[Ffmpeg-devel-irc] ffmpeg-devel.log.20160121

burek burek021 at gmail.com
Fri Jan 22 02:05:02 CET 2016

[00:09:38 CET] <TD-Linux> ffmpeg doesn't detect a tty to enable color?
[00:10:44 CET] <drv> script probably uses a pty
[00:12:31 CET] <J_Darnley> ffmpeg does check stderr
[00:12:58 CET] <J_Darnley> (see libavutil/log.c:125)
[00:57:52 CET] <cone-880> ffmpeg 03Michael Niedermayer 07master:a53fbda9dc92: avcodec/mpeg12enc: Move high resolution thread check to before initializing threads
[02:59:11 CET] <cone-880> ffmpeg 03Michael Niedermayer 07master:19e456d48c90: avcodec/wmaenc: Check ff_wma_init() for failure
[02:59:12 CET] <cone-880> ffmpeg 03Michael Niedermayer 07master:41e07390e04c: avformat/avformat: Replace some references to filenames by urls
[03:37:21 CET] <cone-880> ffmpeg 03Carl Eugen Hoyos 07master:7158ce778f82: lavf/nut: Add forgotten fourccs for YUV4xxP9.
[03:40:30 CET] <cone-880> ffmpeg 03Carl Eugen Hoyos 07master:4f9314ed5835: lavf/riff: Support reading vlc's rawvideo fourccs.
[04:49:54 CET] <cone-880> ffmpeg 03James Almer 07master:8bbb9723508b: configure: add missing library deps for showcqt_filter
[08:01:24 CET] <cone-890> ffmpeg 03Claudio Freire 07master:adc7d2a4ce8f: AAC encoder: check for NaNs/inf in TNS gain
[10:53:32 CET] <durandal_1707> can I get feedback for showwaves patch?
[13:19:27 CET] <cone-469> ffmpeg 03Michael Niedermayer 07master:56c182c4d716: avformat/nutenc: implement deinit()
[13:38:23 CET] <BBB> michaelni: ffmpeg -i fate-date/sgi/lena_rgb48.sgi -i fate-data/sgi/lena_rgb48.sgi -lavfi '[0:v]format=gbrp16le[out1];[1:v]format=gbrp16le[out2];[out1][out2]psnr' -f null -nostats - gives me: Impossible to convert between the formats supported by the filter 'graph 0 input from stream 0:0' and the filter 'auto-inserted scaler 0'
[13:38:26 CET] <BBB> michaelni: why?
[13:40:59 CET] <wm4> missing swscale support?
[13:41:56 CET] <BBB> configuration: --enable-gpl --samples=fate-data --enable-libvpx --enable-libx264
[13:42:01 CET] <BBB> doesnt look like it
[13:46:07 CET] <wm4> I mean, conversion to gbrp16 is probably not supported by swscale
[13:46:26 CET] <wm4>     [AV_PIX_FMT_GBRP16LE]    = { 1, 0 },
[13:46:28 CET] <wm4> yeah
[13:47:59 CET] <michaelni> yes, thats likely the reason
[13:50:05 CET] <kierank> why is the order of those GBR?
[13:50:20 CET] <kierank> cineform uses that too
[13:51:19 CET] <michaelni> btw my fate clients are all down (incompetent electricans in the house) ill start them again once they are gone
[13:51:19 CET] <wm4> (good question)
[13:52:45 CET] <BBB> wm4: ah, I see, ok
[13:52:57 CET] <BBB> kierank: its for h264 and related formats
[13:53:37 CET] <BBB> kierank: rgb colorspace is coded in planar gbr order
[13:53:55 CET] <BBB> kierank: so this way the decoder needs no special case channel order handling for rgb
[13:54:29 CET] <wm4> but why does h264 do this?
[13:54:35 CET] <kierank> subsampling
[13:54:47 CET] <kierank> makes no sense for no subsampled formats like cineform though
[13:55:28 CET] <wm4> I don't think we allow subsampled rgb output either
[14:07:24 CET] <BBB> what do you mean why does it do this?
[14:07:31 CET] <BBB> it decodes planar data into plane 0, 1 and 2
[14:07:39 CET] <BBB> plane 1 is y/g, 1=cb/b, 2=cr/r
[14:07:45 CET] <BBB> it does what the bitstream dictates
[14:07:47 CET] <BBB> brb
[14:08:54 CET] <wm4> yeah, but cr still doesn't have much to do with r
[14:31:55 CET] <cone-469> ffmpeg 03Paul B Mahol 07master:fa04ec728da3: avfilter: add afftfilter
[14:38:44 CET] <nevcairiel> does the order of the components really matter? planar rgb is uncommon enough as it is, its not like another planar rgb format is more popular or anything :D
[14:48:10 CET] <cone-469> ffmpeg 03Paul B Mahol 07master:73c938dd9f3b: doc/filters: fix afftfilt example
[14:53:47 CET] <J_Darnley> I thought it was gbr because g made the most sense to put into the luma plane and retain the most quality.
