[Ffmpeg-devel-irc] ffmpeg-devel.log.20150901

burek burek021 at gmail.com
Wed Sep 2 02:05:02 CEST 2015


[00:03:36 CEST] <lglinskih> kierank: thank you)
[00:05:00 CEST] <cone-790> ffmpeg 03Michael Niedermayer 07master:a1a32fdb0ee6: avcodec/movtextdec: Make tx3g_ptr unsigned
[00:05:01 CEST] <cone-790> ffmpeg 03Michael Niedermayer 07master:a047ccbb9fad: avcodec/movtextdec: Free ftab_temp
[03:13:57 CEST] <cone-517> ffmpeg 03Michael Niedermayer 07master:b54e03c9dc2a: avcodec/mpeg12dec: Set dimensions in mpeg1_decode_sequence() only in absence of errors
[05:42:17 CEST] <cone-517> ffmpeg 03Michael Niedermayer 07master:63fb5a6aefb4: avcodec/truemotion1: Check for even width
[08:09:44 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:5ed5ca706fce: aacenc: populate tns_max_bands
[08:09:45 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:139c2f93fd67: aacenctab: add tns_min_sfb[] to the encoder tables header
[08:09:46 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:a70ee2bdcd88: lpc: remove unused ff_lpc_calc_levinson() function
[08:09:47 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:0fc3a5135350: lpc: add ff_lpc_calc_ref_coefs_f() function
[08:09:48 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:d09f9c45c755: aacenc: allocate a larger buffer for the TNS LPC context
[08:09:49 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:f3f6c6b92822: aacenc_tns: rework coefficient quantization and filter application
[08:09:50 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:a0079aae00a7: aacenc: reorder resetting of cpe->common_window
[08:09:51 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:a8ced764f729: fate: update AAC Encoder TNS test for the new changes
[08:28:08 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:b7eb7cb3a180: aacenc: Enable Perceptual Noise Substitution by default
[08:28:09 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:bc9927b8548b: aacenc: Enable Intensity Stereo by default
[08:28:10 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:555f513769ec: fate: update AAC tests to the new defaults
[08:35:37 CEST] <imaculate> Hi all!
[08:36:11 CEST] <imaculate> I'm interested in doing opw internship from December with ffmpeg.
[08:36:28 CEST] <imaculate> Will ffmpeg be participating this year?
[08:52:48 CEST] <durandal_1707>  dunno ask on mailing list
[08:59:13 CEST] <imaculate> thank you
[09:23:36 CEST] <cone-517> ffmpeg 03Carl Eugen Hoyos 07master:0cb6c0ec488e: lavfi/scale: Pass src_range and dst_range to libswscale.
[09:23:36 CEST] <cone-517> ffmpeg 03Carl Eugen Hoyos 07master:670d8ecfae52: lavf: Remove probesize32 and max_analyze_duration32 on version bump.
[10:15:13 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:b86532810ddc: fate: adjust fuzz of AAC Encoder's TNS test
[10:29:01 CEST] <nevcairiel> now i need to remember to turn PNS back off, should write a note
[10:29:36 CEST] <nevcairiel> i still vote for introducing a mpeg2-lc  profile that limits coding tools to mpeg2 part 7 features
[10:29:45 CEST] <atomnuker> fair enough
[10:30:53 CEST] <atomnuker> I actually have a patch which enabled turning on/off default options per profile
[10:31:17 CEST] <atomnuker> I'll dig it up and merge it
[10:31:36 CEST] <nevcairiel> apparently PNS is the only feature so far affected, IS and TNS are in part 7 already
[10:32:01 CEST] <atomnuker> is there a macro for part 7? like FF_PROFILE_AAC_LTP?
[10:32:20 CEST] <nevcairiel> dont think so
[10:32:34 CEST] <nevcairiel> its technically LC profile as well, so noone made a distinction i suppose
[10:34:05 CEST] <nevcairiel> could also just be a separate private option that enables such a mode by limiting tools, if profiles dont really fit it
[10:34:12 CEST] <nevcairiel> or i just have to remember :D
[10:35:20 CEST] <atomnuker> where do you use the encoder so that you need strict part 7 compliance anyway?
