[Ffmpeg-devel-irc] ffmpeg-devel.log.20120605

[burek]

[00:19] <CIA-119> ffmpeg: 03Clément BSsch 07master * r13a7bd70c8 10ffmpeg/libavfilter/vf_select.c: lavfi/select: 10l: llabs -> fabs after switching types.
[01:28] <cbsrobot> rabeeh: and btw. what decoding engine is it ?
[01:28] <rabeeh> CuBox is based on Marvell Armada510 chip.
[01:28] <rabeeh> It uses a hardware engine called Vmeta.
[01:30] <rabeeh> Originally it had a gstreamer based libraries (that's the official Marvell release), but then Gimli from team XBMC ported that 10K source code driver to ~700 lines native vmeta driver for XBMC
[01:30] <rabeeh> hardware supports h.264, vc-1, mpeg2, mpeg4v2 and others (like jpeg, mjpeg - i think)
[01:30] <cbsrobot> ok
[01:31] <cbsrobot> maybe you need to repeat your issue here again ...
[01:34] <rabeeh> i'm representing Solid-Run which is a company that developed CuBox; an open source platform www.solid-run.com
[01:34] <rabeeh>  i was wondering if there are developers around that are interested in getting ffmpeg ported to use the hardware decode engine that is on the product.
[01:35] <rabeeh> the decoder ported to XBMC is at -
[01:35] <rabeeh> https://github.com/huceke/xbmc/blob/marvell-dove/xbmc/cores/dvdplayer/DVDCodecs/Video/DVDVideoCodecVMETA.cpp
[01:44] <pasteeater> rabeeh: if you don't get a response here then try the ffmpeg-devel mailing list too. 
[03:09] <CIA-119> ffmpeg: 03Andrew Wason 07master * r1ef802d038 10ffmpeg/doc/examples/ (Makefile filtering_audio.c): 
[03:09] <CIA-119> ffmpeg: update filtering_audio example
[03:09] <CIA-119> ffmpeg: The filtering_audio.c example needs to be updated for the new
[03:09] <CIA-119> ffmpeg: "abuffer" filter args.
[03:09] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:09] <CIA-119> ffmpeg: 03Michael Bradshaw 07master * r1b27b8bf6c 10ffmpeg/libswscale/x86/rgb2rgb_template.c: 
[03:09] <CIA-119> ffmpeg: MANGLEd swscale x86 asm to save registers
[03:09] <CIA-119> ffmpeg: register starvation caused gcc4.2 to fail building 32 bit shared libs
[03:09] <CIA-119> ffmpeg: on 64 bit OS X
[03:09] <CIA-119> ffmpeg: Signed-off-by: Michael Bradshaw <mbradshaw at sorensonmedia.com>
[03:09] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:29] <CIA-119> ffmpeg: 03Clément BSsch 07master * r44be67595d 10ffmpeg/libavfilter/vf_select.c: 
[07:29] <CIA-119> ffmpeg: lavfi/select: use exported dsputil_init function.
[07:29] <CIA-119> ffmpeg: This is the deprecated version of the function but it is exported, so it
[07:29] <CIA-119> ffmpeg: should fix the shared FATE box.
[09:55] <CIA-119> ffmpeg: 03Carl Eugen Hoyos 07master * r757d5b9bfd 10ffmpeg/ (libavcodec/Makefile libavformat/Makefile): 
[09:55] <CIA-119> ffmpeg: Fix MP2 muxer Makefile dependencies.
[09:55] <CIA-119> ffmpeg: Found, analysed and tested by trac user Jamal.
[09:55] <CIA-119> ffmpeg: Fixes ticket #1411
[10:05] <ubitux> http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=788a60d9
[10:05] <ubitux> http://git.libav.org/?p=libav.git;a=commitdiff;h=e58b75f7f
[10:05] <ubitux> mpf.
[10:05] <ubitux> and again.
[11:14] <CIA-119> ffmpeg: 03Carl Eugen Hoyos 07master * r1faf0d6a7a 10ffmpeg/libavformat/avisynth.c: Add missing CRLFs to avisynth error messages.
[11:54] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r1488c4dc0b 10ffmpeg/libavfilter/ (avfilter.c internal.h video.c): 
[11:54] <CIA-119> ffmpeg: lavfi: move tracing utilities definition from video.c to avfilter.c
[11:54] <CIA-119> ffmpeg: These utilities are meant to be defined in avfilter.c, since they are
[11:54] <CIA-119> ffmpeg: generic and thus contain both audio and video code.
