[FFmpeg-devel-irc] IRC log for 2010-03-20

irc at mansr.com irc at mansr.com
Sun Mar 21 01:00:52 CET 2010


[00:10:20] <CIA-92> ffmpeg: stefano * r22602 /trunk/doc/ffserver-doc.texi:
[00:10:20] <CIA-92> ffmpeg: List the FFserver specific options under a "Main options" section, to
[00:10:20] <CIA-92> ffmpeg: distinguish them from the generic common options.
[00:53:02] <CIA-92> ffmpeg: michael * r22603 /trunk/libavcodec/error_resilience.c: Error concealment of h264 with multiple references.
[01:03:32] <CIA-92> ffmpeg: mru * r22604 /trunk/ffmpeg.c: 10l: fix build without termios.h
[09:19:34] <DonDiego> moin
[09:19:44] <kshishkov> god morgon
[09:20:05] <DonDiego> can anybody confirm libfaad2 being around twice as fast as ffmpeg when decoding AAC SBR?
[09:20:08] <DonDiego> peloverde: you?
[09:20:51] <ohsix> does faad not ignore sbr?
[09:22:28] <kshishkov> DonDiego: don't you have samples for that?
[09:24:51] <_av500_> faad does sbr iirc
[09:25:29] <kshishkov> it has
[09:29:09] <elenril> ffaac is ~2x faster than faad for me
[09:29:47] <kshishkov> probably SBR decoding is not optimized well enough
[09:30:47] * elenril wonders why do he get  DECAUDIO: AUDIO: 24000 Hz, 2 ch, s16le, 128.0 kbit/16.67% (ratio: 16000->96000) with ffaac
[09:31:05] <elenril> and DECAUDIO: AUDIO: 48000 Hz, 2 ch, s16le, 128.0 kbit/8.33% (ratio: 16000->192000) with faad
[09:32:12] <kshishkov> because it does not report doubled sampling rate with SBR
[09:32:30] <pJok> god morgon kshishkov :)
[09:32:32] <elenril> oh
[09:32:37] <kshishkov> goda morgnar
[09:33:17] <kshishkov> yet what does mplayer -benchmark -ac ffaac -ao null thatfile says at the end?
[09:36:21] <elenril> hmm, with ffaac audio plays at half speed
[09:36:38] <kshishkov> try with ffplay
[09:36:47] <kshishkov> (aka future MPlayer)
[09:37:04] <elenril> yesh, ffplay sounds almost normally
[09:37:17] <elenril> but still somewhat slower
[09:37:30] <elenril> probably because ffplay fails at vfr or something
[09:38:00] * elenril wonders why does it fail with mplayer
[09:40:15] <elenril> DonDiego: anyway except for that ffaac is much faster
[09:48:18] * elenril wonders how to force a decoder with ffmpeg
[09:48:38] <Dark_Shikari> -vcodec X -i whatever
[09:48:43] <Dark_Shikari> or -acodec X -i whatever
[09:49:43] <elenril> thanks
[09:50:48] <elenril> yeah, same results with ffmpeg - ffaac >> faad
[09:50:53] <elenril> all hail peloverde
[09:51:44] <kshishkov> indeed
[10:01:07] <Kovensky> <+elenril> all hail peloverde <-- did it get committed? :o
[10:02:20] <elenril> ofc
[10:02:27] <kshishkov> yes, otherwise there'd be no talk about SBR support in ffaac
[10:02:29] <elenril> a few week ago
[10:02:33] <elenril> +s
[10:03:16] <Kovensky> lol
[10:04:33] <Kovensky> <@verm__> jfs: i remembered the reason why i designed the mru interface the way i did <-- itc: an interface to mru
[10:09:19] <DonDiego> kshishkov: -ao nulll will not work
[10:09:25] <DonDiego> it plays at normal speed
[10:09:52] * elenril used -ao pcm:fast:file=/dev/null
[10:10:10] <kshishkov> whatever
[10:11:44] <DonDiego> elenril: i don't get half speed, sample?
[10:14:35] <DonDiego> http://samples.ffmpeg.org/V-codecs/h264/cathedral-beta2-400extra-crop-avc.mp4
[10:14:39] <DonDiego> take this sample
[10:14:46] <DonDiego> mplayer dumps audio faster with faad
[10:18:02] <elenril> oh, happens only with internal mkv demuxer
[10:18:11] <elenril> works fine with lavf
[10:33:11] <elenril> DonDiego: http://pastebin.ca/1846509
[11:23:40] <DonDiego> elenril: you should add -quiet
[11:24:29] <DonDiego> http://pastebin.ca/1846539
[11:24:35] <DonDiego> updated with my numbers alongside it..
[11:27:25] <DonDiego> maybe it's because i'm running on my trusty old K6-III
[11:27:35] <DonDiego> which lacks some mmx2/sse asm?
[11:27:55] <elenril> yeah, probably
[11:28:03] <elenril> i'm on core duo
[11:28:27] <DonDiego> i'll never see the day when i can get rid of libfaad :-(
[11:28:32] * DonDiego cries his eyes out
[11:29:32] <Dark_Shikari> DonDiego: simple solution
[11:29:33] <Dark_Shikari> get rid of the k6-3
[11:29:38] <Dark_Shikari> then nobody will ever have to care about such shit again
[11:34:33] <saste> DonDiego: where is the nut-container.org web site stuff stored?
[11:34:47] <saste> I'm browsing http://nut-container.org/
[11:35:06] <saste> the link to the spec points to a websvn page which is non-existing
[11:36:12] <saste> I also tried to look for the websvn of mplayer on the Mplayer site with no luck
[11:39:36] <saste> there is a mplayer git repo somewhere?
[11:40:01] <elenril> no unless you count uau's fork
[11:41:00] <DonDiego> saste: you can check out the svn repository
[11:41:06] <saste> I was asking because that would allow to do like in http://git.ffmpeg.org/
[11:41:18] <DonDiego> complain to attila and mans
[11:41:32] <DonDiego> i did not take down the web frontend to svn
[11:41:36] <saste> DonDiego: I know but we should fix the dangling link in nut-container.org
[11:41:44] <DonDiego> and i'm still annoyed it's no longer available
[11:42:03] <saste> if we can't point to the websvn page then we should directly put the static page somewhere
[11:42:07] <jai> saste: this was reported multiple times, noone cares
[11:42:12] <DonDiego> Dark_Shikari: i don't have money to replace my perfectly working computer
[11:42:46] <DonDiego> saste: if you care, upload a copy of the spec to the website directly
[11:43:03] <saste> DonDiego: yes I'll try to do that
[11:43:35] <Dark_Shikari> DonDiego: then we can start a donation fund
[11:43:36] <Dark_Shikari> for you and michael
[11:43:39] <saste> that's why I was asking where I can find the nut-container web site stuff ;-)
[11:43:43] <Dark_Shikari> for $200 we can get you something modern
[11:44:18] <DonDiego> IIRC: svn://svn.ffmpeg.org/nut
[11:44:28] <jai> saste: you could maintain a clone of the repo on one of the git hosting services and link to that
[11:44:38] <saste> DonDiego: thanks
[11:44:42] <jai> i dont think websvn is coming back in the near future
[11:44:58] <saste> jai: yes that's possible just I still don't have experience with that
[11:44:59] <DonDiego> Dark_Shikari: i kinda like my old computer
[11:45:14] <DonDiego> and i think there is value in somebody running an old machine
[11:45:15] <saste> jaj: also I'd like to see ffmpeg migrate to git soon or later
[11:45:28] <DonDiego> not everybody lives in the 1st world, you know..