[10:36:44 CEST] <nevcairiel> local streaming, some devices are a bit stubborn and dont like mpeg4 aac
[10:36:55 CEST] <nevcairiel> just signaling mpeg4 in the adts header will make them bail out
[10:39:10 CEST] <atomnuker> actually there is an option for strict part 2
[10:39:18 CEST] <atomnuker> mpeg2_aac_low - FF_PROFILE_MPEG2_AAC_LOW
[10:39:37 CEST] <atomnuker> you can specify that right now as -profile:a but it doesn't get any special treatment
[10:39:37 CEST] <nevcairiel> that does exist? never noticed
[10:40:30 CEST] <nevcairiel> looks like the profile check in the encoder would call it unsupported
[10:41:21 CEST] <atomnuker> yes, it'll consider it unsupported
[10:42:54 CEST] <atomnuker> I'll merge that patch once I dig it up later today
[10:43:18 CEST] <atomnuker> TNS and IS are in part 7, right? just not PNS
[10:43:44 CEST] <nevcairiel> yes
[10:43:52 CEST] <nevcairiel> on that note, how much does IS help?
[10:44:10 CEST] <cone-517> ffmpeg 03Paul B Mahol 07master:32f217edadf0: avfilter/vf_waveform: implement various filters
[10:44:10 CEST] <cone-517> ffmpeg 03Paul B Mahol 07master:97c7c39d2539: avfilter/vf_waveform: simplify memory allocations
[10:44:11 CEST] <cone-517> ffmpeg 03Paul B Mahol 07master:c3cd1a7496fe: avfilter/vf_waveform: support envelope for all filters
[10:46:22 CEST] <atomnuker> helps quite a lot to reduce artifacts below 5khz since the upper bits cost far less bits to encode
[10:46:58 CEST] <nevcairiel> back in the day when you heard "joint stereo mp3", it was usually considered to be lower quality, but i dont know how the coding tools compare from mp3 to aac
[10:47:24 CEST] <atomnuker> joint stereo translates as M/S in AAC, and yep, we have it, it's just currently not enabled by default
[10:47:42 CEST] <atomnuker> does a fine job though not as good as IS does
[10:48:02 CEST] <nevcairiel> MS is also perfectly lossless, so that wouldnt account for perceived worse quality
[10:48:06 CEST] <atomnuker> (at giving more bits to parts of the signal that actually need it)
[10:48:59 CEST] <nevcairiel> didnt MP3 also support intensity stereo
[10:49:07 CEST] <atomnuker> yes, it does
[10:49:28 CEST] <atomnuker> I remember reading it had a very strict set of conditions for it to be enabled
[10:49:43 CEST] <nevcairiel> after reading up on some of these, its fasinating how small the difference is from mp3 to aac in coding tools
[10:49:58 CEST] <atomnuker> well, yes, in coding tools
[10:50:08 CEST] <atomnuker> though AAC is sane in that it uses a proper MDCT
[11:54:39 CEST] <mouni> Hi hello i am just a newbie to ffmpeg . I am using koding.org to deploy ffmpeg .I have a doubt . Does ffmpeg need the gui to access it ? or to test it ?
[11:55:04 CEST] <mouni> Is this a gui based application ? or can i make the project only from backend ?
