[Ffmpeg-devel-irc] ffmpeg-devel.log.20150910

[burek]

[00:12:47 CEST] <BBB> ubitux: hey, want to review two vp9 patches? :-p
[00:13:09 CEST] <BBB> j-b: pokey
[00:30:52 CEST] <j-b> BBB: pokey
[00:46:04 CEST] <BBB> j-b: see privmsg (if you didnt go to bed yet)
[01:13:47 CEST] <cone-816> ffmpeg 03James Almer 07master:67d81b76b6e8: avformat/avformat: remove dead code
[04:54:13 CEST] <cone-363> ffmpeg 03James Almer 07master:bbc8fcae1dd9: RELEASE: update to 2.8.git
[05:13:00 CEST] <rcombs> ubitux: https://ffmpeg.org/pipermail/ffmpeg-devel/2015-September/178644.html <-- wtb lavfi subtitles
[10:31:24 CEST] <nevcairiel> ubitux: i'll just leave them alone, trying to "help" them by showing them how ffmpeg solved problems usually doesnt do anything positive, so not merging it is
[10:32:51 CEST] <nevcairiel> videolan git seems slow at best, down at worse.. j-b? :)
[10:39:59 CEST] <j-b> nevcairiel: I'll see
[10:40:24 CEST] <nevcairiel> seems to be getting better, still slightly slow though
[10:42:48 CEST] <j-b> nevcairiel: on it, the load of the machine is at 800.
[10:43:12 CEST] <nevcairiel> slightly over the top... :)
[12:14:03 CEST] <durandal_1707> nice libav, very nice
[12:48:51 CEST] <bahador> Hi .. is there any body here?
[12:50:00 CEST] <bahador> i need help about conver a banch of images to video 
[12:50:22 CEST] <bahador> i have a bunch of images. and i want to make video from these images. below is my code for ffmpeg. but the problem is how can i append audio file to filter-complex. with bitrate and..
[12:50:22 CEST] <bahador> ffmpeg \
[12:50:23 CEST] <bahador> -loop 1 -i input0.png \
[12:50:23 CEST] <bahador> -loop 1 -i input1.png \
[12:50:23 CEST] <bahador> -loop 1 -i input2.png \
[12:50:24 CEST] <bahador> -loop 1 -i input3.png \
[12:50:24 CEST] <bahador> -loop 1 -i input4.png \
[12:50:25 CEST] <bahador> -filter_complex \
[12:50:26 CEST] <bahador> "[0:v]trim=duration=15,fade=t=out:st=14.5:d=0.5[v0]; \
[12:50:26 CEST] <bahador> [1:v]trim=duration=15,fade=t=in:st=0:d=0.5,fade=t=out:st=14.5:d=0.5[v1]; \
[12:50:26 CEST] <bahador> [2:v]trim=duration=15,fade=t=in:st=0:d=0.5,fade=t=out:st=14.5:d=0.5[v2]; \
[12:50:27 CEST] <bahador> [3:v]trim=duration=15,fade=t=in:st=0:d=0.5,fade=t=out:st=14.5:d=0.5[v3]; \
[12:50:27 CEST] <bahador> [4:v]trim=duration=15,fade=t=in:st=0:d=0.5,fade=t=out:st=14.5:d=0.5[v4]; \
[12:50:28 CEST] <bahador> [v0][v1][v2][v3][v4]concat=n=5:v=1:a=0,format=yuv420p[v]" -map "[v]" out.mp4
[12:55:35 CEST] <cone-247> ffmpeg 03Vesselin Bontchev 07master:8e28e0721c61: avformat/aa: initialize "header_seed" and "header_key"
[12:57:47 CEST] <cone-247> ffmpeg 03Gregory J. Wolfe 07master:e8c45b96f49d: avcodec/libopenh264enc: Use av_log() to log messages
[13:05:55 CEST] <nevcairiel> BtbN: some guy claimed nvenc doesnt support B frames. . is that something I just missed, or is he full of it?
[13:08:49 CEST] <BtbN> Depends on the settings, in some modes it's indeed I/P only. But it doesn't generaly not support B frames.
[13:09:42 CEST] <BtbN> Normal HQ mode should use B frames.
[13:13:37 CEST] <iive> BtbN: avoid double negatives :)
[13:22:41 CEST] <bahador> ???????????????
