[Ffmpeg-devel-irc] ffmpeg-devel.log.20130327

[burek]

[00:00] <ubitux> saste: ping
[00:00] <saste> ubitux, pong
[00:00] <ubitux> saste: so what's going on for the gsoc page?
[00:01] <ubitux> 2 days left afaik
[00:01] <ubitux> (introduction etc)
[00:01] <saste> ubitux, llogan^
[00:01] <saste> i'll do it tomorrow if  llogan won't have do it in the meantime
[00:02] <saste> will not have done it or whatever
[00:02] <saste> also i should ping mentors and backup mentors
[00:02] <saste> bcoudurier, ping for gsoc
[00:03] <saste> mentors and co-mentors are welcome
[00:03] <bcoudurier> yes
[00:03] <bcoudurier> put me on any project
[00:03] <bcoudurier> I'll mentor
[00:03] <saste> bcoudurier, http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_of_Code_2013
[00:04] <saste> you can choose amongst the several unmentored task, or propose a new task, or volunteer as backup mentor
[00:04] <saste> you need an account from Mike Melanson for editing the wiki, or just tell me or llogan and will fill it
[00:04] <saste> *we'll
[00:05] <llogan> any suggestions, ideas, requirements for intro?
[00:06] <saste> llogan, a small section about telling what the project consists (pay for twisting bits etc.)
[00:07] <saste> some guidelines about how and where to contact mentors, recommending to discuss with the candidate mentors or with the community
[00:07] <saste> you should mention that the students can propose a task, even in that case they need to find a mentor willing to help
[00:08] <saste> then we should probably add some guidelines about the qualification task
[00:08] <saste> for example student who do a good task, who interact with the community have more chances of being accepted
[00:10] <saste> i/we can tweak the text once after you write a first draft
[00:10] <llogan> ok. i'll try to get something within the next 3.5 hours
[00:10] <saste> the soon we start the better, since we only have two days and i'm busy 90% of my time
[00:19] <kierank> i don't think mvc is a gsoc worth of work
[00:20] <llogan> we'll bikeshed it until it is
[00:29] <ubitux> wtf
[00:30] <llogan> so many people stopping by and someone hammering next door makes me not write well.
[00:37] <michaelni> kierank i dont mind droping MVC if people think its too easy
[00:37] <michaelni> kierank, btw do you want to mentor something ?
[00:37] <kierank> dunno what i'd mentor
[00:38] <michaelni> you know alot about a lot of things 
[00:38] <michaelni> something you like to mentor, like to see in ffmpeg or think would be important and realistic for a gsoc task
[00:43] <iive> if mvc is too easy, why it haven't been done yet?
[00:43] <kierank> 3d is dead
[00:44] <kierank> and there's a guy doing it afaik
[00:44] <iive> it is not dead, it just smells funny.
[00:44] <iive> ;) ok, if there is somebody working on it, then it is entirely different matter.
[00:45] <michaelni> iive do you want to mentor something ?
[00:46] <iive> nope. i probably won't be able to navigate ffmpeg myself.
[00:47] <michaelni> iive you contributed code in the past and you managed to navigate it ...
[00:48] <iive> i should first contribute some new code, before trying to help somebody else.
[00:49] <saste> damn sent two times the same mail..
[00:49] <llogan> welcome to the club
[00:56] <saste> iive, backup mentors are welcome as well
[00:57] <saste> which possibly mean to guide the student when the poor mentor takes a well-deserved one-week vacation
[01:10] <ubitux> saste: no coverage for lut* filters? :(
[01:10] <saste> ubitux, return AVERROR_PAWE
[01:11] <ubitux> pawe?
[01:11] <ubitux> "Power for Association with Error"?
[01:11] <saste> AVERROR_PATCHWELCOME
[01:11] <ubitux> oh.
[01:12] <ubitux> :(
[01:16] <michaelni> if i hear no objections then ill follow kieranks suggestion and remove MVC
[01:17] <ubitux> i remember seeing a guy talking about mvc recently
[01:17] <ubitux> maybe he has some opinion on the matter
[01:17] <ubitux> michaelni: we don't have much tasks btw, would be nice to replace it with another one if you drop it
[01:19] <ubitux> his nick was "jojva"
[01:19] <ubitux> "< jojva> Well, I'm a student (master's) and with a few people we have this project. The purpose is to film an object with the 3d cam, then transform it into a 3d model"
[01:20] <cone-383> ffmpeg.git 03Carl Eugen Hoyos 07release/0.10:5dd5cfd0b8d3: Only test the first frame for missing aac_adtstoasc bistream filter.