[11:55:25 CEST] <mouni> i mean from terminal without the requirement of gui
[11:57:23 CEST] <atomnuker> nevcairiel: Here's the patch to select coding tools by profile
[11:57:24 CEST] <atomnuker> https://paste.debian.net/309925/
[11:58:05 CEST] <atomnuker> I'm concerned about the fact that because I override the default options with -1 the "default" field in -help encoder=aac is just -1 for all
[11:58:42 CEST] <atomnuker> so I have to add "(default)" in the option description to point out what's the default
[11:59:27 CEST] <nevcairiel> hm yeah
[11:59:56 CEST] <atomnuker> it's a hacky solution alright, but I'm thinking it might be too hacky
[12:00:07 CEST] <nevcairiel> but profiles change default behaviour in other codecs, so its probably not too bad
[12:00:51 CEST] <atomnuker> yes, but the reason I have to set -1 to be the default for all options is just for detection purposes
[12:01:08 CEST] <mouni> please do help me
[12:01:11 CEST] <atomnuker> because you'll probably want to be able to override by stating -aac_option 1 or whatever
[12:02:03 CEST] <nevcairiel> its a shame we dont have a "soft" 1 and a "hard" 1 :D
[12:02:18 CEST] <nevcairiel> so you could detect differences between default or user-specified
[12:02:23 CEST] <nevcairiel> without having to change the default
[12:03:18 CEST] <atomnuker> yeah, if you wanted to do that you'd probably have to modify the main options struct and every single option for every single encoder/decoder
[12:14:48 CEST] <wm4> for the record, I can 100% understand why Libav people aren't interested in a merge in the slightest, with people like cehoyos around
[12:15:30 CEST] <nevcairiel> I said as much when the topic came up first, but others assured me he aint that bad
[12:16:04 CEST] <wm4> once in a while he's outdoing himself
[12:25:26 CEST] <durandal_1707> mouni: wrong channel
[12:38:02 CEST] <michaelni> atomnuker, "ffmpeg -i matrixbench_mpeg2.mpg -strict experimental -aac_coder 0 -acodec aac out.aac" fails with Assertion diff >= 0 && diff <= 120 failed at libavcodec/aacenc.c:316
[12:38:38 CEST] <atomnuker> must be the scalefactor indices, on it
[12:39:16 CEST] <nevcairiel> anything but the twoloop coder is kinda faulty, isnt it
[12:39:23 CEST] <nevcairiel> or is 0 twoloop
[12:39:57 CEST] <atomnuker> anmr/faac are useful for extremely low bandwidth voice
[12:41:10 CEST] <atomnuker> fast isn't even a proper coder, it just sets all scalefactor indices to some constant but is suprisingly coherent for full-bandwidth audio
[12:43:29 CEST] <atomnuker> michaelni: actually the faac coder has always been broken with some files and settings
[12:44:06 CEST] <atomnuker> I have a WIP fix for the case when it gets stuck with high bandwidths, I'll try to fix the scalefactor index problem too
[13:11:31 CEST] <cone-517> ffmpeg 03Rostislav Pehlivanov 07master:8ffe1cb4d7d4: aacenc: disable bandtype modifying extensions when coder != twoloop
[13:12:16 CEST] <atomnuker> that should fix it for now, of course it's encode_window_bands_info() failing
[13:12:58 CEST] <atomnuker> Claudio told me he once tried to merge all coders to use the trellis bandtype setter but gave up because it was too difficult
[13:14:09 CEST] <atomnuker> I'll give it a go, should be better than repairing that broken algorithm
[13:34:44 CEST] <michaelni> atomnuker, thanks
[13:56:20 CEST] <cone-517> ffmpeg 03Andrew Stone 07master:dc926ab518a3: rtmp: support the AMF_DATE tag
[14:12:41 CEST] <cone-517> ffmpeg 03Ganesh Ajjanagadde 07master:d79c20000052: avfilter/vf_spp: use the name 's' for the pointer to the private context
[14:57:30 CEST] <cone-517> ffmpeg 03Petri Hintukainen 07master:757cb0f286b5: mpegts: fix demuxing PES private stream 2
[14:57:31 CEST] <cone-517> ffmpeg 03Petri Hintukainen 07master:5d12d7de2ce5: mpegts: demux BluRay text subtitles
[15:40:39 CEST] <BBB> Daemon404: trac 4819 may be for you (its x265 related)
[15:40:48 CEST] <BBB> (it comes from stackoverflow)
[15:43:53 CEST] <Daemon404> looks internal to x265
[15:44:33 CEST] <Daemon404> g 44
[15:44:35 CEST] <Daemon404> woops
[15:46:06 CEST] <Daemon404> BBB, weird - looks like an implementation detail of dlopen...
[15:46:59 CEST] <Daemon404> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=700899
[15:47:02 CEST] <Daemon404> looks like same as this
[16:29:38 CEST] <BtbN> Is it only me or is "hevc - fix split function of parser" lacking the patch?