[13:23:26 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:ec5b13fbe4b2: avcodec/libopenh264enc: remove useless cast
[13:23:27 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:fb000b0f0f32: avcodec/libopenh264enc: Fix formating to match FFmpegs style
[13:23:28 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:35b7bd25580c: avcodec/libopenh264enc: Remove some unneeded intermediate variables
[13:24:41 CEST] <bahador> Helloooooooo
[13:24:52 CEST] <bahador> i need some help
[13:52:23 CEST] <cone-247> ffmpeg 03Ronald S. Bultje 07master:4bb9dbe4d795: vp9: permanently delete segmentation refmap on any size change.
[13:52:24 CEST] <cone-247> ffmpeg 03Ronald S. Bultje 07master:fd8b90f5f63d: vp9: fix overflow in 8x8 topleft 32x32 idct ssse3 version.
[13:54:14 CEST] <xxthink> how to build graph2dot.c
[13:55:22 CEST] <xxthink> which lib I should use?
[13:55:48 CEST] <xxthink> one of the error message is: undefined reference to `av_pix_fmt_desc_get'
[14:01:49 CEST] <ubitux> xxthink: make tools/graph2dot
[14:04:12 CEST] <xxthink> ubitux, great!
[14:04:17 CEST] <xxthink> thank you very much!
[14:05:10 CEST] <ubitux> i had a patch to make it cute but it wasn't cute enough at that time
[14:12:14 CEST] <xxthink> :)
[15:15:24 CEST] <rcombs> https://gist.github.com/09817ef4010e447859b5 <-- is this evil?
[15:15:46 CEST] <rcombs> the idea is to let `make fate` and `make fate-rsync` work without specifying SAMPLES on the command line
[15:16:15 CEST] <nevcairiel> i think its better to actually let users specify where it lives
[15:16:22 CEST] <nevcairiel> you can setup env variables like FATE_SAMPLES
[15:16:27 CEST] <nevcairiel> or pass --samples during configure
[15:16:33 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:dfd3fd67bbfd: avcodec/libopenh264enc: use AV_LOG_TRACE instead of DEBUG+1
[15:16:34 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:816cf5181e35: swscale/utils: If the source has no alpha then disable alpha blendaway
[15:16:35 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:319898bba254: swscale/utils: handle gray+alpha formats like gray for luma ranges
[15:17:28 CEST] <Daemon404> agree w/ nevcairiel 
[15:18:03 CEST] <nevcairiel> oddly enough the FATE_SAMPLES env variable breaks libav fate
[15:18:08 CEST] <nevcairiel> i should figure that out so its less annoying
[15:19:43 CEST] <nevcairiel> ah they fail and renamed it LIBAV_SAMPLES, and use FATE_SAMPLES in the makefile somewhere, way to make it hard =p
[15:21:53 CEST] <rcombs> nevcairiel: oh, didn't realize it was a configure flag
[15:21:59 CEST] <rcombs> never mind, then
[15:22:25 CEST] <nevcairiel> --samples=<path> of course
[15:22:41 CEST] <nevcairiel> i tend to use the env variable for ffmpeg
[15:24:21 CEST] <Daemon404> nevcairiel, i cant fathom the point of that change
[15:56:00 CEST] <philipl> nevcairiel: https://github.com/xbmc/xbmc/pull/7751#issuecomment-139170355
[15:56:16 CEST] <philipl> This sounds like what you were describing before the rps idx fix.
[15:56:20 CEST] <philipl> but that fix is in 2.8
[16:02:22 CEST] <nevcairiel> possibly, one would hope someone tested with the most up to date code
[16:14:10 CEST] <philipl> nevcairiel: I'll dig into it.
[16:14:44 CEST] <ubitux> anyone with a NUC here?
[16:15:03 CEST] <ubitux> i'm curious if it's able to playback 10-bit 1080p footage
[16:15:56 CEST] <Daemon404> ubitux, i have a NUC, but it's running a head linux server, and is 2 years old
[16:15:59 CEST] <Daemon404> headless*
[16:16:07 CEST] <Daemon404> i3 i think.
[16:16:10 CEST] <Daemon404> (my irc client runs on it)
[16:16:40 CEST] <ubitux> yeah
[16:16:50 CEST] <ubitux> an irc client is slightly less consuming than decoding video though
[16:17:57 CEST] <philipl> I have an ivy bridge nuc. I've been able play 10bit anime (what else?...) on it ok.