[11:45:31] <Dark_Shikari> DonDiego: people running old machines gets us stuff like ffmpeg's asm
[11:45:39] <Dark_Shikari> where most of the asm is ancient mmx
[11:45:41] <Dark_Shikari> i.e. swscale
[11:45:46] <Dark_Shikari> because devs can't be bothered to get new computers
[11:45:46] <jai> saste: ah, well dont hold your breath :)
[11:45:46] <DonDiego> no
[11:46:06] <DonDiego> that's because the asm was written for the machines available back then
[11:46:13] <Dark_Shikari> no, that's because michael still has a pentium dual
[11:46:16] <Dark_Shikari> >_>
[11:46:37] <DonDiego> no, mn hasn't written asm for libswscale in many years
[11:46:40] <Dark_Shikari> exactly
[11:46:45] <Dark_Shikari> because it's good enough for his computer
[11:46:50] <DonDiego> i think mans has an old computer lying around
[11:46:54] <Dark_Shikari> but if he got a new computer, he would go improve it
[11:46:55] <DonDiego> donate that to michael
[11:47:00] <Dark_Shikari> I can say from experience that I do not write much asm for anything that doesn't work on my computer
[11:47:05] <Dark_Shikari> or even things that do, but won't run on it
[11:47:10] <Dark_Shikari> e.g. I almost never optimize for phenom
[11:47:15] <Dark_Shikari> because I don't have any phenom boxes, not even on ssh
[11:47:26] <Dark_Shikari> pengvado is the same
[11:47:30] <DonDiego> that's why i say we need devs with old computers
[11:47:35] <Dark_Shikari> and with a sample size of 2, I generalize to everyone
[11:47:49] <Dark_Shikari> "having an old computer" != "not having a new computer"
[11:48:05] <DonDiego> the stuff you actually use counts
[11:48:12] <DonDiego> i can compile on faster boxes nowadays
[11:48:19] <Dark_Shikari> no, it's what you have
[11:48:22] <DonDiego> though admittedly i don't always do it
[11:48:24] <Dark_Shikari> I optimized for nehalem for a year before I got one
[11:48:27] <Dark_Shikari> because I had one on ssh
[11:48:37] <Dark_Shikari> I optimized for a penryn for over a year and I have yet to ever get one
[11:48:58] <Dark_Shikari> anyways ffmpeg devs should not be slowed down by ancient hardware
[11:49:02] <Dark_Shikari> and "3rd world" is no excuse
[11:49:10] <DonDiego> and i make sure stuff continues to run decently on old x86 boxes
[11:49:11] <jai> well, patches are welcome
[11:49:19] <Dark_Shikari> I have a friend who lives in estonia, their median income is 10x worse than the US
[11:49:20] <jai> so people with newer cpus could write the asm
[11:49:23] <Dark_Shikari> and their internet is faster than us.
[11:49:44] <DonDiego> that's besides the point
[11:50:00] <DonDiego> the USA lives in the network dark ages
[11:50:19] <Dark_Shikari> furthermore, someone who spends a large portion of his life on a computer can afford to spend $200 on a new one ;)
[11:50:23] <DonDiego> still there are many many people that cannot afford modern computers
[11:50:32] <DonDiego> they use their old stuff or get them second hand
[11:50:43] <Dark_Shikari> "second-hand computers"are now athlon 64s and athlon xps
[11:50:50] <Dark_Shikari> and you can find them in dumpsters
[11:50:52] <DonDiego> where you live, yes
[11:51:04] <Dark_Shikari> Then getting one to where you live is a shipping fee away.
[11:51:22] <Dark_Shikari> We had an entire cart full of pentium 4s at this year's dorm auction
[11:51:25] <Dark_Shikari> none even got bid on
[11:51:27] <Dark_Shikari> nobody even wanted them
[11:51:29] <Dark_Shikari> not even for $1
[11:51:42] <Dark_Shikari> they weren't even worth chopping up for spare parts.
[11:51:43] <DonDiego> that processor sucks..
[11:51:55] <Dark_Shikari> having a k6-3 is just not excusable.
[11:52:13] <DonDiego> anyway, do you have a point to make?
[11:52:29] <Dark_Shikari> ffmpeg devs should be treated better.
[11:52:32] <Dark_Shikari> that's my point.
[11:52:39] <Dark_Shikari> and they should treat themselves better too.
[11:53:00] <Dark_Shikari> you know that if you asked for money for an upgrade, you'd get it.
[11:53:02] <pross-au> shucks, my main comps are P4 (dothan) and G4
[11:53:15] <Dark_Shikari> still much better than a K6-3  :>
[11:53:29] <DonDiego> i'll think about it
[11:53:29] <pross-au> My K6-2 PSU died.
[11:53:45] <DonDiego> i just upgraded my disks in any case ;)
[11:54:01] <DonDiego> and now i have around 100GB total capacity..
[11:54:02] <pross-au> Otherwise its operational :D
[11:54:09] <jai> well, gsoc is on and DonDiego could get enough for a nehalem :)
[11:54:20] <jai> he did say once that he wanted to participate
[11:54:28] <Dark_Shikari> 100gb? lol
[11:54:43] <DonDiego> i might try to get a used g5, they look interesting
[11:54:45] <Dark_Shikari> I don't even have enough usb slots for all my hard disks, I could send you a 500gb
[11:54:46] <pross-au> Cloud computing means that you dont need a powerful desktop comp
[11:55:08] <DonDiego> ok, i need to get some work done
[11:55:09] <DonDiego> bbl
[11:59:46] * kshishkov is glad he skipped all that discussion
[12:03:32] <Dark_Shikari> blegh, I've spent the past hour reading papers on intra mode decision in h264
[12:03:35] <twnqx> <Dark_Shikari> having a k6-3 is just not excusable. <- where should i send my spare athlon xp 1800+?
[12:03:36] <Dark_Shikari> it's staggering how bad these things are
[12:03:55] <Dark_Shikari> twnqx: michael or diego? =p  but seriously, I'd want to make sure michael has a nehalem
[12:07:45] <KotH> osoi-yo!
[12:07:51] <KotH> janneg: fogir is idle ;)
[12:08:07] <kshishkov> Dark_Shikari: make somebody send him a notebook with it
[12:08:36] * KotH would volunteer to bring it to him personaly
[12:09:00] <kshishkov> indeed
[12:09:20] <twnqx> why a nehalem?
[12:09:27] <twnqx> isn't that some P4?
[12:09:38] <Dark_Shikari> i7
[12:10:46] <twnqx> that one i'll keep.