[01:20] <cone-383> ffmpeg.git 03Carl Eugen Hoyos 07release/0.11:8538811a1fef: Only test the first frame for missing aac_adtstoasc bistream filter.
[01:20] <cone-383> ffmpeg.git 03Carl Eugen Hoyos 07release/0.9:402bc31d8787: Only test the first frame for missing aac_adtstoasc bistream filter.
[01:20] <cone-383> ffmpeg.git 03Carl Eugen Hoyos 07release/1.0:fe995dcd84c6: Only test the first frame for missing aac_adtstoasc bistream filter.
[01:20] <cone-383> ffmpeg.git 03Carl Eugen Hoyos 07release/1.1:fc7071cb53de: Only test the first frame for missing aac_adtstoasc bistream filter.
[01:20] <cone-383> ffmpeg.git 03Carl Eugen Hoyos 07release/1.2:9ecfd7daa37d: Only test the first frame for missing aac_adtstoasc bistream filter.
[01:21] <ubitux> and a WIP seems available here: https://github.com/jojva/FFmpeg/commits/master
[01:21] <ubitux> anyway, i have no opinion on whether mvc should be part of gsoc or not
[01:21] <ubitux> but it seems there is some slight interest in it
[01:27] <michaelni> ubitux, if jojva is a student maybe he would want to work on MVC under gsoc ?
[01:27] <ubitux> that's what i wonder yes
[01:30] <llogan> latest message in -devel appears to be a student wanting to add SILK support (i'm not familair with that or know if it is "big" enough for a project)
[01:39] <kierank> that became opus, no?
[01:43] <Skyler_> Yeah, SILK is now part of opus, so
[01:43] <cone-383> ffmpeg.git 03Michael Niedermayer 07master:e370b6589778: tests/tiny_ssim: include the 2 macros instead of a header
[01:44] <cone-383> ffmpeg.git 03Clément BSsch 07master:1a3f7ded147f: doc: "av* tools" -> "ff* tools".
[01:45] <saste> llogan, did you join #gsoc btw?
[01:45] Action: llogan apparently wasn't paying attention to SILK 'n friends
[01:46] <llogan> saste: i've been there twice
[01:46] <saste> llogan, i wonder if candidate mentors need to register before application
[01:46] <saste> but i don't think so
[01:46] <llogan> might be mentined in the FAQ and/or manual(s)
[01:47] <llogan> Skyler_: want to mentor anything?
[01:47] <Skyler_> Um, I'm not sure I'd do a very good job; do we have any idea what the projects are?
[01:47] <llogan> http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_of_Code_2013
[01:48] <Skyler_> oh, cool, APNG!
[01:48] <llogan> of course you can make your own too
[01:49] <Skyler_> oh nice, there's a VP7 spec now :o
[01:50] <ubitux> looks like it's xmas for Skyler_ 
[01:52] <kierank> don't shoot the messenger but it is probably not a good idea for Skyler_ to be involved in gsoc
[01:52] <saste> llogan, #gsoc confirms that no registration is required for candidate mentors
[01:53] <highgod> Hello, for the multi-thread synchronization of OpenCL patch, in the our original design, the thread synchronzation is controled by the user application,so we didn't add the the thread controle in the APIs related to the OpenCL
[01:53] <llogan> saste: ok, thanks.
[01:55] <ubitux> would it make sense to have a filter to randomize linesizes?
[01:55] <ubitux> just like (a)perms, in order to test filters etc
[01:55] <ubitux> typically here in the filter i'm working on, i was assuming that successive frames had the same linesize
[01:56] <ubitux> while it's true in practice, it might not be in some cases
[01:56] <cone-383> ffmpeg.git 03Carl Eugen Hoyos 07master:7857ddceecdd: Fix a typo in the Monkey's Audio demuxer.
[01:56] <ubitux> can anyone think of filters where it would make sense to test such thing?
[01:59] <michaelni> ubitux, maybe you should add a vf_jerk that randomizes everything 
[01:59] <ubitux> :D
[02:00] <ubitux> do you see other things to randomize?
