[Ffmpeg-devel-irc] ffmpeg-devel.log.20180210
burek
burek021 at gmail.com
Sun Feb 11 03:05:03 EET 2018
[01:23:30 CET] <feliwir> does av_seek_frame seek my custom IO context? When i am seeking my IO context manually aswell things don't work anymore
[01:23:35 CET] <feliwir> so i guess that shouldn't be done?
[01:35:06 CET] <rcombs> tmm1: I just need a sample
[01:46:00 CET] <cone-809> ffmpeg 03Muhammad Faiz 07master:909e00ae816d: avformat/format: temporarily use old next api
[01:47:52 CET] <tmm1> rcombs: try https://s3.amazonaws.com/tmm1/cc.mpg or https://s3.amazonaws.com/tmm1/h264.ts
[01:59:05 CET] <rcombs> thanks
[02:13:38 CET] <wm4> michaelni, jamrial: can one of you upload the sample here: http://ffmpeg.org/pipermail/ffmpeg-devel/2018-February/224906.html
[02:18:43 CET] <jamrial> wm4: done
[02:19:40 CET] <wm4> thanks
[03:57:48 CET] <cone-809> ffmpeg 03Xiaohan Wang 07master:dd5208aaf5bf: avformat/mov: Fix ctts_index calculation
[05:25:38 CET] <gdpinname> _ ____ _____ ______ _____ _____ _____ ____ _ _ ____ _ _ _____
[05:25:49 CET] Last message repeated 2 time(s).
[05:25:49 CET] <gdpinname> | | / __ \| __ \| ____| | __ \ /\ | __ \_ _/ __ \ | | | |/ __ \| | | | __ \
[05:25:49 CET] Last message repeated 2 time(s).
[05:25:49 CET] <gdpinname> | | | | | | | | | |__ | |__) | / \ | | | || || | | | | |__| | | | | | | | |__) |
[05:25:49 CET] Last message repeated 2 time(s).
[05:25:49 CET] <gdpinname> | | | | | | | | | __| | _ / / /\ \ | | | || || | | | | __ | | | | | | | _ /
[05:26:00 CET] Last message repeated 2 time(s).
[05:26:08 CET] <gdpinname> | |___| |__| | |__| | |____ | | \ \ / ____ \| |__| || || |__| | | | | | |__| | |__| | | \ \
[05:26:08 CET] Last message repeated 2 time(s).
[05:26:08 CET] <gdpinname> |______\____/|_____/|______| |_| \_\/_/ \_\_____/_____\____/ |_| |_|\____/ \____/|_| \_\
[05:26:09 CET] Last message repeated 2 time(s).
[05:26:09 CET] <gdpinname> https://www.youtube.com/watch?v=JfinBRO_pVc&feature=youtu.be
[05:26:11 CET] Last message repeated 2 time(s).
[05:26:11 CET] <gdpinname> CALL 315-505-4666
[05:26:12 CET] Last message repeated 2 time(s).
[05:26:12 CET] <gdpinname> gix- c3r1c3-Win lucaswang jrmuizel ilove11ven durandal_1707 mateo` moser RT|Chatzilla philipl jamrial microchip_ relaxed raytiley_ atomnuker michaelni Maxz ldts Maverick|MSG bigpet peloverde Kei_N cone-809 Rathann Nightrose _dev_zero telex jpabq chance83 gagandeep_ gagandeep Tzimmo merbanan tyng oorm PaulCapestany jfmcarreira lrusak aballier BtbN Adarfive Blubberbub Compn laxask reynaldo Chloe j0sh tmatth u0m3 keith Gramner JEEB RushPL kepstin bofh
[05:33:58 CET] <philipl> Does that say LOAD BALANCER? :-P
[05:34:57 CET] <Compn> wat
[05:34:59 CET] <Compn> awesome
[05:35:12 CET] <Compn> hilarious that spammer hit a bunch of my other channels
[05:35:43 CET] <Compn> someone joined his irc server once
[05:35:59 CET] <Compn> and one of the channels made his client join 5000 channels or something and crashed him out haha
[05:36:36 CET] <Compn> also i dunno that ascii is all broken here
[05:48:29 CET] <philipl> LODE RADOHOUR
[05:48:35 CET] <philipl> that makes even less sense.