[12:13:55] <Compn> Dark_Shikari : write something up on x264 project page 'in an effort to speed up all projects using ffmpeg, we would like to donate i7's to all ffmpeg asm devels. please pony up some cash for this'
[12:14:21] <twnqx> lol
[12:16:36] <pross-au> Atoms
[12:17:07] <pross-au> i3s
[12:18:04] <kshishkov> pross-au: Atom was good for me and allowed compiling FFmpeg for BeagleBoard. Too bad its packaging was shite. Strewth!
[12:18:09] <Compn> kshishkov : someone on -dev-eng just mentioned vp6 interlaced ...
[12:18:23] <kshishkov> Compn: and?
[12:18:51] <Compn> oops, i guess that is aurel's code
[12:18:56] <Compn> nm
[12:19:37] <pross-au> Just pointing out that not every intel owner is an i7 owner
[12:20:24] * kshishkov has 80i186
[12:20:31] <saste> DonDiego: I need the nut specification pages published somewhere on the web so that I can create a link to them in nut/homepage/index.html
[12:21:00] <saste> DonDiego: hopefully all the nut/docs dir should be accessible somewhere
[12:22:32] <saste> DonDiego: there is no hurry i'll be away from my computer till tomorrow afternoon
[12:27:00] <jai> well, thats why i said fork a copy onto guthub or sth :)
[12:32:00] <saste> jai: well I'll explore that option i can't find a more lazy solution ;-)
[12:45:17] <Dark_Shikari> hmm.  some of these papers are actually pretty cool.  an IP core for an intra analysis engine that takes 3.5 milliwatts for 1080p
[12:48:07] <KotH> what papers?
[12:48:18] <twnqx> D_S: what is actually done during ratecontrol lookahead, just calculate a metric for picture complexity?
[12:48:23] <Dark_Shikari> was just randomely browsing shit on google scholar
[12:48:29] <Dark_Shikari> twnqx: oh ho ho, quite a bit more complicated than that
[12:48:32] <Dark_Shikari> frametype decision
[12:48:34] <Dark_Shikari> macroblock-tree
[12:49:04] <Dark_Shikari> which includes up to (--bframes + 3) motion searches per frame
[12:49:14] <Dark_Shikari> and by motion search I mean a search for all blocks in that frame
[12:49:16] <Dark_Shikari> and intra analysis
[12:49:21] <twnqx> wow, so already the motion vector search is there?
[12:49:24] <Dark_Shikari> yes
[12:49:27] <Dark_Shikari> it's not the main search
[12:49:28] <twnqx> i see
[12:49:30] <Dark_Shikari> it's a simplified search
[12:49:34] <Dark_Shikari> downscaled by 2 in each direction
[12:49:54] <Dark_Shikari> no multiref, no partitions, SAD only, final destination
[12:50:23] <twnqx> so only refined later. makes sense
[12:50:35] <Dark_Shikari> it isn't refined
[12:50:36] <Dark_Shikari> it's thrown away
[12:50:42] <Dark_Shikari> the lookahead data is only used for lookahead
[12:50:53] <KotH> hmm.. can someone tell me what "folk psychology" vs "folk physics" is?
[12:50:53] <Dark_Shikari> the only thing re-used is that a few of the MVs get used as predictors later.
[12:51:49] <kshishkov> KotH: system of beliefs not based on facts covering different areas of "knowledge"
[12:53:00] <KotH> hmm.. and what has that to do with child development? ^^'
[12:53:41] <kshishkov> read Pratchett - "lies to children" is a good way to explaining them something in completely wrong way
[12:57:36] <twnqx> what the FUCK
[12:57:42] <twnqx> i went to seagate.com
[12:57:52] <kshishkov> BTW, does this mean you've finally decided to take responsibility and raise one of your children?
[12:57:53] <twnqx> and the thing... selected my language as ru-ru :S
[12:57:54] <KotH> kshishkov: the wiki entry?
[12:58:13] <KotH> kshishkov: nah.. just reading something on authism and genetics
[12:59:35] <kshishkov> twnqx: http://xkcd.com/713/
[13:07:37] <twnqx> heh
[13:16:26] * KotH wonders whether there are any differential studies of brain activity between vegetarians and carnivors
[13:17:52] <superdump> ?
[13:17:54] <superdump> why?
[13:18:06] <kshishkov> for me it setlearn biology
[13:18:15] <kshishkov> s/.*set//
[13:18:56] <kshishkov> food (among other things) defines man
[13:19:20] <Dark_Shikari> so stop drinking so much soda.
[13:20:06] <kshishkov> it's water or tea for me
[13:20:13] <Dark_Shikari> good.
[13:20:37] <kshishkov> it's mostly because everything else is bad
[13:21:46] <KotH> superdump: oh.. just because glutamat is one of the most important neurotransmitters, which is a protein. so if vegetarians have less intake of proteins, they'll have less glutamat which should result in an impared brain function
[13:21:58] <Dark_Shikari> since when do vegetarians have less protein?
[13:22:15] <Dark_Shikari> I'd be more worried about less fat and too much carbs
[13:22:16] <Dark_Shikari> but protein is easy
[13:22:19] <Dark_Shikari> just eat lots of nuts and soy
[13:22:21] <KotH> ok.. depends on your definition of vegetarian :)
[13:22:33] <Dark_Shikari> and also, "vegan" vs "vegeterian"
[13:22:35] <Dark_Shikari> vegetarians eat cheese
[13:22:36] <kshishkov> KotH: not mentioning unique aminoacids of animal origin
[13:22:44] <Dark_Shikari> vegans don't
[13:22:47] <superdump> i eat lots of cheese
[13:22:54] <superdump> and beans
[13:23:10] <superdump> and more recently nuts
[13:23:26] <kshishkov> you should've tried eating nuts earlier
[13:23:41] * elenril thinks nut needs more attention
[13:23:42] * elenril runs
[13:23:53] <Dark_Shikari> cashews, btw, are amazing
[13:24:09] <Dark_Shikari> like 70% fat, 35% protein, -5% carbs
[13:24:16] * KotH curses wikipedia
[13:24:40] <kshishkov> Dark_Shikari: nothing beats macadamia
[13:24:45] <KotH> i haven't even finished the first page of the paper i wanted to read... but several wp articles instead ^^'
[13:25:11] <Dark_Shikari> oh wow.  that's pretty impressive
[13:25:22] <Dark_Shikari> 74g fat, 9.2g protein, 7.9g carbs, per 100g.
[13:33:49] * justlooking_ always thought that if god wanted you to be a "vegan" or "vegeterian" then he would have not invented Appendicitis, and he/she would have made everyone green and be able to use Photosynthesis as a backup.
[13:36:47] * kshishkov lives in a region where most holy food is considered treif
[13:37:50] <CIA-30> ffmpeg: jbr * r22605 /trunk/ (12 files in 3 dirs):
[13:37:50] <CIA-30> ffmpeg: Add VorbisComment writing to FLAC files.
[13:37:50] <CIA-30> ffmpeg: Patch by James Darnley <james darnley at gmail>.
[13:37:52] <kshishkov> so IMNSHO just eat what you want
[13:38:02] <elenril> \o/
[13:39:37] <superdump> kshishkov: what's treif
[13:39:40] <superdump> ?
