'm using it ends up using far too many bits for the requested bitrate
[22:35:45] <saintdev> that's about all i know of that would be necessary to make it _usable_ in the strict sense of the word
[22:36:31] <mru> if someone who knows aac were to work on it, how long before it would compete with faac in quality? compete with good encoers?
[22:36:37] <mru> weeks, monts, years?
[22:36:44] <saintdev> of course there could be other things peloverde may know of that i don't
[22:37:00] <mru> but he's not here...
[22:37:08] <saintdev> yeah :/
[22:38:58] <saintdev> to compete with faac, i would probably say a month.
[22:39:34] <saintdev> although i'm not familiar with most of the code. really i've only dealt with the psymodel in any detail.
[22:42:26] <saintdev> speaking of, i should probably look into the channel issues while i'm waiting to hear something from gabriel
[22:52:34] <mru> fate is looking sad...
[22:53:02] <saintdev> poor fate, would some ice cream cheer it up?
[22:58:17] <Dark_Shikari> gg yadif patch
[22:58:59] <lu_zero> Dark_Shikari: is it usable for you now?
[23:02:12] <jannau> mru: is there a way to guess the libswscale revision? the date is UTC?
[23:02:26] <mru> the last commit broke it
[23:10:25] <Dark_Shikari> lu_zero: it doesn't compile
[23:11:09] <Dark_Shikari> https://roundup.ffmpeg.org/issue2250
[23:11:09] <Dark_Shikari> lol
[23:11:11] <Dark_Shikari> I broke seeking
[23:11:18] <Dark_Shikari> why the fuck is mplayer trying to seek to non-keyframes
[23:13:35] <twnqx> mhhh
[23:13:51] <twnqx> what else is there to try on mpeg-ps other than ffmpeg, mplayer and mediainfo?
[23:17:05] <j0sh> jannau: spelling of michael's name
[23:17:08] <j0sh> http://git.jannau.net/git/FFmpeg.git.convert/tree/merge_FFmpeg_libswscale_git.sh?id=HEAD#n171
[23:28:50] <mru> bcoudurier: ping
[23:28:50] <mru> bcoudurier: your swscale commit broke something
[23:31:45] <bcoudurier> rah
[23:48:00] <bcoudurier> mru, fixed
[23:51:38] <lu_zero> bleah installing Xcode takes as much as compiling a gentoo base system... what the hell are they doing!?
[23:51:51] <lu_zero> </rant>


More information about the FFmpeg-devel-irc mailing list