[16:30:33 CEST] <wm4> BtbN: seems to be missing
[16:31:49 CEST] <BBB> Daemon404: isnt it x265_cleanup() missing?
[16:33:35 CEST] <wm4> BtbN: now it's there
[16:34:35 CEST] <BtbN> oh, it's in a second mail
[16:34:38 CEST] <BtbN> interesting
[16:38:18 CEST] <wm4> probably used --cover-letter
[16:39:05 CEST] <Daemon404> BBB, hmmm not 100% sure
[16:39:14 CEST] <Daemon404> get_api support wasnt written by me, but an x265 dev
[16:39:21 CEST] <Daemon404> i sent a poke to muggs.
[16:41:05 CEST] <BBB> got it, ty
[16:42:52 CEST] <Daemon404> actually teh dlopen bit was written by muggs too
[17:33:05 CEST] <atomnuker> e2222
[17:33:21 CEST] <atomnuker> sorry, wrong channel
[17:38:49 CEST] <kierank> http://aomedia.org/
[17:42:08 CEST] Action: wm4 barfs all over kierank 
[17:42:51 CEST] <wm4> continuing the tradition that all things that have "Open" in the name are bad
[17:45:45 CEST] <wm4> I like how they have "content encryption" in their name
[17:45:51 CEST] <wm4> even the nazis were better
[17:52:54 CEST] <BBB> wm4: tsk tsk tsk
[17:53:05 CEST] <wm4> s/name/goals
[17:54:43 CEST] <BBB> you know, people read these logs and scan for godwin terms and alike
[17:54:51 CEST] <wm4> they better do
[17:55:19 CEST] <BBB> & its ok to barf over this stuff, but please, can we do so civilly&?
[17:56:01 CEST] <wm4> well, this site just reads like a parody
[17:56:23 CEST] <wm4> "From the companies behind the open web" "Microsoft"
[17:56:26 CEST] <wm4> etc.
[17:56:52 CEST] <BBB> I know& barfing is ok, really
[17:56:56 CEST] <BBB> we all do that all the time
[17:57:02 CEST] <BBB> but lets leave godwin aside, ok?
[17:57:12 CEST] <wm4> what's wrong with godwin
[18:03:48 CEST] <rcombs> they also fail to use an oxford comma
[18:04:19 CEST] <iive> https://www.youtube.com/watch?v=N4vf8N6GpdM 
[18:43:04 CEST] <j-b> kierank: http://www.aomedia.org/
[18:43:12 CEST] <j-b> arf, you posted it already
[18:44:37 CEST] <Daemon404> \o/
[18:45:49 CEST] <nevcairiel> since both google and microsoft are in there, maybe microsoft will finally acknowledge that vp8/9/10 exist
[18:47:50 CEST] <j-b> well, expect a mix of VP10/Daala and Thor...
[18:48:55 CEST] <Daemon404> inb4 no spec
[19:01:45 CEST] <wm4> "We believe that Daala, Ciscos Thor, and Googles VP10 combine to form an excellent basis for a truly world-class royalty-free codec."
[19:03:28 CEST] <j-b> Seeing that Google still refuses to open the spec of VP9
[19:03:34 CEST] <j-b> I doubt this will be good...
[19:03:52 CEST] <wm4> they'll make a code dump and everyone will use it, maybe?
[19:04:48 CEST] <j-b> like for vp9
[19:15:09 CEST] <philipl> j-b: even if it's a useless dump, with amazon, netflix and youtube, that have enough influence on the content creation side to force adoption to a large extent.
[19:16:17 CEST] <j-b> philipl: some people did not like hevc patents terms
[19:16:58 CEST] <philipl> indeed, but given that there won't be horrible patent terms here (however they end up achieving that goal), that's not an issue.