[13:39:49] <kshishkov> non-kosher
[13:39:59] <kshishkov> in Judaism
[13:40:10] <superdump> heh
[13:40:13] <CIA-30> ffmpeg: jbr * r22606 /trunk/libavformat/ (Makefile oggenc.c):
[13:40:13] <CIA-30> ffmpeg: Add VorbisComment writing to Ogg/FLAC and Ogg/Speex files.
[13:40:13] <CIA-30> ffmpeg: Patch by James Darnley <james darnley at gmail>
[13:40:54] <elenril> \o/ ^2
[13:41:32] * superdump always thought that people had the freedom to choose who they wanted to be and how they wanted to live so there wasn't any dictation that one must eat animals and as one doesn't need to, one can choose not to and be healthy
[13:42:16] <kshishkov> superdump: exactly, but then came those vagens, Greenpeace activists and other loonies
[13:42:19] <superdump> maybe eating animals was the simplest way for human kind to develop to the point where we could understand that
[13:42:21] <kshishkov> *vegans
[13:43:07] <kshishkov> it just quite effective food
[13:43:30] <kshishkov> and happens in larger quantities too
[13:43:37] <CIA-30> ffmpeg: jbr * r22607 /trunk/ (Changelog libavformat/avformat.h):
[13:43:37] <CIA-30> ffmpeg: Add Changelog entry and bump lavf micro version for the addition of
[13:43:37] <CIA-30> ffmpeg: VorbisComment writing for FLAC, Ogg/FLAC, and Ogg/Speex files.
[13:45:39] <kshishkov> just like certain activists must not limit our choice of codecs to use to the family mantioned in last commit message
[13:47:48] <J_Darnley> elenril: there you go, all applied now
[13:48:06] <elenril> cool
[13:48:46] <J_Darnley> Now I just need to sort out comments for ogg-vorbis files
[13:49:26] <J_Darnley> I've got a very hacky patch that crashes when using libvorbis
[13:49:29] <elenril> wait, metadata format is different for each codec?
[13:49:50] <J_Darnley> Some what
[13:49:56] <Kovensky> <+elenril> wait, metadata format is different for each codec? <-- surprised?
[13:50:15] <J_Darnley> The problem is a combination of ogg and ffmpeg
[13:50:48] <J_Darnley> _My_ problem is
[13:51:41] <kshishkov> don't use ogg then ;)
[13:52:11] <elenril> av500: what's up with writing asf chapters?
[13:54:37] <DonDiego> kshishkov: btw, you were looking for something to RE, i have a suggestion: QDM2
[13:54:44] <DonDiego> it's not quite finished
[13:54:54] <DonDiego> there is a mode that does not decode exactly
[13:55:01] <DonDiego> there are noticeable artifacts
[13:55:39] <DonDiego> the sound is somewhat metallic
[13:55:41] <DonDiego> http://samples.ffmpeg.org/A-codecs/QDM2/reason_300.mov
[13:55:52] <DonDiego> i think this is one of the samples that shows the artifacts
[14:01:37] <BBB> hi spyfeng
[14:01:46] <BBB> good work on the mms tree, I'll look at the state tomorrow
[14:02:01] <BBB> spyfeng: if you're confident and it works, feel free to submit a first version for more general review already
[14:02:03] <twnqx> http://img534.imageshack.us/img534/8765/bbbx.png that's what BBBs do to your comps :S
[14:02:08] <BBB> it'd be good to get it up to speed
[14:02:14] <DonDiego> http://a1852.g.akamai.net/5/1852/184/3e1e2226/1a1a1a25940a73d84b79ea5ed40263d343af1442a51481b01d8c026bd04372d847bfee56ca3eaedd50b51688eb53bf2082b11778ea57ca36ab1f92/farmsluts_300.mov
[14:02:15] <spyfeng> hi BBB
[14:02:35] <BBB> holy shit 700% cpu
[14:02:42] <DonDiego> kshishkov: direct link to a failing sample
[14:02:52] <BBB> DonDiego: farmsluts?
[14:02:55] <BBB> suggestive....
[14:03:00] <spyfeng> I'll review it for myself and simply the code for serveral times :)
[14:03:09] <BBB> spyfeng: that's a good idea also
[14:03:17] <BBB> spyfeng: did you submit your soc proposal yet?
[14:03:18] <kshishkov> DonDiego: I'll put it to box
[14:03:19] <spyfeng> then submit it a first version asap
[14:03:20] <BBB> or isn't it open?
[14:03:25] <spyfeng> no
[14:03:29] <BBB> oh
[14:03:37] <BBB> oh I guess it opens monday
[14:03:43] <spyfeng> it doesn't open yet.
[14:04:21] <DonDiego> BBB: go watch the movie
[14:04:37] <DonDiego> it's a very funny short about a guy who gets a link to some special pr0n at work
[14:04:58] <DonDiego> his life takes a downward turn after that, very black humor :)
[14:05:06] <DonDiego> kshishkov: put it to box?
[14:05:23] <kshishkov> yep, "filed"
[14:05:58] <DonDiego> are you interested in working on this?
[14:06:07] <kshishkov> not particularly
[14:06:26] <kshishkov> but I may work on it in future
[14:08:03] <DonDiego> it would be appreciated
[14:08:11] <DonDiego> those incomplete codecs are a PITA
[14:08:16] <DonDiego> bbl
[14:08:31] * kshishkov "maintains" another one - WMV3
[14:08:51] <DonDiego> yeah, but i guess qdm2 would be simpler to fix
[14:09:03] <DonDiego> there's also qdmc waiting to be REed
[14:10:28] <kshishkov> a lot of other stuff too
[14:18:16] <BBB> spyfeng: at the top, MMSCSPacketType, SC_PACKET_.._TYPE, you can rename them all to SC_PKT_... (without TYPE)
[14:18:19] <BBB> saves another couple of bytes
[14:26:09] <twnqx> hm
[14:26:35] <twnqx> i have 8 80mm fans, spinning at ~10krpm, next to me
[14:26:59] <twnqx> rendering is noisy.
[14:27:34] <twnqx> and power consumption jumped from 400W to 600W for this box :S
[14:27:37] <kshishkov> and I thought that 3D Studio MAX 1.1 was a CPU/RAm-hungry beast
[14:28:08] <kshishkov> that's more than total comsumption for all my boxes
[14:28:43] <justlooking> perhaps yu should do some case modding and get more effective airflow so you can remove 6 of them...
[14:28:58] <twnqx> mh.... no.
[14:29:37] <twnqx> i don't mod my 2U server chassis :P
[14:30:38] <twnqx> anyway, time to change rooms
[14:30:41] <twnqx> football time \o/
[14:33:31] <KotH> .o0(it isnt summer yet, why is twnqx sitting in the server room?)
[14:33:52] <twnqx> because i pulled my servers from the data center yesterday
[14:33:55] <twnqx> to fix them.
[14:34:02] <twnqx> gonna return them tomorrow
[14:34:46] <twnqx> hm
[14:34:53] <twnqx> actually, this i7 is not doing anywthing...