[12:14] <CIA-119> ffmpeg: 03Piotr Bandurski 07master * ra489db6cde 10ffmpeg/ (libavcodec/raw.c libavformat/isom.c): isom: add Radius DV YUV FourCCs
[12:15] <CIA-119> ffmpeg: 03Piotr Bandurski 07master * r48fdaa1cff 10ffmpeg/libavcodec/snowdec.c: snowdec: add new line to some error messages
[12:52] <CIA-119> ffmpeg: 03Carl Eugen Hoyos 07master * ra3bc7f916d 10ffmpeg/libavcodec/h264.c: Make H264 reorder buffer size message less verbose.
[13:44] <CIA-119> ffmpeg: 03Nicolas George 07master * r6a4c5c730e 10ffmpeg/libavfilter/avfiltergraph.c: avfiltergraph: check query_formats return value.
[13:44] <CIA-119> ffmpeg: 03Nicolas George 07master * raba8065adb 10ffmpeg/libavfilter/avfiltergraph.c: 
[13:44] <CIA-119> ffmpeg: avfiltergraph: call query_formats on inputs first.
[13:44] <CIA-119> ffmpeg: This is a temporary workaround for amerge,
[13:44] <CIA-119> ffmpeg: until format renegociation is implemented.
[13:44] <CIA-119> ffmpeg: Should fix trac ticket #1400.
[13:48] <ubitux> j-b: it seems vlc is pushing the libavresample wrapper; do you plan to make comparison between libswresample and libavresample?
[13:48] <j-b> VLC is not pushing anything
[13:48] <ubitux> i mean libav is pushing on vlc*
[13:48] <ubitux> yep sorry, bad wording
[13:48] <j-b> we have received both a libswresample and a libavresample patch.
[13:49] <ubitux> yes i remember the libswr one getting flamed
[13:49] <ubitux> is there any plan to make a comparison at some point?
[13:49] <j-b> we do not have plans, but yes
[13:50] <ubitux> i think a few ppl would be interested in an objective benchmark :)
[13:50] <j-b> I think a lot of ppl would love if people stopped fighting...
[13:51] <Compn> i think vlc is angry they have to support two apis
[13:51] <j-b> we are not angry
[13:51] <av500> they are french
[13:51] <Compn> you're french, you're always angry :)
[13:51] <ubitux> j-b: what do you propose, on ffmpeg side?
[13:51] <j-b> so far, I use FFmpeg, because libav does not backport security fixes
[13:51] <ubitux> s/security // :)
[13:51] <j-b> and therefore if I do not have those, Microsoft people from security flame me by mail.
[13:52] <j-b> ubitux: yes, but I care about security
[13:52] <ubitux> ok :)
[13:52] <ubitux> Microsoft is lurking VLC?
[13:52] <Compn> microsoft is trying to fix all software that sends crash reports, i think
[13:52] <j-b> yep, we received 4 advisrories so far
[13:52] <Compn> which is strange :)
[13:52] <j-b> unfortunately, FFmpeg has bugs that are not present in libav
[13:53] <ubitux> Compn: that's an honorable quest actually :)
[13:53] <ubitux> j-b: can you raise them?
[13:53] <j-b> no.
[13:53] <Compn> ubitux : yes, but motive must be questioned
[13:53] <ubitux> j-b: eeh~ :(
[13:53] <j-b> Compn: motives from MSVR are very nice.
[13:53] <ubitux> (j-b: why not?)
[13:54] Action: Compn afk
[13:54] <j-b> ubitux: because it is too complex to reproduce
[13:54] <michaelni> j-b, if you want to benchmark swr, wait a bit, there are some optimizations i planed but didnt do yet
[13:55] <j-b> michaelni: OK
[13:55] <ubitux> j-b: is there that much bugs too complex to reproduce?
[13:55] <j-b> ubitux: this one is quite annoying to be honest.
[13:55] <ubitux> and we can't help?
[13:55] <j-b> ubitux: anyway, what will very likely happen is that both will get merged.
[13:55] <av500> michaelni: x86?
[13:55] <j-b> NEON?
[13:56] <ubitux> j-b: you mean merge libavr and libswr?