[05:49:04 CET] <philipl> LODE RADIOHOUR
[05:55:53 CET] <Compn> its a radio show
[05:56:14 CET] <Compn> call in chat kind of thing live youtube stream
[05:56:22 CET] <Compn> public access
[05:56:27 CET] <Compn> more of an american thing
[05:56:34 CET] <philipl> huh
[05:56:39 CET] <Compn> but from the 80s/90s
[05:56:54 CET] <Compn> no one does public access anymore
[09:22:58 CET] <thardin> set /mode +p to get rid of spam
[12:47:00 CET] <DHE> thardin: mode +s is usually better
[14:17:54 CET] <atomnuker> jdarnley: check ml
[15:31:16 CET] <jfmcarreira> which version introduced the following function in libswscale sws_getContext?
[15:31:50 CET] <JEEB> that sounds like a very old version
[15:31:57 CET] <JEEB> as I remember using that function in 2013
[15:33:40 CET] <jfmcarreira> JEEB: so something weird is going on on my app.. Getting error linking against in in ubuntu xenial (while trying to build a debian pkg) but in my gentoo machine everthyng is ok
[15:34:16 CET] <JEEB> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=doc/APIchanges;h=91f41a5020ad31140b75cf149d30b820d62ce652;hb=HEAD
[15:34:36 CET] <JEEB> the only entry with sws_getContext is from 2014 where it's noted that it's not getting removed
[15:34:53 CET] <JEEB> jfmcarreira: I hope you're using pkg-config for getting the flags?
[15:35:46 CET] <jfmcarreira> JEEB: cmake but pkg-config
[15:38:11 CET] <sfan5> haven't verified this myself, but looks like the aptX commits broke building with --disable-muxers
[15:38:12 CET] <sfan5> https://github.com/mpv-android/mpv-android/pull/98#issuecomment-364654277
[15:39:21 CET] <JEEB> lemme test
[15:43:40 CET] <jfmcarreira> JEEB: rookie mistake. I was not requesting for swscale module
[15:44:03 CET] <JEEB> jfmcarreira: yea, that's the classic one :)
[15:44:27 CET] <JEEB> sfan5: replicated
[15:44:52 CET] <JEEB> so it's in the headers and utilized, but not in the library
[15:58:22 CET] <JEEB> right
[15:58:28 CET] <JEEB> the demuxer is under CONFIG_APTX_MUXER
[15:58:34 CET] <JEEB> that should be simple to fix
[16:05:05 CET] <JEEB> ok, confirmed it fixing the build
[16:05:32 CET] <JEEB> will quickly test with disabling demuxers as well
[16:05:42 CET] <JEEB> to see if we have any breakage there, too
[16:08:41 CET] <JEEB> ok, that's fine
[16:12:02 CET] <JEEB> sfan5: thanks for the report, posted a patch on -devel
[16:12:12 CET] <sfan5> nice
[16:54:15 CET] <cone-139> ffmpeg 03Aurelien Jacobs 07master:3e21f285ccde: MAINTAINERS: add myself for aptx.c
[17:21:16 CET] <cone-139> ffmpeg 03Jan Ekström 07master:063be4fe2cc6: lavf/aptxdec: fix the limit on the APTX demuxer definition
[17:22:15 CET] <JEEB> durandal_1707: thanks :)
[17:22:48 CET] <JEEB> sfan5: should be fixed now
[17:23:26 CET] Action: JEEB focuses back on hakumei&mikochi
[17:26:59 CET] <Chloe> michaelni: https://0x0.st/scIh.patch fairly sure this should give the correct output but if you could check it before I send it to the ML that'd be great
[17:56:28 CET] <fjsanchez> I have a question, according to the definition on libavtuil/pixdesc.c AV_PIX_FMT_GRAY10BE is not a bitstream, does it mean it should use 2bytes per pixel?