[14:48:32] <janneg> KotH: the problem of manual scheduling
[14:50:41] * janneg has not enough trust in the queue systems to run any of them over the internet
[15:15:03] <justlooking> "trust in the queue systems" what systems ? and why not write your own if thats the case anyway? use something like this rebol script as you base for instance http://itmmetelko.com/blog/2009/07/03/rebols-actor-net-used-for-real-distributed-system/  http://musiclessonz.com/rebol_tutorial.html
[15:15:55] <kshishkov> I don't like that language name. Reminds me of 60s horrors
[15:16:58] <kshishkov> and syntax looks like tcl/tk
[15:17:25] <justlooking> i dont like green eggs and ham, so im not going to try it....
[15:18:23] * kshishkov does not like virtually everything, so don't mind
[15:23:20] <justlooking> well Carl Sassenrath was the Amiga lead OS dev so he knows a thing or two about small light productive codeing, try it if you like cross plafrom scripting perhaps, http://www.rebol.com/oneliners.html
[15:23:49] <justlooking> platform
[15:23:57] <kshishkov> we have _Perl_
[15:25:12] <mru> shell script and perl is all we need
[15:25:37] <mru> shell script obviously including sed and awk
[15:25:52] <kshishkov> can be replaced with Perl in case of trouble
[15:28:51] <justlooking> you have a perl script for that BBB scheduling job ? and you can do it in a few pages of script and a 600Kbyte exe, so why doesnt janneg trust it then ?,
[15:30:39] * BBB grmbls
[15:38:40] <twnqx> lol
[15:38:57] * twnqx pats the bunny
[15:39:16] <twnqx> now stop eating my ram and CPUs :P
[15:54:42] <kshishkov> twnqx: that's because you have them in abundance. My <= 1 GB RAM boxes are good to fend off any Blender
[15:54:54] <twnqx> :P
[16:04:23] <twnqx> <APTX> int i; [&]() { [&]() { i; }; }; - C++ almost like perl now :)
[16:04:41] <twnqx> yey for lambda expressions in c++ :P
[16:05:00] <kshishkov> we also have _Perl_ for that
[16:05:17] <twnqx> but but... native!
[16:05:25] <mru> wtf does that do?
[16:05:59] <twnqx> what *will* it do, once the next spec is throug and implemented
[16:06:06] * KotH thinks that perl is easier to understand
[16:06:33] <mru> fine, wtf will that do?
[16:06:35] <kshishkov> another reason to forget C++
[16:07:26] <twnqx> dunno. APTX says it does nothing, but it will be valid code :P
[16:08:01] <kshishkov> of course
[16:08:16] <kshishkov> and what if it was i++ inside?
[16:08:41] <twnqx> i never understood lambda functions.
[16:08:58] <kshishkov> but it's easy
[16:10:22] <mru> in lisp a lambda function is simply an unnamed function typically passed as an argument to something expecting a function reference
[16:10:35] <mru> and no, "reference" isn't correct lisp parlance
[16:12:24] <kshishkov> Java took it further - you create whole unnamed class for single function as an argument
[16:13:44] <mru> ugh, don't remind me
[16:13:50] <mru> the syntax is horrible too
[16:13:58] <kshishkov> of course
[16:14:17] <kshishkov> "everything is an object except for atomic types" was a wonderful idea
[16:19:47] <twnqx> janneg: are there actually scenes you can render with 6G ram? 10&12 need >6 each
[16:35:03] <justlooking> <kshishkov> "everything is an object except for atomic types" was a wonderful idea < you like that object idea but wont even try rebol http://www.rebol.com/docs/core23/rebolcore-10.html "REBOL stands for relative expression based _object language_ ." http://www.rebol.com/what-rebol.html , twnqx i take it people didnt try setting up sliced  rendering so you dont need all this ram in a single pc for the hard scenes then!
[16:38:52] <kshishkov> justlooking: there was a thing invented lately called "sarcasm"
[16:41:03] * justlooking cant find the US patent on that ;) Ohh "sarcasm" nm
[16:49:31] <peloverde> DonDiego, ffaac is about 25% faster than faad for streams with SBR
[16:49:35] <peloverde> for me
[16:50:36] <kshishkov> peloverde: as we found out, it's slower for him since lavc does not have 3DNow! SIMD speedups for it
[16:50:46] <peloverde> ahhh
[16:50:57] <peloverde> I was going to say that it prbably is a little slow on non SSE systems
[16:52:46] <CIA-30> ffmpeg: michael * r22608 /trunk/libavcodec/error_resilience.c:
[16:52:46] <CIA-30> ffmpeg: Hotfix for regtest-error failure
[16:52:46] <CIA-30> ffmpeg: a proper fix appears to require changing h263/mpeg2/motion est to index ref_index like
[16:52:46] <CIA-30> ffmpeg: h264. Ill do this change and then revert this hotfix but it will take a bit time.
[17:17:17] <siretart`> DonDiego: ping
[17:23:52] <Kovensky> <@mru> the syntax is horrible too <-- new JButton().addActionListener(new ActionListener() { public void actionPerformed(ActionEvent evt) {} });
[17:24:01] <Kovensky> why do I still remember this D:
[17:24:52] <DonDiego> siretart`: pong
[17:25:32] <DonDiego> peloverde: my numbers indicate that that faad is about twice as fast as ffaac with sbr on non-sse systems :(
[17:25:34] <mfg> is make test failing for anyone else?
[17:25:37] <DonDiego> what's the exact reason?
[17:25:42] <mfg> I just updated, and it fails on seek tests
[17:26:16] <siretart`> DonDiego: hey there. what's the status? :-)
[17:26:19] <mru> make test passes here
[17:26:22] <mfg> hmmm
[17:26:30] <DonDiego> siretart`: 0.5.2? 0.6?
[17:26:31] <kshishkov> Kovensky: the same reason as I - it's so utter crap that you do it once and remember forever. Java GUI code sucks even more than plain Java
[17:26:40] <kierank> java gui itself also sucks
[17:26:45] <siretart`> DonDiego: 0.6. I didn't really start working on the remaining patches
[17:26:47] <siretart`> for 0.6.2
[17:26:51] <siretart`> for 0.5.2
[17:26:59] <DonDiego> Yuvi: what's the status? :)
[17:27:05] <peloverde> DonDiego, A new filterbank would help other than that I'd probably need to see some profiling data
[17:27:14] <Kovensky> <@kshishkov> Kovensky: the same reason as I - it's so utter crap that you do it once and remember forever. Java GUI code sucks even more than plain Java <-- I guess
[17:27:24] <DonDiego> peloverde: what do you mean by new filterbank?
[17:27:42] <DonDiego> can i (easily) provide some profiling to you?
[17:27:44] <Kovensky> Swing is also the only GUI toolkit where I can remember the basic classes and functions without needing a reference >_>
[17:28:03] <peloverde> The SBR analysis/synthesis filterbank could probably be done better
[17:28:29] <peloverde> as far as profiling goes, just make sure the top level functions in ff_sbr_apply() don't get inlined
[17:29:33] <DonDiego> speak to me in layman's terms please, i think i've profiled once or twice in my life :)
[17:29:47] <DonDiego> (running oprofile under guidance or similar)
[17:32:41] <janneg> twnqx: probably half of all currently rendered frames are rendered with 6G or less
[17:32:42] <peloverde> I don't know how to get oprofile to work either
[17:32:57] <DonDiego> you have it running on your machine though?