[13:56] <michaelni> x86
[13:56] <j-b> ubitux: yes
[13:56] <ubitux> j-b: from where?
[13:56] <j-b> ubitux: http://people.videolan.org/~jb/libavcodec_plugin.dll vs http://people.videolan.org/~jb/libavcodec-libav_plugin.dll in latest vlc 2.0.2 NB
[13:56] Action: av500 does not care much how fast a resample is on quad core i7...
[13:56] <ubitux> j-b: not reproducible on *nix?
[13:57] <j-b> when piped from rtmpdump streams from popular programs like TVo (win32) only, it does not work in FFMpeg's while it does in the libav one.
[13:57] <j-b> and taking the same flvdec does not change a thing
[13:59] <ubitux> ah that's yhe bug you were talking about the other day
[14:00] <j-b> yes
[14:00] <michaelni> j-b, how can it be reproduced on windows then ?
[14:07] <j-b> michaelni: I'll do a proper testing when I have time
[14:08] <michaelni> j-b, thx
[14:08] <ubitux> j-b: anyway, if you have any suggestion on what ffmpeg could do to help the situation... also i'm still a bit surprise about the libavr/libswr merge you were talking about
[14:08] <ubitux> (help the situation refering to the flaming above)
[14:09] <j-b> I have not seen flaming yet
[14:10] <ubitux> the fighting stuff you were talking about
[14:10] <ubitux> with libswr vs libavr i guess
[14:20] <CIA-119> ffmpeg: 03Paul B Mahol 07master * r02827a3c99 10ffmpeg/libswscale/swscale_unscaled.c: 
[14:20] <CIA-119> ffmpeg: swscale: RGBA64LE <-> RGBA64BE support
[14:20] <CIA-119> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[14:43] <michaelni> av500, do you want to donate some NEON hw to me ? i cant promisse i would find the time to actually write NEON code but without means to test/benchmark except qemu iam kinda handicaped ...
[14:43] <michaelni> or donate a NEON developer maybe :)
[14:45] <merbanan> michaelni: what kind of hw would you need ?
[14:46] <merbanan> android phone or pnda board like ?
[14:47] <michaelni> merbanan, I am not ARM expert, but NEON support is mandatory and the easier it is to devlop /test/maintain it the better
[14:48] <michaelni> somoen offered me a non-NEON ARM yesterday so i might get a non-NEON board in a box
[14:49] <merbanan> k, then I think beagle/panda board should be the easiest ones to use
[14:51] <michaelni> fine, do you happen to have a spare one, ive just checked and it seemed the latest beagle bord is out of stock everywhere i checked
[14:53] <av500> it's always out of stock
[14:53] <av500> since they sell every single one they make
[14:53] <av500> so there is never stock
[14:53] <av500> only orders and shipments
[14:54] <merbanan> I gave one to andoma, I'll see if he uses it anymore
[15:48] <Compn> michaelni : someone would donate one if you asked on the list i'm sure
[15:50] <av500> michaelni: if you need an A8 with neon, the beaglebone is ~$90
[15:53] <ubitux> i have a dreamplug iirc
[15:53] <ubitux> armv5 though afaik
[15:54] <ubitux> i can send it to you if you think you can make something out of it
[16:00] <michaelni> merbanan, Compn av500 ubitux ... thilo has a beaglebord XM that he would be happy to send me as he never managed to get it working
[16:01] <av500> michaelni: perfect
[16:01] <Compn> just hope you can get it working :P
[16:01] <michaelni> Compn, yes thats my fear too ;)
[16:04] <av500> Compn: there is no issue to get one working
[16:06] <Compn> av500 : i meant get thilo's board working
[16:06] <Compn> which is reported as != working
[16:08] <av500> well, is its broken, its broken
[16:10] <michaelni> ill try to find out if thilos board is broken, he just spoke of some USB-LAN issue
[16:10] <ubitux> mine should work, but never had the time to get something out of it
[16:10] <ubitux> so if you need one&
[16:11] <av500> usb-lan could be a too old kernel issue
[16:11] <av500> the polarity of the enable pin changed
[16:11] <av500> especially for people that insist on running ubuntu
[16:15] <michaelni> ubitux, yours doesnt have NEON does it ?
[16:16] <ubitux> no idea, never look closely
[16:16] <ubitux> (http://www.globalscaletechnologies.com/t-dreamplugdetails.aspx)
[16:17] <av500> no, its marvell armv5 on stereoids
[16:20] <ubitux> = pointless then?