[19:17:16 CEST] <wm4> philipl: plus they will provide a DRM solution
[19:17:19 CEST] <wm4> all across the board
[19:17:24 CEST] <wm4> including the OS
[19:18:16 CEST] <philipl> That's inevitable, and largely orthogonal, I think. widevine can work with anything right? ;-)
[19:21:58 CEST] <Daemon404> i cant wait for the future
[19:22:11 CEST] <Daemon404> vc-5 mezzanine and code-dump aomedia codec prod
[19:25:57 CEST] <rcombs> people come up with new video codecs to get better compression
[19:26:13 CEST] <rcombs> people come up with new DRM setups because somebody figured out how the old one works
[19:26:43 CEST] <Daemon404> the future is drm built into skylake
[19:26:46 CEST] <Daemon404> supported by the OS
[19:26:55 CEST] <Daemon404> praise satan, etc
[19:27:04 CEST] <rcombs> time for some decapping
[19:27:21 CEST] <philipl> Daemon404: that's different from hdcp?
[19:27:40 CEST] <rcombs> meanwhile you can still output video over HDMI
[19:27:44 CEST] <rcombs> and HDCP is still broken
[19:27:58 CEST] <philipl> We've already had hardware that takes encrypted video in and pushes it out with hdcp for years and OS support since vista.
[19:28:02 CEST] <Daemon404> the current solution used by netflix et al is to simply not support HD on PC.
[19:28:04 CEST] <philipl> I don't think anything's meaningfully changing there.
[19:28:05 CEST] <Daemon404> only plastic toys get it
[19:28:16 CEST] <Daemon404> or for 4k, only smart tvs
[19:28:22 CEST] <philipl> And we all shrug.
[19:28:47 CEST] <Daemon404> the real solution is quantum computing
[19:28:49 CEST] <Daemon404> duh
[19:28:58 CEST] <philipl> Then you're never sure what show you're watching.
[19:30:43 CEST] <rcombs> the real solution is to never let anyone watch anything
[19:30:54 CEST] <rcombs> just take people's money and don't actually distribute any content
[19:30:55 CEST] <philipl> They would be happiest if you just paid them and they didn't show you anything, yes.
[19:32:01 CEST] <rcombs> in addition to solving the analog hole, this prevents you from recounting the content to your friends verbally
[19:32:10 CEST] <rcombs> or from reenacting scenes
[19:32:43 CEST] <philipl> Genius.
[19:33:04 CEST] <rcombs> I should get a patent on this
[19:33:13 CEST] <rcombs> and see if I can sell licenses to studios
[19:33:26 CEST] <rcombs> you haven't even heard the best part yet, though
[19:33:43 CEST] <rcombs> I thought I'd save it for once the patent was pending but here goes
[19:34:01 CEST] <rcombs> if you just take people's money without actually showing them anything, you don't have to _make_ the content either
[19:34:20 CEST] <rcombs> so all the piracy is dealt with and your expenses drop dramatically
[19:37:06 CEST] <Daemon404> 'noflix'
[19:39:27 CEST] <wm4> the real solution is the theatre
[19:39:35 CEST] <wm4> let actors perform live
[19:39:41 CEST] <wm4> nobody can pirate this
[19:42:58 CEST] <nevcairiel> ever heard of a camera
[19:43:56 CEST] <wm4> nobody likes to watch recorded theatre
[19:51:10 CEST] <ubitux> when the show is good, it doesn't really matter
[19:51:34 CEST] <ubitux> even i enjoyed the rip of mei to koneko basu
[19:52:50 CEST] <nevcairiel> i think there is a big chunk of piracy that only exists because content providers are dumb idiots and dont provide what the users want and/or as hassle-free as users would want it
[19:54:17 CEST] <nevcairiel> and most of that is a factor of DRM in the first place, since it destroys everyones freedom, and secondly regional lock-in, i dont want to watch dub'ed shows two years later on our national TV, i want my shows in english when they air in the US
[19:59:03 CEST] <wm4> nevcairiel: of course, this happened to all areas of content... music, games, now video
[19:59:30 CEST] <nevcairiel> at least music learned, games somewhat too
[19:59:35 CEST] <nevcairiel> video is still in the stone age
[20:03:12 CEST] <durandal_1707> What about drawass filter
[20:04:08 CEST] <wm4> I'm not opposed to seeing more asses
[20:04:28 CEST] <wm4> (what would it do?)
[20:08:23 CEST] <philipl> wm4: did you ever get into a position to run vainfo?