[17:33:22] <DonDiego> getting oprofile up and running should be doable
[17:33:36] <DonDiego> if you later tell me what exactly you want/need profiled, i can do that
[17:33:56] <peloverde> Nearly every oprofile guide I can find is about profiling the kernel itself
[17:34:10] <janneg> twnqx: 05_throwing-0[3789] rendered without problems
[17:34:10] <DonDiego> what do you use for profiling?
[17:34:33] <peloverde> I use callgrind and verify with time
[17:34:57] <DonDiego> callgrind_annotate  callgrind_control
[17:35:02] <DonDiego> i have these two in my path
[17:41:13] <DonDiego> peloverde: what do you do exactly with callgrind and time?
[17:41:51] <DonDiego> (there is no callgrind_* man page on my system)
[17:42:30] <peloverde> valgrind --tool=callgrind ./ffmpeg_g -vn -i file.m4a -f null /dev/null
[17:43:26] <DonDiego> i'll try that later, after a recompile, i need to get to do some work now..
[17:45:27] <peloverde> ok
[17:45:32] <twnqx> janneg: http://img534.imageshack.us/img534/8765/bbbx.png maybe my blender is a bit more hungry than others...
[17:48:34] <twnqx> then again... mem usage seems to be lower, with just 9G total
[17:53:04] <peloverde> Why are japanese broadcasters broadcasting something that seems a lot like aac but isn't?
[17:53:43] <mru> not aac at all?
[17:53:46] <kierank> dunno, japanese tv is weird
[17:54:11] <kierank> coreaac plays the actual program audio ok, but not the adverts and last time i tried ffmpeg did it the other way round
[17:54:28] <peloverde> Well it doesn't conform to 14496-3 or 13818-7 so it's not AAC
[17:54:42] <kierank> it's meant to be the latter
[17:54:55] <kierank> but they could have modified it in their national specs
[17:56:12] <peloverde> why not just follow the rules in the spec?
[17:56:33] <peloverde> they are there to guarantee interoperability
[17:57:10] <twnqx> they are afraid of streamrippers and reencoders
[17:57:11] <janneg> but then everybody could play^Wsteal that precious intellectual property
[17:57:29] <peloverde> I don't think it's obsfucation
[17:57:34] <peloverde> I think it's a lazy muxer
[17:58:23] <peloverde> It's the classic element instance tags are changing midstream issue
[17:59:20] <janneg> twnqx: it's strange, I somehow suspect blender has some automatic memory usage tuneables or a lazy GC if there's no memory pressure
[18:00:17] <superdump> peloverde: maybe we should add a 'treat element instance tags naively' mode?
[18:00:34] <superdump> so that crap streams can be played in some form
[18:00:50] <kierank> hmmm version.sh doesn't work in mingw any more
[18:01:01] <superdump> because they don't conform, we can not give a crap about the channel order of the output
[18:01:19] <superdump> but we can at least allow them to decode, as it seems to be a common issue
[18:02:01] <peloverde> What worries me about that is channel counts changing on the fly
[18:02:13] <superdump> hmm
[18:04:04] <peloverde> I guess the trick is to save a locked configuration, and if the new configuration has the same count and locks then replace the old one and treat it naively, otherwise assume it's corruption and revert to the locked configuration
[18:07:56] <peloverde> I guess we are going to have to deal with on the fly channel count changes for backwards compatible PS
[18:08:20] <superdump> hrm
[18:09:55] <peloverde> because backwards compatible PS is signaled inside the SBR data payload
[18:10:25] <peloverde> So if there is no SBR Data yet, there is no PS
[18:13:19] <superdump> they really did this with tape and glue didn't they?
[18:13:59] <peloverde> There was no field for backwards compatible extensions inside SBR but outside of the data payload
[18:14:23] <peloverde> So I guess the choices were to put it straight in AAC or put it in the payload
[18:14:54] <peloverde> Really makes me wish i had access to USAC, I feel like I could find some spec defects before it's frozen
[18:15:05] <peloverde> but sadly they insist on developing it in secret
[18:15:10] <superdump> most probably
[18:16:04] <peloverde> Another bad article on <video> codecs: http://www.zdnet.co.uk/news/cloud/2010/03/19/sting-in-the-tail-for-webs-video-codec-search-40082396/
[18:16:24] <kshishkov> peloverde: probably people like you is the reason they develop it in secret
[18:17:18] <kshishkov> peloverde: looks like they don't want spec to be usable for developing independent implementation
[18:17:58] <peloverde> I guess so
[18:18:31] <peloverde> I miss independent coding technologies, they needed independent implementations, it was part of their buisness model
[18:18:59] <kshishkov> but they made SBR!
[18:19:05] <peloverde> and they were actually helpful re: specification defects
[18:19:08] <peloverde> SBR is great
[18:20:41] <peloverde> implicit SBR is awkward, signaled SBR is fine
[18:21:23] <peloverde> They should have never allowed implicit signaling, hierarchical backwards compatible is good enough, it just doesn't support ADTS
[18:22:01] <peloverde> Also if anyone else wants to learn about SBR the mp3pro xmms plugin seems very RE friendly
[18:22:32] <ramiro> kierank: what's wrong with it (version.sh)?
[18:22:59] <_av500_> peloverde: just that nobody uses mp3pro...
[18:23:17] <peloverde> I always see people complaining on the ubuntu forums about lack of mp3pro
[18:23:24] <peloverde> so it appears a few suckers fell for it
[18:23:26] <kshishkov> good, MP3 is crap
[18:23:30] <kierank> ramiro: for whatever reason it doesn't work in mingw for me, not a big deal though. probably mingws fault
[18:23:54] <_av500_> peloverde: our users never asked for it
[18:24:03] <ramiro> kierank: could you be more specific as in what doesn't work? even if it's mingws fault I'm interested
[18:24:06] <mru> one of the coding technologies guys taught my signals&systems class
[18:25:02] <siretart`> is vrml really part of mpeg4?
[18:25:04] <_av500_> peloverde: when do you do the 960 sbr?
[18:25:10] <_av500_> for DRM
[18:25:12] <mru> siretart`: _everything_ is part of mpeg4
[18:25:33] <peloverde> When someone adds a non-power-of-2 fft to ffmpeg
[18:25:35] <kierank> ramiro: it doesn't output a version.h file at all but i haven't investigated why
[18:25:49] <peloverde> we don't even support 960 LC
[18:26:05] <peloverde> The reference software didn't support 960 until last year!
[18:26:20] <kshishkov> so who cares?
[18:26:52] <superdump> no one
[18:27:23] <_av500_> so i have to use faad for it, shame :)
[18:27:44] <superdump> you mean you actually have some 960 aac?
[18:27:59] <_av500_> you can receive it easily
[18:28:26] <_av500_> my coworker did the dream dm project
[18:29:50] <peloverde> Why did DRM standardize on 960?
[18:30:06] <mru> because it's non-standard of course
[18:30:08] <ramiro> kierank: worked fine here.