[16:20] <cbsrobot> yesterday someone asked if we could support his hw: http://www.solid-run.com/
[16:21] <michaelni> cbsrobot, i know i chated with him
[16:21] <ubitux> cbsrobot: kawaii
[16:21] <michaelni> he said he would donate one to me but it has no NEON
[16:21] <michaelni> if i understood him correctly at least :)
[16:21] <cbsrobot> ah sorry - missed that
[16:23] <michaelni> ubitux, if you want one too only thing you need to do is port some xbmc hw decoder to ffmpeg IIRC
[16:23] <av500> cubox is another marvell thing
[16:24] <av500> again, no neon
[16:24] <av500> it has some binary blob video decoder
[16:24] <av500> like they all have these days
[16:24] <ubitux> michaelni: i pass. :)
[16:59] <Daemon404>             "r_frame_rate": "180000/2",
[16:59] <Daemon404>             "avg_frame_rate": "18495000/891763",
[16:59] <Daemon404>             "time_base": "1/90000",
[16:59] <Daemon404> ^ phones generate this un purpose? wtf.
[16:59] <Daemon404> on*
[16:59] <kierank> 1/90000 is pretty standard
[17:00] <av500> its THE standard
[17:00] <av500> http://26-26-54.hardwarebug.org/61
[17:01] <kierank> didn't know i wrote that
[17:01] <kierank> or am on that thing
[17:01] <Daemon404> kierank, thats pretty funny then
[17:02] <Daemon404> [mpeg4 @ 0x25437c0] Invalid pixel aspect ratio 65536/65536, limit is 255/255 reducing
[17:02] <Daemon404> [mpeg4 @ 0x25437c0] removing common factors from framerate
[17:02] <Daemon404> [mpeg4 @ 0x25437c0] timebase 1/90000 not supported by MPEG 4 standard, the maximum admitted value for the timebase denominator is 65535
[17:02] <Daemon404> and then it just refuses to continue
[17:02] <kierank> oh dunno about mpeg4
[17:16] <Compn> its funny everyone uses binary blob, but when ffmpeg gets support, everyone switches ... :)
[17:16] <Compn> guess its more to do with ffmpeg being clusterable
[17:18] <av500> Daemon404: from what file do you get the above values?
[17:20] <Daemon404> av500, 3gp from a 3 year old android phone
[17:31] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r787c395a30 10ffmpeg/libswresample/swresample-test.c: 
[17:31] <CIA-119> ffmpeg: swr-test: support "--help"
[17:31] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:31] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r63b1c08073 10ffmpeg/libswresample/ (Makefile swresample_test.c): 
[17:31] <CIA-119> ffmpeg: buildsys: fix rules for swresample-test
[17:31] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:31] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r5a7a9f02de 10ffmpeg/libswresample/Makefile: 
[17:31] <CIA-119> ffmpeg: libswresample/Makefile: put each object on its own line
[17:31] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:16] <CIA-119> ffmpeg: 03Andrew Wason 07master * rc527027c27 10ffmpeg/libavfilter/ (avfiltergraph.c formats.c): 
[18:16] <CIA-119> ffmpeg: aconvert filter needs to configure samplerates
[18:16] <CIA-119> ffmpeg: I changed the *_set_common_* functions to only set unset formats, then
[18:16] <CIA-119> ffmpeg: added a wrapper that calls them after the filters query_formats.
[18:16] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:45] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r0da9bce5a3 10ffmpeg/libavfilter/ (avcodec.c buffer.c): 
[18:45] <CIA-119> ffmpeg: lavfi: move avfilter_copy_frame_props() definition from buffer.c to avcodec.c
[18:45] <CIA-119> ffmpeg: The new location is more suited, as it is where the lavfi/lavc glue is
[18:45] <CIA-119> ffmpeg: defined.
[18:45] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r62b39d41cd 10ffmpeg/ (4 files in 2 dirs): lavc: add pkt_duration field to AVFrame
[19:02] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r647e2e070f 10ffmpeg/libavfilter/ (src_movie.c vsrc_testsrc.c): 
[19:02] <CIA-119> ffmpeg: lavfi: remove more instances of avfilter_default_filter_name()
[19:02] <CIA-119> ffmpeg: They were discarded in 0f0f3bd1e0.
[00:00] --- Wed Jun  6 2012