[02:00] <llogan> random curves preset for lols
[02:00] <ubitux> llogan: hehe :)
[02:01] <michaelni> it could inject random command & metadata maybe optionally change w/h
[02:01] <ubitux> oh that would be evil
[02:02] <michaelni> aspect, timestamps, position
[02:02] Action: iive likes the idea
[02:02] <ubitux> let's add a GSoC
[02:03] <ubitux> "be a jerk and write tools to break ffmpeg by all means"
[02:04] <saste> ubitux, curves=bikeshed?
[02:04] <ubitux> saste: i doubt this is going to break much stuff
[02:04] <ubitux> unless you meant something else?
[02:05] <Skyler_> kierank: yeah, one failed project later and after the drama in code-in, I figure it'd be easier if not...
[02:05] <ubitux> saste: i'm still waiting for you pink profile btw ;)
[02:06] <saste> ubitux, really we need a bikeshed curves profile
[02:06] <saste> even if sadly it won't break nothing
[02:06] <ubitux> saste: feel free to propose anything
[02:07] <Skyler_> (and given this is an identity I kind of want to limit the use of, I don't want to have to dedicate hours more a week to it...)
[02:07] <saste> bcoudurier, ping for the gsoc task, mail if also fine
[02:08] Action: saste sleeps
[02:19] <highgod> Hi, michaelni, is there any multi-thread code whitch I can reference?
[02:19] <highgod> in libavutil?
[02:20] <highgod> I have little experience in ffmpeg thread operations
[02:27] <michaelni> highgod, for atomic operations look at atomic.h, for a use example look at buffer.c, for mutex examples just grep for pthread_mutex_lock its used in various parts of avcodec and avformat
[02:32] <funman> school didn't teach me much but grep did.. :)
[02:39] <cone-383> ffmpeg.git 03Carl Eugen Hoyos 07master:92a9a3020d2c: Support pcm_mulaw in Hikvision CCTV mpeg program streams.
[02:46] <michaelni> kierank, I dont see a problem with Skyler_ mentoring if he wants to but saste / llogan are gsoc admins this year so this is just my oppinon also i dont know any details of the "drama" (which isnt my buisness anyway) just that there was some drama
[02:54] <highgod> michaelni, Thanks
[04:00] <cone-383> ffmpeg.git 03d s 07master:444001bda6ad: Fix avisynth_context_create constructor error handling.
[04:00] <cone-383> ffmpeg.git 03d s 07master:914f48ddf022: Update MAINTAINERS with information about avisynth demuxer.
[04:07] <llogan> instead of writing the intro like i should have i made a FFmpeg gsoc logo... beer time and then maybe i'll finish the intro tonight.
[04:35] <ubitux> yay, fieldmatch fixed
[04:37] <ubitux> now i wonder why the the micmatch is mismatching so much..
[04:37] <ubitux> Daemon404: would you mind testing something for me in tfm?
[04:38] <KSV> hi guys
[04:39] <KSV> how many frames are decoded by ffplay in advance
[04:39] <KSV> and can we control this value?
[06:32] <ubitux> Daemon404: with your kyoani_cm_mechagirl.ts sample, i get a lot of still combed frames in the output as soon as the micmatching is enabled (default), whatever the mode
[06:32] <ubitux> OTOH, when micmatching is disabled, it works perfectly fine, not a single combed frame
[06:33] <ubitux> (both vivtc and the ffmpeg's one i'm writing based on it)
[06:33] <ubitux> so i wonder if this is a problem with vfm, or if it's the same with tfm
[06:33] <ubitux> unfortunately, it's hard for me to test
[06:39] <ubitux> Myrsloik_: you might be interested in this btw ^
[06:40] <Myrsloik_> ok
[06:40] Action: Myrsloik_ probably got some condition mixed up
[06:41] <ubitux> do you have the sample?