[18:30:14] <mru> peloverde: it's french
[18:30:18] <mru> do I need to say more?
[18:30:22] <kshishkov> no
[18:30:23] <mru> remember secam?
[18:30:47] <kierank> so how does 960 relate to french protectionism?
[18:31:14] <peloverde> I never really understood 960 AAC, your short window is 8 samples shorter, that doesn't get you anything useful
[18:31:48] <_av500_> peloverde: no idea why...
[18:32:33] <peloverde> somebody who has an expensive library for awkward sized FFTs must have been on the committee
[18:33:46] <aaronl> i wonder if such a library exists when fftw works great for weird fft sizes, and a bunch of commercial software uses it
[18:33:48] <kshishkov> FFT-15 FTW!
[18:34:18] <peloverde> fftw is GPL/commercial
[18:34:24] <peloverde> so maybe them
[18:35:00] <aaronl> ah, i didn't realize that
[18:35:20] <kshishkov> reminds me one company asked the company I worked at if they could cook up something faster than Intel IPP FFT and FFTW
[18:37:07] <peloverde> Totally possible if you are running on interesting hardware and can exploit a feature they don't
[18:37:54] <kshishkov> so no - they asked for x86
[18:38:20] <kshishkov> and I was the only one there who knew something about FFT anyway
[19:08:51] <ramiro> hmm, why is gcc sub'ing and add'ing to rsp in this function (win64): http://ffmpeg.pastebin.com/9hGE8wkQ
[19:11:11] <jai> ramiro: it shouldnt have done that at -O3
[19:11:36] <ramiro> it's gcc 4.4.2
[20:03:01] * KotH just found a funny quote
[20:03:14] <mru> enlighten us
[20:03:16] <KotH> "... nobody programs in machine language anymore..."
[20:03:22] <mru> rotfl
[20:03:26] <KotH> from a book from 1971
[20:03:39] * mru did just that less than a month ago
[20:04:07] <mru> wrote instructions directly in hex, that is
[20:04:40] <KotH> i suspect that he meant asm with that
[20:04:52] <mru> oh, I did that yesterday
[20:04:59] <KotH> the language in programming changed quite considerably in the last 40 years
[20:05:26] <KotH> and he is still talking about punch cards ^^'
[20:07:07] <mru> KotH: btw, http://newteevee.com/2010/03/20/happy-birthday-gnutella-pioneering-p2p-protocol-turns-ten/
[20:07:17] <mru> apropos our discussion in brussels
[20:12:38] <KotH> it's already been 10y?
[20:12:39] <KotH> wow
[20:14:02] * _av500_ feels older now
[20:15:02] <_av500_> uni sent me new student card, i am in 40th semester now...
[20:15:14] <KotH> lol
[20:15:55] * _av500_ should apply for gsoc
[20:16:06] <mru> lol
[20:16:10] <mru> where's your qual task?
[20:16:26] <_av500_> damn
[20:16:31] <Dark_Shikari> lol
[20:16:56] <Dark_Shikari> reminds me, the other day facebook sent me a care package with a guide to getting accomodated to the area/etc
[20:17:00] <Dark_Shikari> it was only what, 9 months late?
[20:17:13] <KotH> rotfl
[20:17:40] <mru> I didn't know fb was that big a company
[20:18:10] <_av500_> or has a very overworked hr team
[20:18:14] <Dark_Shikari> 1000 employees last I was there
[20:18:25] <KotH> night boys
[20:18:41] <_av500_> dltbbby
[20:18:50] <KotH> gesundheit
[20:18:54] <mru> ah, that's the size when they start acting like they're 5x bigger than they are
[20:19:09] <mru> and failing even worse than if they really were that size
[20:58:44] <DonDiego> ffserver.c:1376: warning: implicit declaration of function ‘av_match_ext’
[20:58:48] <DonDiego> haha
[20:58:51] <DonDiego> internal API abuse..
[20:59:40] <mru> that one is currently the subject of bikeshedding
[21:01:43] <mru> as soon as that and the other one like it in ffserver are fixed I'll make such abuse an error
[21:15:59] <DonDiego> please do
[21:27:56] <CIA-30> ffmpeg: mru * r22609 /trunk/libavcodec/ (fft.h rdft.c):
[21:27:56] <CIA-30> ffmpeg: Call rdft by function pointer
[21:27:56] <CIA-30> ffmpeg: Call the RDFT by a function pointer like other FFT related transforms.
[21:27:56] <CIA-30> ffmpeg: This makes instruction set optimized versions possible.
[21:27:56] <CIA-30> ffmpeg: Based on patch by Alex Converse.
[21:27:57] <CIA-30> ffmpeg: mru * r22610 /trunk/libavcodec/rdft.c: rdft: reorder functions to avoid static prototype
[21:40:40] <CIA-30> ffmpeg: diego * r22611 /trunk/ffplay.c:
[21:40:40] <CIA-30> ffmpeg: Move variable declaration, fixes the warning:
[21:40:40] <CIA-30> ffmpeg: ffplay.c:1703: warning: unused variable `pos'
[23:08:14] <peloverde> Have any specifics released about the Aldi DVD lawsuit?
[23:15:06] <Dark_Shikari> peloverde: what's that?
[23:17:13] <peloverde> "(DENVER, CO, US – 16 March 2010) – MPEG LA, LLC, world leader in alternative one-stop patent licenses, today announced that several patent holders in MPEG LA’s MPEG-2 Patent Portfolio License have filed separate patent enforcement actions in Landgericht Düsseldorf, Germany against Aldi Nord GmbH & Co. oHG (“Aldi Nord”) and Aldi Süd GmbH & Co. oHG (“Aldi Süd”), as well as Lidl Stiftung & Co. KG and Lidl Dienstleist
[23:17:13] <peloverde> ung GmbH & Co. KG (“Lidl”), for infringing patents essential to the MPEG-2 digital video compression standard used worldwide in digital television broadcasting and DVD."
[23:18:59] <CIA-30> ffmpeg: aurel * r22612 /trunk/libavcodec/error_resilience.c: fix compilation with --disable-everything --enable-decoder=mpeg2video
[23:21:10] <Dark_Shikari> peloverde: ooh.
[23:23:51] <peloverde> Except all I can find (in english) is that dumb press release that doesn't say who the plaintiffs are and which patents were violated
[23:25:58] <DonDiego> all of the mpeg-2 portfolio i guess
[23:26:22] <DonDiego> if you ship hardware you better watch out for the mpeg la
[23:27:32] <peloverde> DonDiego, certainly not all of the portfolio
[23:29:14] <peloverde> Despite claims of essentiality not everything there is truly essential
[23:29:29] <peloverde> and it says that only "several" of the licensors are suing
[23:29:30] <DonDiego> yes of course
[23:29:57] <Dark_Shikari> honestly though, if you are shipping hardware, you can afford to pay license fees
[23:30:06] <peloverde> Much like the RIAA, the MPEG-LA doesn't sue anybody, the actual owners wind up suing
[23:30:09] <Dark_Shikari> (and equally, hardware is much easier to sue over)
[23:30:10] <kierank> aldi dvd players are dirt cheap
[23:30:20] <Dark_Shikari> kierank: and now we know why... ;) ;)
[23:30:26] <Dark_Shikari> then again, aren't mpeg-2 fees really cheap?