[14:54:32 CET] <J_Darnley> Make me wonder though.  How far back does it go?  Was h264 the first?
[15:08:19 CET] <kierank> nevcairiel: it matters because my way of figuring out the component order of cineform is just to change the ordering until it looks right
[15:17:45 CET] <JEEB> was aresample=async=1 still a special case?
[15:18:38 CET] <durandal_1707> what for?
[15:20:30 CET] <JEEB> I just remember it being useful with broken or otherwise non-monotonically rising tiemstamps
[15:22:16 CET] <JEEB> I think it was the one that didn't limit the amount of emptiness that it should pad
[15:24:31 CET] <jkqxz> G as "first" component of RGB has been around forever - consider component video and sync-on-green.  Not sure when that got transposed into video formats in-memory, but it totally makes sense than G ends up first there (use identical hardware setup for YPbCr and GBR planes).
[15:48:41 CET] <cone-469> ffmpeg 03Rostislav Pehlivanov 07master:8248b51e0b94: diracdec: add support for 12 bit videos
[16:04:56 CET] <cone-469> ffmpeg 03Michael Niedermayer 07master:b65efbc0f419: avcodec/mpegvideo_enc: Check for integer overflow in ff_mpv_reallocate_putbitbuffer()
[16:07:46 CET] <durandal_1707> come on, try and comment patches
[16:16:31 CET] <BBB> which one& streamselect?
[16:18:04 CET] <durandal_1707> showwaves
[16:21:15 CET] <cone-469> ffmpeg 03Rostislav Pehlivanov 07master:a849ebb54e18: dirac_parser: Improve parsing and silence pointless warnings
[16:24:28 CET] <BBB> durandal_1707: the colors one? thats pretty minor, lgtm
[16:25:55 CET] <cone-469> ffmpeg 03Ronald S. Bultje 07master:8c9103c4aff4: checkasm: add videodsp emulated_edge_mc test.
[16:26:04 CET] <durandal_1707> no, you are supposed to try it and tell how much it is awesome
[16:27:08 CET] <durandal_1707> it major thing like color tv
[16:28:44 CET] <BBB> how do I test it?
[16:28:54 CET] <BBB> ffmpeg -i file.mp3 -lavfi showwaves -f null - ?
[16:29:45 CET] <BBB> ffplay file.mp3 -vf showwaves also doesnt work
[16:31:03 CET] <durandal_1707> ffmpeg and pipe
[16:31:15 CET] <cone-469> ffmpeg 03Rostislav Pehlivanov 07master:7d8891593c28: dirac_parser: add exceptions for Dirac inter parse codes
[16:32:20 CET] <BBB> pipe what?
[16:33:37 CET] <BBB> I dont even see a video stream in the ffmpeg output
[16:34:12 CET] <atomnuker> I usually use "ffmpeg -i file.mp3 -filter_complex <FILTER> -strict -2 -f yuv4mpegpipe - | mpv -"
[16:35:05 CET] <BBB> Output file #0 does not contain any stream
[16:35:21 CET] <BBB> ah, -vf vs -lavfi
[16:35:23 CET] <BBB> so complicated
[16:35:24 CET] <wm4> hm I should get back to implementing "complex" filter support in mpv
[16:36:04 CET] <BBB> how do I get this to work in ffplay?
[16:36:11 CET] <BBB> I mean, video without audio is kind of & researchy
[16:36:44 CET] <wm4> ffplay apparently doesn't have "complex" filter support either
[16:37:22 CET] <BBB> right
[16:37:28 CET] <BBB> Im doing the worst I can think of here
[16:37:42 CET] <BBB> -f nut
[16:37:53 CET] <BBB> with -c:v rawvideo -c:a pcm_s16le
[16:37:58 CET] <BBB> and then pipe into ffplay
[16:37:59 CET] <BBB> that works
[16:38:09 CET] <BBB> I suppose ffm would work also
[16:38:21 CET] <BBB> yeah ffm is fine also
[16:38:30 CET] <BBB> so now I have to test the patch?
[16:41:07 CET] <durandal_1707> yes
[16:41:26 CET] <BBB> it doesnt work
[16:41:32 CET] <BBB> if I say colors=orange, I get something whiteish
[16:41:38 CET] <BBB> if I do the default, its brown
[16:41:54 CET] <BBB> blue gives me very light blue
[16:42:06 CET] <BBB> red gives me pink
[16:44:23 CET] <durandal_1707> BBB: how many channels have audio?
[16:45:03 CET] <BBB> 2?
[16:47:08 CET] <durandal_1707> BBB: put two colors: pink¦cyan
[16:48:48 CET] <BBB> ah, ok, now it makes sense
[16:49:06 CET] <BBB> (it also helps to select a audio file where left/right are not identical copies or nearly-so)
[16:49:19 CET] Action: BBB pats durandal_1707 on the back - good job dude!
[16:50:55 CET] <BBB> so, whos gonna hook this up in ffplay so ffplay -vis showwaves file.mp3 does this?