[20:08:41 CEST] <durandal_1707> use ass to draw with libass
[20:10:13 CEST] <wm4> philipl: not yet
[20:10:20 CEST] <wm4> sorry
[20:10:27 CEST] <wm4> durandal_1707: sounds like a good idea
[20:10:39 CEST] <philipl> no worries. I know the universe doesn't want me to know.
[20:10:59 CEST] <wm4> durandal_1707: libass is gaining support for running fontconfig-free on windows and osx, so that might help
[20:11:05 CEST] <Compn> wm4 : i watched spamalot and book of mormon cammed from a theatrical performance :P
[20:11:10 CEST] Action: Compn is a jerk cam-watcher
[20:11:46 CEST] <wm4> Compn: the point is that people would still be ready to pay a lot of money to see the real thing
[20:14:13 CEST] <kierank> very cultured of wm4
[20:15:01 CEST] <wm4> sorry
[20:37:14 CEST] <Daemon404> why am i rwading about libav vs ffmpeg on gentoo forums, in a thread about ME
[20:37:20 CEST] <Daemon404> ME API*
[20:38:53 CEST] Action: Daemon404 also wonders if carl is a paranoid schizophrenic, based on his 'betrayal
[20:38:57 CEST] <Daemon404> and 'spies' posts
[20:39:29 CEST] <wm4> michael is probably already traumatized by what his attempts of extending the API resulted in
[20:39:47 CEST] <Daemon404> probably.
[20:40:01 CEST] <Daemon404> this is where one says "i feel bad for the submitter".
[20:41:32 CEST] <wm4> definitely
[21:06:47 CEST] <BBB> trollwars& :(
[21:08:09 CEST] <iive> well, you started them
[21:08:30 CEST] <BBB> I did :(
[21:08:34 CEST] <BBB> Im sorry
[21:09:26 CEST] <BBB> on the fun spectrum end: http://stackoverflow.com/questions/32272200/error-ffmpeg-not-found-on-android-device
[21:09:34 CEST] <BBB> error ffmpeg not found on android device
[21:09:50 CEST] <BBB> the guy doesnt seem to understand that ffmpeg is not installed on android by default
[21:10:52 CEST] <iive> well, it should :)
[21:14:18 CEST] <iive> btw, the gentoo thread is good one. I do recommend reading it.
[21:14:33 CEST] <iive> it might give you different perspective on the things.
[21:16:57 CEST] <BBB> wait why do I need a different perspective?
[21:17:03 CEST] <BBB> Im confused now
[21:17:24 CEST] <BBB> is this because Im a spy from gstreamer?
[21:17:31 CEST] <BBB> (really, Ive been told these exact words)
[21:19:02 CEST] <wm4> BBB: I'm also sorry
[21:19:09 CEST] <wm4> and iive should be sorry too
[21:19:20 CEST] <BBB> nah, lets just ignore this whole thing
[21:19:26 CEST] <BBB> waste of time :-/
[21:19:32 CEST] <wm4> anyway, seems like at least some engineer part of this ao thing have indeed no plans/intention/obligation of developing DRM
[21:19:41 CEST] <Daemon404> Let's go to the Winchester, have a pint, and wait for this to blow over.
[21:19:47 CEST] <iive> wm4: i have nothing to sorry, because i haven't sent my draft mail...
[21:20:00 CEST] <wm4> iive: heh
[21:20:17 CEST] <wm4> maybe I should just wait 10 minute before sending a mail reply
[21:20:25 CEST] <iive> wm4: make them 30
[21:21:17 CEST] <Daemon404> bikeshed.org still applies.
[21:30:43 CEST] <durandal_1707> adding float to swscale is impossible it would need bunch of float pixel formats
[21:31:07 CEST] <durandal_1707> swscale need complete overhaul
[21:32:02 CEST] <durandal_1707> and it appears avscale is vapourware
[21:33:24 CEST] <durandal_1707> michaelni: ^
[21:33:42 CEST] <Compn> durandal_1707 : adding float pixels wont be that hard ?
[21:34:28 CEST] <iive> it's basically writing second swscale.
[21:34:41 CEST] <Compn> yes but based on other one, isnt it like porting c to asm ?
[21:34:44 CEST] <iive> new formats, new code, new assembly
[21:37:30 CEST] <Compn> what did we need float pixels for again ? faster on modern cpu ?