[16:18:03 CEST] <philipl> core i3 version
[16:18:18 CEST] <ubitux> http://ark.intel.com/products/75107/Intel-Core-i3-4010U-Processor-3M-Cache-1_70-GHz?q=4010U
[16:18:21 CEST] <ubitux> so with this one?
[16:18:25 CEST] <philipl> yeah.
[16:18:28 CEST] <ubitux> 10-bit 1080p?
[16:18:42 CEST] <philipl> No, the one before that.
[16:18:42 CEST] <ubitux> avx2, it's kinda nice
[16:18:51 CEST] <philipl> so even older.
[16:18:56 CEST] <ubitux> i guess i'll give it a try then
[16:19:08 CEST] <philipl> but I go out of my way to avoid 10bit content.
[16:19:13 CEST] <philipl> so my experience is limited.
[16:19:20 CEST] <philipl> but it worked the times I was forced to use it
[17:10:16 CEST] <philipl> BtbN: veepeenine
[17:31:09 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:aec340c53329: avcodec/bmp: Analyze BGRA files alpha channel to choose pixel format
[17:31:10 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:67fe1a2b61a6: avcodec/cscd: Use BGR0 instead of BGRA
[17:31:11 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:a4d17c9a706b: avcodec/truemotion1: use BGR0/0RGB
[17:31:12 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:21c727285913: swscale/utils: Forward luma range to the cascaded context for alphablending
[17:31:13 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:d730dd8c53f5: swscale: Forward colorspace updates to the 2nd cascaded context in case of alphablend
[17:41:08 CEST] <BtbN> philipl, in vdpau?
[17:59:44 CEST] <cone-247> ffmpeg 03Harshit Mittal 07master:4f7189f09e5c: .gitignore: ignore temp files
[18:25:32 CEST] <philipl> BtbN: No - just wondering how your progress was going.
[18:25:47 CEST] <philipl> also - that url I pointed nevcairiel to.
[18:25:59 CEST] <philipl> The guy says it was hevc problems on Intel - vaapi and dxva2.
[18:26:16 CEST] <philipl> Could you look at it? I'm not in a position to test and I've not seen that stuttering on vdpau
[18:28:19 CEST] <BtbN> I am seeing that stuttering here. Can only test with VAAPI though. Haven't found time to look into it further. I just confirmed it's there. And it's not frame drops/skips.
[18:28:27 CEST] <philipl> Ok.
[18:28:55 CEST] <philipl> maybe nevcairiel can test dxva2 but I think he did all his work on nvidia - maybe he can test intel?
[18:29:04 CEST] <philipl> maybe it's a hardware problem...
[18:32:15 CEST] <BtbN> It only happens with those odd resolutions
[18:32:19 CEST] <BtbN> 4K ultra-wide
[18:32:29 CEST] <philipl> that file isn't normal 4k?
[18:32:34 CEST] <philipl> I haven't had a chance to check it
[18:32:49 CEST] <BtbN> 4096x1744
[18:32:52 CEST] <philipl> ah.
[18:33:03 CEST] <philipl> Maybe Intel hardware can't do that then.
[18:33:13 CEST] <BtbN> Well, it generaly works
[18:36:10 CEST] <philipl> Can you tell if Intel officially support 4096? Maybe it's not something they claim to support so any success is just luck...
[18:37:29 CEST] <BtbN> hm? 4096 is not what's odd about this resolution.
[18:38:05 CEST] <philipl> 1744 divides by 16. Seems pretty reasonable to me.
[18:39:00 CEST] <BtbN> No idea what's happening, but i doubt it's strange that the same stuff happens on dxva and vaapi
[18:39:52 CEST] <philipl> Yeah. Points to something low level - which makes sense if other resolutions work fine.
[19:40:14 CEST] <nevcairiel> philipl: I tested hevc dxva both on nvidia 960 and haswell gpu, don't have anything more recent
[20:01:46 CEST] <philipl> nevcairiel: ok, so the 'fake' intel hevc in that case, right?
[20:35:10 CEST] <ubitux> curl -s 'http://sprunge.us/KMhY' | grep -A6 9999
[20:35:13 CEST] <ubitux> lol wtf...