[06:42] <ubitux> (chromashift.org/<samplename>)
[11:39] <cone-985> ffmpeg.git 03Martin Storsjö 07master:a2acadd0584b: x86: vc1dsp: Fix indentation
[11:39] <cone-985> ffmpeg.git 03Martin Storsjö 07master:3b2d0ec473b0: configure: Remove the mpegvideo dependency from svq1
[11:39] <cone-985> ffmpeg.git 03Michael Niedermayer 07master:ef8ab2f95307: Merge commit '3b2d0ec473b036bdd0a5bc0d896fd5292915f44d'
[11:50] <cone-985> ffmpeg.git 03Martin Storsjö 07master:2ece5bf3ec8b: configure: Remove a stray msmpeg4v1 encoder declaration
[11:50] <cone-985> ffmpeg.git 03Janne Grunau 07master:c24469e81250: utils: add workaround for AVHWAccel in ff_get_buffer compat code
[11:50] <cone-985> ffmpeg.git 03Michael Niedermayer 07master:12f203a09879: Merge commit 'c24469e812501903a46a06eff9722a82e136e841'
[11:59] <cone-985> ffmpeg.git 03Janne Grunau 07master:05fa79b84480: hwaccel: fix use with frame based multithreading
[11:59] <cone-985> ffmpeg.git 03Michael Niedermayer 07master:ace0cfea60a3: Merge remote-tracking branch 'qatar/master'
[12:46] <highgod> how can I use filter options, for example "deshake_options", how can I use command to input it?
[13:07] <ubitux> highgod: deshake=foo=bar:bla=xxx
[13:07] <ubitux> check the documentation
[13:25] <highgod> @ubitux:can you tell more details?which document, I tried your command
[13:25] <highgod> but failed
[13:25] <ubitux> what did you try?
[13:27] <ubitux> http://ffmpeg.org/ffmpeg-filters.html
[13:31] <durandal_1707> ubitux: what is funny in phasescope output?
[13:32] <ubitux> durandal_1707: i got strange cases of different "layouts" depending on the music
[13:33] <durandal_1707> "layouts"?
[13:34] <ubitux> layouts, patterns, ..
[13:34] <ubitux> i'll show you in a moment
[13:38] <durandal_1707> those should be lissajous curves, my goal is to implement something like this: http://www.izotope.com/support/help/ozone/pages/meters_vectorscope.htm
[13:39] <ubitux> durandal_1707: http://ubitux.fr/pub/pics/_phasescope.png
[13:39] <ubitux> got a music which gives me this
[13:39] <ubitux> almost all the song
[13:40] <durandal_1707> well default is really slow fading of previous frames
[13:41] <ubitux> there is no fading problem
[13:41] <ubitux> it's more like the horizontal/vertical patterns
[13:43] <durandal_1707> unless this is because of broken max/min_samples (and thus reading wrong data) there is afaik nothing wrong with filter
[13:44] <ubitux> i'm not saying there is anything wrong
[13:44] <ubitux> i find the pattern funny
[13:44] <ubitux> at least different from some other songs: http://ubitux.fr/pub/pics/_phasescope2.png
[13:44] <ubitux> this one looks more common for instance ^
[13:45] <ubitux> i need to test with more songs :)
[13:45] <ubitux> anyway, i'm looking forward some fancy outputs
[13:45] <durandal_1707> do you like this correlation meter bellow?
[13:46] <durandal_1707> i dunno would i put its fade out as different option
[13:46] <ubitux> with colors it would be even more awesome ;)
[13:46] <durandal_1707> or put it as separate mode
[13:46] <highgod> thanks
[13:46] <highgod> ubitux
[13:47] <durandal_1707> colors? you mean instead of white use green? or multiple colors at same time
[13:47] <ubitux> color gradiants for instance
[13:47] <ubitux> like showspectrum
[13:47] <ubitux> a nice rainbow
[13:48] <durandal_1707> your music is very mono friendly , unlike mine
[13:48] <ubitux> what do you mean? :)
[13:48] <durandal_1707> correlation meter on bottom is 99% on right side
[13:49] <durandal_1707> so when you convert to mono you would not experience strange music
[13:50] <durandal_1707> what is left to do is polar mode (current one is just lissajous) and balance meter
[13:51] <ubitux> i have a more symmetric looking here: http://ubitux.fr/pub/pics/_phasescope3.png
[13:52] <ubitux> zsh: segmentation fault (core dumped)  ./ffplay -f lavfi 'amovie=requiem.flac,asplit[out0],phasescope[out1]'
[13:52] <ubitux> :(
[13:52] <durandal_1707> this is still good
[13:52] <durandal_1707> i usually have very bright left side
[13:53] <ubitux>     at libavfilter/avf_phasescope.c:184
[13:53] <ubitux> 184	            x = ((ptr[1] - (int)ptr[0]) / (float)UINT16_MAX + 1) * hw;
[13:53] <ubitux> want that sample? :)
[13:53] <durandal_1707> what happens?