[16:51:19 CET] <durandal_1707> Original code used only gray and changed brightness if multiple channels where 
[16:51:29 CET] <durandal_1707> present
[16:51:29 CET] <BBB> (as opposed to ffmpeg -i file.mp3 -lavfi showwaves -f nut - | ffplay -)
[16:51:54 CET] <BBB> or -f ffm, or whatever
[16:52:19 CET] <BBB> and then -c:v rawvideo -c:a pcm_s16le or whatever because by default itll compress to mp2/mpegvideo, I AM NOT KIDDING YOU
[16:52:49 CET] <durandal_1707> BBB: well adding visualization should be easy, ffplay already have some code but general filter complex is harder
[16:55:32 CET] <jamrial> atomnuker: mind if i also replace av_clip with av_clip_uintp2 for the 12bit function you just added as part of the 10bit patch you reviewed?
[16:57:40 CET] <atomnuker> jamrial: sure, go ahead, sorry you had to rebase
[16:58:20 CET] <BBB> durandal_1707: patch still fine btw
[17:19:33 CET] <cone-469> ffmpeg 03James Almer 07master:033e7dbd31c5: avcodec/diracdsp: use av_clip_uintp2
[17:27:31 CET] <ubitux> jamrial: you can do better templating for these
[17:28:00 CET] <ubitux> http://blog.pkh.me/p/20-templating-in-c.html "Mixing full functions mechanism and macros"
[17:29:10 CET] <ubitux> it should still be considered constant
[17:31:13 CET] <jamrial> well, i used the same kind of template as other dsp functions in this file
[17:31:55 CET] <ubitux> ok :)
[17:35:36 CET] <[1]merlink> Is there any possibility to set decklink output cards to interlaced?
[17:49:54 CET] <cone-469> ffmpeg 03Rostislav Pehlivanov 07master:053900ca4c72: dirac: add more parse codes from the specifications
[17:49:55 CET] <cone-469> ffmpeg 03Rostislav Pehlivanov 07master:5776344a7bc9: diracdec: fix original Dirac Low Delay profile
[17:59:28 CET] <atomnuker> I have a Dirac Low Delay sample here, would be nice to get that into the samples repo to add a test
[18:00:44 CET] <atomnuker> who did I need to talk to/send email to do that?
[18:05:09 CET] <BBB> atomnuker: if all else fails, ask michael :-p
[19:27:47 CET] <cone-469> ffmpeg 03Michael Niedermayer 07master:1dd11160eafb: fate: Add test for colorkey
[19:59:14 CET] <jamrial> nice, new dca patchset
[20:18:58 CET] <Gramner> BBB: are you ok with the updated "x86inc: Preserve arguments when allocating stack space" patch?
[20:20:23 CET] <BBB> yes
[20:20:29 CET] <BBB> ty
[20:40:16 CET] <Gramner> great, I'll push it later today then
[20:47:41 CET] <durandal_1707> michaelni: why colorkey test have audio?
[20:49:07 CET] <cone-469> ffmpeg 03Paul B Mahol 07master:6ed3a10f644b: avfilter/avf_showwaves: color support
[20:49:43 CET] <michaelni> does it cause a problem ? i didnt think about it but its more "real world" with audio
[20:51:30 CET] <cone-469> ffmpeg 03James Almer 07master:a2e1b6646079: x86/intmath: disable sse av_clip functions when using ICC
[20:59:01 CET] <durandal_1707> michaelni: nope, its just strange
[21:03:36 CET] <michaelni> i think we should maybe have more real world tests
[23:36:15 CET] <cone-469> ffmpeg 03Henrik Gramner 07master:6cbd0fdf2859: x86inc: Be more verbose in assertion failures
[23:36:16 CET] <cone-469> ffmpeg 03Henrik Gramner 07master:b1496008ee17: x86inc: Improve FMA instruction handling
[23:36:17 CET] <cone-469> ffmpeg 03Henrik Gramner 07master:dfe771dc5a09: x86inc: Preserve arguments when allocating stack space
[23:36:18 CET] <cone-469> ffmpeg 03Henrik Gramner 07master:2d60b18cf082: x86inc: Use more consistent indentation
[23:36:19 CET] <cone-469> ffmpeg 03Henrik Gramner 07master:87b587d4fe91: x86inc: Simplify AUTO_REP_RET
[23:36:20 CET] <cone-469> ffmpeg 03Henrik Gramner 07master:d3662777e070: x86inc: Avoid creating unnecessary local labels
[23:36:21 CET] <cone-469> ffmpeg 03Geza Lore 07master:d39c229e54d6: x86inc: Add debug symbols indicating sizes of compiled functions
[23:58:31 CET] <cone-469> ffmpeg 03Michael Niedermayer 07master:509c9e74e548: avcodec/mjpegdec: Check for end for both bytes in unescaping
[00:00:00 CET] --- Fri Jan 22 2016

More information about the Ffmpeg-devel-irc mailing list