[20:35:23 CEST] <ubitux> hello darkness my old friend
[20:39:42 CEST] <durandal_1707> ?
[20:40:02 CEST] <ubitux> look at the font tag
[20:40:07 CEST] <ubitux> in the middle of the timing information
[20:40:15 CEST] <ubitux> i mean, like, wtf
[21:17:46 CEST] <cone-247> ffmpeg 03Clément BSsch 07master:4f26c9991245: avformat/subtitles: drop duplicated events
[21:38:21 CEST] <cone-247> ffmpeg 03Clément BSsch 07master:a634649a36d8: avformat/subtitles: make sure we don't drop "duplicated" events from different streams
[21:38:22 CEST] <cone-247> ffmpeg 03Clément BSsch 07master:af924fd9e469: avformat/subtitles: forward log context in ff_subtitles_queue_finalize() for logging
[21:38:23 CEST] <cone-247> ffmpeg 03Clément BSsch 07master:342ead449166: avformat/subtitles: improve ff_subtitles_queue_finalize() doxy
[22:07:17 CEST] <ubitux> http://b.pkh.me/g176.jpg xface codec?
[22:37:17 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:151312890c77: avcodec/truemotion1: Simplify code using AV_PIX_FMT_0RGB32
[22:37:18 CEST] <cone-247> ffmpeg 03Kevin Wheatley 07master:c3cd6dd106b1: avutil/pixfmt: Add additional primaries and transfer characteristic enumerations from ITU-T Rec H.265
[22:37:19 CEST] <cone-247> ffmpeg 03Kevin Wheatley 07master:47570a265ed0: avcodec/options_table: Add CLI options for SMPTE ST 2084 and ST 428-1 transfer characteristics
[22:37:20 CEST] <cone-247> ffmpeg 03Kevin Wheatley 07master:7061a30c372a: avcodec/options_table: Add SMPTE ST428-1 colour primaries (CIE 1931 XYZ) to command line options
[22:37:21 CEST] <cone-247> ffmpeg 03Kevin Wheatley 07master:6d4b394b5d6c: avutil/pixdesc: Add SMPTE ST 2084 and ST 428-1 pixel descriptions
[22:38:36 CEST] <BBB> j-b: still poke
[22:42:53 CEST] <durandal_1707> ubitux: xface is not codec
[23:03:23 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:b5939ba09cfa: avutil/pixdesc: Use "-" in color_primaries_names to be consistent with color_transfer_names
[23:03:24 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:24426172d3fd: avutil/pixfmt: Name "SMPTE ST 428-1" in AVColorPrimaries the same as in AVColorTransferCharacteristic
[23:03:25 CEST] <cone-247> ffmpeg 03Michael Niedermayer 07master:3ffa77d301a4: avcodec/options_table: Fix TRC / PRI typo
[23:07:20 CEST] <cone-247> ffmpeg 03Ganesh Ajjanagadde 07master:5064b0622412: avfilter/f_perms: use the name 's' for the pointer to the private context
[23:07:59 CEST] <ebail> hi guys. I would like to access to queue_len of RTPDemuxContext from my app. The goal is to have my jitter buffer filled enough. I tried to use "av_opt_get_int(inputCtx_->priv_data, "queue_len"..) but did not success
[23:08:23 CEST] <ebail> is it the right way to do it ?
[23:16:24 CEST] <durandal_1707> no
[23:17:42 CEST] <ebail> durandal_1707: how would it be possible. If I am right this information is not exposed
[23:18:43 CEST] <durandal_1707> There is no way, unless you modify code
[23:18:45 CEST] <ebail> durandal_1707: I want to have packet reordering efficient, and so keep datas in my buffer (I am reading it too often)
[23:19:14 CEST] <ebail> durandal_1707: I would be happy to do it. And I think it can be usefull
[23:40:01 CEST] <cone-247> ffmpeg 03Ganesh Ajjanagadde 07master:f6e598afa822: avfilter/vf_transpose: use the name 's' for the pointer to the private context
[23:56:11 CEST] <BBB> ebail: I would do it inside the rtp code, not inside the applicatio
[23:56:19 CEST] <BBB> ebail: isnt this useful to all applications, not just yours?
[00:00:00 CEST] --- Fri Sep 11 2015