[13:53] <ubitux> crash
[13:54] <ubitux> almost immediately
[13:54] <durandal_1707> i would prefer patch
[13:54] <ubitux> :D
[13:54] <ubitux> ok give me a moment :)
[13:57] <durandal_1707> how is balance between L & R calculated? it is just some average of sum?
[13:59] <ubitux> i have a lot of invalid read here
[13:59] <ubitux> with any file
[13:59] <ubitux> and use of uninit data
[14:00] <durandal_1707> hmm, than max/min is broken
[14:00] <durandal_1707> or you mean invalid write
[14:00] <ubitux> nope invalid read first
[14:01] <ubitux> uninit:
[14:01] <ubitux> ==27839== Use of uninitialised value of size 8
[14:01] <ubitux> ==27839==    at 0x469F29: filter_frame (avf_phasescope.c:187)
[14:01] <ubitux> invalid read:
[14:01] <ubitux> ==27839== Invalid read of size 2
[14:01] <ubitux> ==27839==    at 0x469EE0: filter_frame (avf_phasescope.c:184)
[14:03] <durandal_1707> x/y calculation is buggy
[14:04] <ubitux> yup :p
[14:05] <ubitux> i'll let you fix your code yourself
[14:05] <ubitux> got some stuff to do
[14:09] <durandal_1707> actually it read too much, so that probably reason why some strange straight lines appears
[14:17] <durandal_1707> ubitux: crash happens because of invalid read or write?
[14:17] <durandal_1707> anyway upload sample that causes crash somewhere
[14:22] <ubitux> ’ pm
[14:27] <durandal_1707> thx
[14:37] <ubitux> mmh
[14:37] <ubitux> + * Copyright (C) 2012 Peng Gao <peng at multicorewareinc.com>
[14:37] <ubitux> + * Copyright (C) 2012 Li   Cao <li at multicorewareinc.com>
[14:37] <ubitux> + * Copyright (C) 2012 Wei  Gao <weigao at multicorewareinc.com>
[14:37] <ubitux> i wonder if the 'C' is not meant to be a 'G'
[14:38] <ubitux> and if so... fear.
[14:40] <durandal_1707> i beat there is at least 10 Wei Gao in china at least
[14:41] <av500> 10?
[14:41] <iive> 10k
[14:41] <ubitux> Gao reminds me the cry of misuzu in Air btw...
[14:42] Action: ubitux wonders if pasting https://www.youtube.com/watch?v=IPbtKMYlV7U on the mailing list will be taken as a bad joke
[14:45] <durandal_1707> those animes all like same
[14:46] <ubitux> i found the "gao~" only once in another anime
[14:46] <ubitux> (’ http://lolicri.es/#touch-minami-asakura)
[15:03] <durandal_1707> ubitux: i think i fixed strange patterns and crash
[15:26] <kierank> i wonder who is paying for opencl filtering
[15:27] <ubitux> durandal_1707: indeed that's better
[15:28] <ubitux> durandal_1707: OTOH the sample from my first screenshot now only displays a vertical line
[15:28] <ubitux> not a single pixel outside that vertical line
[15:28] <ubitux> want a sample?
[15:31] <ubitux> i got some pretty outputs with some samples though
[15:31] <ubitux> like this one http://ubitux.fr/pub/pics/_pretty-phasescope.png
[15:31] <durandal_1707> ubitux: perhaps its mono
[15:32] <ubitux> durandal_1707: what do you mean by mono? all channels the are the same?
[15:33] <durandal_1707> yes
[15:34] <ubitux> durandal_1707: Assertion x < p->w failed at libavfilter/avf_phasescope.c
[15:34] <ubitux> want another sample? :)
[15:35] <durandal_1707> yes
[15:35] <ubitux> uploading..