[23:30:27] <DonDiego> hardware does not have much of a profit margin
[23:30:29] <kierank> they presumably got advice that the patent licence was unnecessary
[23:30:43] <ShadowJK> or just thought that the manufacturer paid it
[23:30:50] <_av500_> or are presumably clueless
[23:30:55] <DonDiego> it's like paying protection money to the mafia, nothine more nothing less
[23:30:59] <DonDiego> you work, they profit
[23:31:01] <_av500_> ShadowJK: the chinese, no
[23:31:13] <kierank> i wonder if aldi's dirt cheap satellite systems will be affected too
[23:31:17] <Dark_Shikari> ShadowJK: the rule btw is that the seller pays
[23:31:25] <Dark_Shikari> i.e. if I license my product X to company Y
[23:31:31] <ShadowJK> I wonder how much the patents cost, must be a significant chunk of 19.90E DVD players :)
[23:31:32] <Dark_Shikari> and company Y sells product X in box Z to users
[23:31:37] <Dark_Shikari> company Y must pay the fees, not me.
[23:32:03] <Dark_Shikari> ShadowJK: licensing fees are like $20-40 in a top-end fancy blu-ray player
[23:32:03] <_av500_> he who imports to e.g. eu pays
[23:32:06] <Dark_Shikari> with support for the kitchen sink
[23:32:13] <Dark_Shikari> per unit
[23:32:25] <_av500_> per snk
[23:32:45] <kierank> Dolby E is $2-3k per device...
[23:33:30] <kierank> aka daylight robbert
[23:33:32] <kierank> robbery*
[23:33:39] <_av500_> now that mpegla has extended h264 grace to 2016, they need money from elsewhere :)
[23:33:51] <_av500_> so, aldi, blame theora!
[23:34:05] <peloverde> the mpeg la isn't suing anyone
[23:34:14] <_av500_> i know
[23:34:25] <Dark_Shikari> _av500_: it's mostly a matter of setting an example
[23:34:37] <Dark_Shikari> if you want everyone making DVD players to pay you 10 cents per unit
[23:34:41] <Dark_Shikari> you have to go after the people who don't
[23:34:47] <Dark_Shikari> even if it's not profitable to do so
[23:34:52] <Dark_Shikari> because if you didn't, nobody would pay
[23:35:04] <peloverde> agreed
[23:35:11] <_av500_> yes
[23:35:27] <Dark_Shikari> thus a lawsuit is effectively a signal of "we care about this particular market"
[23:36:16] <_av500_> i guess most stuff here is licensed, aldi being a discounter maybe thought they get a discount :)
[23:36:36] <peloverde> Still in theory you could draw up a list of actually essential patents and secure licenses independently of the MPEG-LA
[23:36:56] <_av500_> if licensor would talk to you
[23:37:06] <_av500_> and not send u to mpegla
[23:37:14] <peloverde> Sometimes I wonder if the patent cross licensing agreements that microsoft strong arms people into include their media patents
[23:37:23] <kierank> aldi are massive anyway so this will be interesting to watch
[23:37:57] <peloverde> And some would argue that Samsung can't enforce thier patents against LGPL implementations
[23:38:07] <_av500_> maybe they will switvh to ogg players...
[23:38:43] <peloverde> I can't wait to go to Aldi and pick up some blurry Ogg-Ray discs
[23:38:49] <ohsix> then what about the content people actually want to play
[23:38:58] <kierank> peloverde, or AVS-Ray
[23:39:02] <twnqx> janneg: 05 02 just hit 8.9G
[23:39:08] <_av500_> it is aldi, they buy coz it is cheap
[23:39:10] <twnqx> oh, 9.3 now
[23:39:19] <ohsix> snownutray
[23:39:34] <kierank> is aldi a reputable store in germany? because it isn't over here
[23:39:40] <twnqx> no
[23:39:46] <twnqx> it's a cheap store
[23:40:01] * _av500_ sopsthere
[23:40:09] * _av500_ shops there
[23:40:21] <twnqx> me too, just... rarely, and specific
[23:40:28] * _av500_ larts cheap aldi keyboard
[23:40:44] <twnqx> i wouldn't buy computers there :P
[23:40:51] <_av500_> nope
[23:41:00] <microchip_> they come from Medion mostly
[23:41:12] <twnqx> of course
[23:41:13] <microchip_> and Medion is sh*t
[23:41:13] <_av500_> their computers are expensive actualy
[23:41:27] <twnqx> they are cheaper than mine
[23:41:27] <_av500_> coz they come with kitchen sink that u dont need
[23:41:30] <twnqx> (in every regard)
[23:42:02] <twnqx> hm these scenes could be rendered if i upgraded my i7 to 12G...
[23:42:13] <twnqx> 6G won't do...
[23:42:20] <_av500_> twnqx: the last aldi 16core was not too ad, no?
[23:42:22] <janneg> twnqx: that explains why it oom-ed on the 8G machine. the first take though which needs more than 8G
[23:43:00] <twnqx> well, i'll shut the box for the night
[23:43:04] * janneg must have missed the aldi 16core with 48G Ram
[23:43:16] <twnqx> when the raid finished rebuilding... wonder why the one drive was kicked from both raids :X
[23:43:18] <_av500_> aldi sued ftw :)
[23:43:31] <twnqx> _av500_: i shop at aldi süd during the week
[23:43:36] <twnqx> and north during the weekend :P
[23:44:00] <microchip_> Lidl > Aldi :p
[23:44:07] <kierank> lol
[23:44:12] <_av500_> so u have the 16core and the track suit
[23:44:29] <kierank> microchip_, theres about 100 levels of crap stores over here
[23:44:29] <twnqx> :]
[23:44:35] * janneg has to pass two bundesländer to get to aldi süd
[23:44:41] <twnqx> heh
[23:44:56] <twnqx> i'm passing between hessen (for work) and niedersachsen (for the weekend)
[23:45:01] <twnqx> commuting*
[23:45:08] <microchip_> kierank: not too many here :)
[23:45:23] <kierank> there's netto and makro all sort of others
[23:45:27] <_av500_> gee, nordhessen
[23:45:52] <twnqx> actually
[23:46:00] <twnqx> bonn is in NRW, not hessen, right?
[23:46:20] <_av500_> last time i checked yes
[23:46:23] <twnqx> so yeah
[23:46:31] <twnqx> i'm in hessen only when i have to fix my servers >_>
[23:47:23] <DonDiego> gnite
[23:49:10] <twnqx> ok, blender doesn't die to ctrl-c
[23:49:11] <twnqx> Oo
[23:49:20] <Dark_Shikari> kill -9
[23:51:15] <janneg> twnqx: double ctrl-c
[23:51:40] <twnqx> i sent like 100 :P
[23:51:45] <twnqx> anyway, dead now
[23:52:15] <twnqx> jobs 10 and 11 are completed, i hope to put them online tomorrow


More information about the FFmpeg-devel-irc mailing list