[21:38:27 CEST] <durandal_1707> more colors
[21:39:12 CEST] <Compn> ah , more colors.
[21:39:25 CEST] <Compn> 16 should be enough...
[21:39:37 CEST] <iive> it's better when you have more processing of the image. e.g. film effects use it.
[21:40:43 CEST] <iive> otherwise each filtering stage adds its own sampling artifacts.
[21:52:11 CEST] <wm4> durandal_1707: there's also zlib2
[22:35:31 CEST] <ubitux> Compn: float pixels can be useful for GPU textures
[22:35:36 CEST] <ubitux> i suppose.
[22:37:06 CEST] <wm4> some decoders produce float output, at least this exr thing
[22:37:36 CEST] <wm4> and all kinds of ridiculus hacks are piling up because that decoder can't export the data without loss (has to convert to int)
[22:38:19 CEST] <wm4> so considering we even have crap like packed 16 bit RGB or 2 monochrome 1 bit formats (!), float format sounds justified
[22:40:31 CEST] <Compn> wm4 : you will probably hate me for saying so, but dont bother arguing history or 2011 with random users on the ml , please? 
[22:40:47 CEST] <Compn> i dont care how wrong they are, no point in arguing with people 
[22:41:24 CEST] <wm4> I already stopped
[22:41:50 CEST] <Compn> no worries :)
[22:41:57 CEST] <ubitux> good reaction from Ganesh
[22:41:59 CEST] <Compn> thanks to everyone for not arguing on the ml
[22:42:25 CEST] <Compn> yes, although i think banning people would be negative attention...
[22:42:26 CEST] <wm4> Compn: which multi-national corps?
[22:42:51 CEST] <wm4> Ganesh did the right thing
[22:43:23 CEST] <Compn> i know google at least :p
[22:44:12 CEST] <ubitux> i wonder what's this anarchy thing is all about too
[22:44:30 CEST] <Compn> lol
[22:44:48 CEST] <ubitux> it's like he's talking about something, and in the middle "hey i like anarchy"
[22:44:52 CEST] <ubitux> cool story bro :D
[22:45:19 CEST] <ubitux> but yeah, one more message from this guy and we'll need to take actions
[22:45:36 CEST] <wm4> must be drugs
[22:45:45 CEST] <Daemon404> worse
[22:45:47 CEST] <Daemon404> he is a gentoo user
[22:45:52 CEST] <Compn> (also dont know why you guys argue with carl)
[22:46:00 CEST] <wm4> Compn: neither do I
[22:47:52 CEST] <Compn> could make wiki page on trac of bugs that carl closed but you think should be open. 'immortal developer bugs' :)
[22:48:02 CEST] <Compn> then those get high priority on fix scale...
[22:48:15 CEST] <Compn> if you think you could use that workaround ?
[22:48:41 CEST] <Daemon404> the proper fix is for carl to stop acting like an obnoxious child.
[22:48:59 CEST] <Daemon404> but of course you dont consider that valid.
[22:49:14 CEST] <ubitux> well, you guys are hardly objective when you're talking with him either
[22:49:34 CEST] <ubitux> not trying to take his defense, but it's starting to look like scapegoating
[22:49:47 CEST] <iive> agree
[22:50:28 CEST] <Daemon404> what are making him the scapegoat of?
[22:50:38 CEST] <ubitux> that is only from an external PoV, i haven't followed the technical arguments
[22:50:48 CEST] <ubitux> i'm just talking about the aggressive attitude there is toward him
[22:51:19 CEST] <ubitux> i definitely understand that you get pissed of (i, too, had my share of fun with such exchanges)
[22:51:23 CEST] <Daemon404> recipricol attitude perhaps
[22:51:58 CEST] <ubitux> i'm just saying to won't really help anything
[22:52:06 CEST] <ubitux> s/to/it/
[22:52:16 CEST] <Daemon404> i agree, which is why youll note i stayed out of the last shitfets
[22:52:18 CEST] <Daemon404> shitfest*
[22:52:25 CEST] <Daemon404> my single post was unrelated to carl
[22:53:00 CEST] <Daemon404> )ignoring carl's behavior doesnt help anything either fwiw)
[22:54:28 CEST] <ubitux> i wasn't really targetting anyone
[22:54:34 CEST] <ubitux> but sure ok
[22:54:51 CEST] <wm4> ubitux: do you think my other reply to him today was unjustified? (something about fixing something from a ticket in which he didn't understand what the user wanted, or something)
[22:55:03 CEST] <Daemon404> also stop and think: such a large portion of people dislike him, such that there is large animosity
[22:55:07 CEST] <Daemon404> perhaps the problem os him.