[15:35] <durandal_1707> i couldn't get that with random()
[15:35] <ubitux> i got it with my 3rd test ;)
[15:37] <ubitux> ’ pm
[15:37] <ubitux> it doesn't happen in the first second, you can reproduce easily with ffmpeg
[15:37] <ubitux> ./ffmpeg -i hoc.flac -filter_complex phasescope -f null - triggers it immediately
[16:05] <cone-985> ffmpeg.git 03Michael Niedermayer 07master:a2b58256490d: vsrc_mandelbrot: support coloring the outside white
[16:16] <durandal_1707> ubitux: that file is very loud at end
[16:19] <ubitux> that's the power of sound
[16:19] <ubitux> so loud it breaks softwares
[16:20] <durandal_1707> actually this was of by one overflow
[16:23] <durandal_1707> fixed, added balance mode (dunno if it is correct)
[16:26] <durandal_1707> now should i rotate those one by default, if user wants to overlay several ones it can use transpose filter
[16:27] <cone-985> ffmpeg.git 03Michael Niedermayer 07master:79938a4e9748: mandelbrot: add outer coloring method showing bailouted z
[19:30] <llogan> would it be preferrable for students to use github or whatever for simple pull requests (assuming they finish something)?
[19:31] <ubitux> llogan: whatever they prefer i guess
[19:32] <cone-985> ffmpeg.git 03Michael Niedermayer 07master:0d6e5a17124d: vsrc_mandelbrot: Mandel morphing support
[19:39] <ubitux> hehe
[19:39] <ubitux> michaelni: maybe you could add a julia mode now :)
[19:41] <llogan> is Small FFmpeg Tasks page fairly up to date?
[19:42] <ubitux> michaelni: i'm looking forward a buddhabrot btw ;)
[19:42] <ubitux> llogan: url?
[19:43] <ubitux> llogan: btw, the "Proposing ideas" and various other developers annotation could be dropped from the page
[19:44] <ubitux> michaelni: https://upload.wikimedia.org/wikipedia/commons/5/51/Nebulabrot.jpg :)
[19:47] <michaelni> ubitux, patch welcome :)
[19:48] <durandal_1707> there is bunch of different fractals
[19:48] <durandal_1707> so separate filter for each of them is littly sillz
[19:49] <ubitux> michaelni: i don't want to steal the fun :p
[19:50] <ubitux> a pretty animated nebulabrot would be nice though
[19:52] <durandal_1707> there is xaos, with bunch of assembly i think
[19:52] <llogan> ubitux: http://wiki.multimedia.cx/index.php?title=Small_FFmpeg_Tasks
[19:52] <llogan> i was waiting (hoping) for bcoudurier to add something
[19:52] <durandal_1707> rewrite swscale
[19:53] <llogan> ..before archiving Proposing to the page discussion 
[20:04] <michaelni> J_Darnley, do you want to mentor something in ffmpeg google summer of code ?
[20:06] <michaelni> TimNich, Tjoppen, mateo`_ if you want to mentor something please add yourself to the gsoc wiki page
[20:08] <llogan> or ask me or saste and we can add anything you want if you don't yet have an account
[20:08] <durandal_1707> you repeated this at least two times
[20:08] <durandal_1707> nobody wants to mentor
[20:09] <J_Darnley> michaelni: I'm not sure I'm experienced enough but I will have a look at the ideas list
[20:09] <llogan> i think the ideas page is much better than last years
[20:10] <michaelni> J_Darnley, thx
[20:14] <ubitux> frame_rate: 0/0 | time_base: 1/1
[20:14] Action: ubitux wonders how useful these fields are
[20:15] <ubitux> (context: lavfi filter links)
[20:16] <ubitux> and in my case the input is supposed to be cfr
[20:42] <J_Darnley> michaelni: I could manage to mentor someone about their interactions with FFmpeg and what we would expect from them...
[20:43] <J_Darnley> but I don't think I could judge how well they were doing their task
[20:44] <J_Darnley> The proposed tasks are above my level of experience with codecs
[20:56] <michaelni> J_Darnley, i was thinking that you maybe would want to mentor something libavfilter related, also google pays 500USD to each mentor IIRC
[20:57] <durandal_1707> i doubt that
[21:02] <michaelni> durandal_1707, http://www.google-melange.com/document/show/gsoc_program/google/gsoc2012/faqs
[21:02] <michaelni> "Google will provide a stipend of 5500 USD per accepted student developer, of which 5000 USD goes to the student and 500 USD goes to the mentoring organization."
[21:06] <ubitux> isn't this only if the student succeed?
[21:07] <michaelni> ubitux, probably yes,
[21:07] <michaelni> s/,//
[21:08] <michaelni> but then it wont be much work for the mentor if the student just disappears or does nothing ...
[21:09] <ubitux> sure
[21:09] <ubitux> btw, if the frame rate was correctly injected in lavfi, we could make select and similar filters adjust pts properly 
[21:10] <ubitux> that would be nice...
[21:10] <ubitux> with cfr of course
[21:10] <michaelni> ubitux, do you have a testcase ?
[21:10] <ubitux> i have various use case
[21:11] <ubitux> first one is extracting pictures from a video with select
[21:11] <ubitux> it doesn't work unless you -vsync cfr for example
[21:11] <ubitux> another example is the decimate filter i'm writing currently
[21:11] <michaelni> do you have a command line, sample file and such so i can see where it fails ?
[21:12] <ubitux> i'll come up with one
[21:12] <ubitux> give me about an hour, i'd like to do some tests before
[21:12] <michaelni> ok
[21:27] <ubitux> michaelni: the output of ffmpeg -lavfi testsrc -t 60 -vsync cfr test-cfr.avi is considered to be cfr, right?
[21:28] <ubitux> and same with other container such as mpg or mkv?
[21:29] <michaelni> ubitux, looks like cbr
[21:30] <michaelni> cFr :)
[21:32] <ubitux> mmh, it might be a problem in my code
[21:32] <ubitux> the frame rate is indeed well populated
[21:40] <ubitux> michaelni: ok, got it; with ffplay it's not set
[21:41] <ubitux> michaelni: with http://pastie.org/7142307 you can compare "./ffmpeg -i test-cfr.avi -vf select -f null -" vs "./ffplay test-cfr.avi -vf select"
[21:55] <michaelni> ubitux, patch posted
[21:56] <ubitux> wow great
[21:56] <ubitux> michaelni: isn't this code shared with ffmpeg?
[21:57] <michaelni> could be factored out yes
[21:57] <ubitux> i'd be happy to such such magical code factorized
[21:58] <ubitux> a lavf util function or such
[21:58] <ubitux> and eventually, automatically done by lavf in one random field
[22:00] <michaelni> ill factor it out
[22:00] <ubitux> anyway, thx for the patch, that will ease my testing
[22:02] <ubitux> seems to work fine
[22:02] <ubitux> in[tb:1/90000 rate:30000/1001] out[tb:1/90000 rate:30000/1001]
[22:02] <ubitux> just as expected, thx
[22:46] <ubitux> my ivtc branch should be ready for testing for those who want to play with it
[22:47] <ubitux> i need to do some more work on the decimate filter, but it's working
[22:58] <ubitux> michaelni: the frame doesn't seem to be used at all for guessing the rate; is this supposed to change later?
[22:59] <ubitux> same for the format
[23:00] <ubitux> sorry for reviewing on irc; maybe i should post on the ml instead
[23:00] <michaelni> i thought it could become usefull in the future saving us from s func2()
[23:01] <michaelni> also its similar to the aspect guessing func
[23:01] <ubitux> while i can imagine some potential improvements based on format, i wonder how the frame could be help
[23:06] <durandal_1707> ffplay -f lavfi -i "amovie=/tmp/hoc.flac,asplit=3[a][b][c],[a]phasescope=s=400x400:d=20[o0],[b]phasescope=m=balance:d=25:s=20x400[o1],[c]phasescope=m=correlation:d=25:s=20x400[o2],[o0][o1]overlay[t0],[t0][o2]overlay=380:0"
[23:08] <ubitux> oh
[23:08] <ubitux> that's nice
[23:08] <ubitux> with a split=4 to listen at the same time that's perfect :)
[23:14] <durandal_1707> now imagine frame full of such meters
[23:14] <durandal_1707> vu, ppm, sco and probably others left to write
[23:20] <cone-985> ffmpeg.git 03Michael Niedermayer 07master:460d30406bca: avformat: change some seeking related assert to av_asserts
[23:20] <cone-985> ffmpeg.git 03Michael Niedermayer 07master:e9d3b4005293: snow: drop commented out asserts
[23:22] <ubitux> http://lucy.pkh.me/phasescope.webm
[23:22] <ubitux> for those who are too lazy to check out durandal_1707's branch ^
[23:23] <ubitux> i like the 'V' appearing around 34sec
[23:29] <ubitux> 'night
[23:47] <michaelni> ubitux, a bit early ;)  but n8
[23:59] <cone-985> ffmpeg.git 03Dale Curtis 07master:81e85bc95cb1: Fix heap-buffer-overflow in matroska_parse_block
[00:00] --- Thu Mar 28 2013