[22:55:09 CEST] <Daemon404> is*
[22:55:42 CEST] <Daemon404> [21:44] <@ubitux> i wonder what's this anarchy thing is all about too <-- btw that is definitely a troll account
[22:55:46 CEST] <Daemon404> the dude's name is shen long.
[22:55:52 CEST] <Daemon404> the biggest troll of SF
[22:56:31 CEST] <Compn> you are sure its the same person?
[22:56:44 CEST] <Compn> (everyones out to get Daemon404)
[22:56:46 CEST] <ubitux> Daemon404: it's a celebrity?
[22:56:49 CEST] <Compn> ehe
[22:56:50 CEST] <Daemon404> ubitux, eh
[22:57:05 CEST] <Daemon404> the name is a reference to: https://en.wikipedia.org/wiki/Sheng_Long
[22:57:37 CEST] <Daemon404> SF -> street fighter
[22:57:49 CEST] <ubitux> wm4: i was just sharing a feeling; but you were definitely aggressive recently ;)
[22:57:54 CEST] <Compn> oh i thought you meant sourceforge
[22:57:56 CEST] <Compn> lol
[22:58:05 CEST] <Compn> https://en.wikipedia.org/wiki/Shenlong_(disambiguation)
[22:58:06 CEST] <ubitux> ah i thought sourceforge as well
[22:59:33 CEST] Action: iive too
[23:00:04 CEST] <Compn> Daemon404 is a street fighter master ?
[23:00:30 CEST] <Daemon404> no i suck at fighting games
[23:00:36 CEST] <Daemon404> it's just a famous video game myth
[23:00:37 CEST] <Compn> (how come we dont have video game tourneys at vdd?) 
[23:00:48 CEST] <cone-517> ffmpeg 03Michael Niedermayer 07master:c7c207aecde0: avformat/mxg: Use memmove()
[23:01:25 CEST] <Compn> Daemon404 : assuming he meant 'sheng long' the character in street fighter vs 'shen long' chinese dragon myth (and dragonball anime character)
[23:01:37 CEST] Action: Compn runs
[23:01:46 CEST] <Daemon404> isnt that shenron
[23:04:52 CEST] <Compn> maybe in the dub
[23:06:03 CEST] <Daemon404> i do see how it could be, since japanese doesn't have an 'l' sound
[23:06:06 CEST] <Daemon404> dont*
[23:06:37 CEST] <Daemon404> could be chinese
[23:06:46 CEST] Action: Daemon404 wonders why theyre debating the pseudonym of a random troll
[23:09:37 CEST] <Compn> summer insanity.
[23:11:03 CEST] <iive> btw, i think he called libav anarchist... that's why he mentions it.
[23:12:53 CEST] <cone-517> ffmpeg 03Michael Niedermayer 07master:642d2f9d558b: avcodec/version: Minor version bump due to 5d12d7de2ce548a3c9f5c175a9d1c5bfbdbcbfa8
[23:13:18 CEST] <wm4> iive: well, anything that guy wrote was dumb shit
[23:15:20 CEST] <iive> i think i know who he is...
[23:16:38 CEST] <iive> that guy who wrote cdrtools.. schilling 
[23:17:14 CEST] <wm4> he'd use his name, not a pseudonym
[23:17:32 CEST] <wm4> and he wouldn't be a gentoo user
[23:18:44 CEST] <iive> he should be user of some distro... debian and ubuntu are even less likely :)
[23:19:49 CEST] <iive> anyway, not important.
[00:00:00 CEST] --- Wed Sep  2 2015


More information about the Ffmpeg-devel-irc mailing list