[17:56:53 CET] <wm4> yes
[17:57:44 CET] <fjsanchez> Ok, but when I try to output rawvideo it is calculating the kb/s as it were a packed bit array
[17:57:58 CET] <fjsanchez> For example, Stream #0:0: Video: rawvideo (Y1[0][10] / 0xA003159), gray10le, 640x480, q=2-31, 3072 kb/s, 1 fps, 1 tbn, 1 tbc
[17:58:14 CET] <fjsanchez> 3072 kb/s
[17:58:20 CET] <wm4> whatever underlying raw codec it uses could still be different from the pix_fmt
[17:58:54 CET] <fjsanchez> I am using rawvideo
[17:59:04 CET] <fjsanchez> what codec would that use_
[17:59:33 CET] <wm4> no idea... although considering that it seems to simply dump the pixel memory, maybe I'm wrong in this case
[18:00:32 CET] <fjsanchez> I tought that this was the case with rawvideo
[18:00:47 CET] <fjsanchez> This is why it doesn't match what I was expecting
[18:00:59 CET] <fjsanchez> *thought
[18:04:32 CET] <atomnuker> yep, bitrate for rawvideo is calculated through ff_guess_coded_bitrate() in libavcodec/utils.c
[18:05:07 CET] <wm4> that's pretty lol
[18:05:24 CET] <wm4> is there any purpose?
[18:06:04 CET] <wm4> or is that just a bug
[18:08:27 CET] <kierank> michaelni: if you have time can you review my patch since you are maintainer
[18:13:05 CET] <fjsanchez> Ok, there is this line in libavcodec/util.c: avctx->bits_per_coded_sample = av_get_bits_per_pixel(desc);
[18:13:18 CET] <fjsanchez> But av_get_bits_per_pixel(desc) will return the bits per pixel, not per sample
[18:13:22 CET] <atomnuker> yes, it should be using av_get_padded_bits_per_pixel()
[18:13:48 CET] <atomnuker> also the correct way to fix it would be to make rawenc.c set the correct bits_per_coded_sample
[18:14:30 CET] <atomnuker> just sent a patch to the ml to fix it
[18:16:13 CET] <fjsanchez> Ok, I will try I have never done it before but I guess it is a good opportunity
[18:16:26 CET] <wm4> doesn't it depend on the codec? there are some other packed codecs which call this
[18:16:35 CET] <wm4> so your change could be wrong for some
[18:22:46 CET] <fjsanchez> With the change on rawenc.c -> Stream #0:0: Video: rawvideo (Y1[0][10] / 0xA003159), gray10le, 640x480, q=2-31, 4915 kb/s, 1 fps, 1 tbn, 1 tbc
[18:23:39 CET] <atomnuker> fjsanchez: sorry, I meant I've just sent a patch to fix it
[18:24:52 CET] <fjsanchez> oh, that's ok
[18:26:18 CET] <fjsanchez> I suspect there might be something else, basically I am trying to output using v4l2 format and gray10 pixel format, and that part works after adding a few lines to libavdevice/v4l2-common.c
[18:27:29 CET] <fjsanchez> But it still is receivign the wrong amoung of bytes I think
[18:27:34 CET] <fjsanchez> Format Video Output: Width/Height : 640/480 Pixel Format : 'Y10 ' Field : None Bytes per Line : 800
[18:27:54 CET] <fjsanchez> I would expect 1280 bytes per line
[18:28:33 CET] <fjsanchez> But I don't know if the problem could be in ffmpeg or v4l2loopback, which is the kernel module that I am using
[18:32:48 CET] <fjsanchez> I get his error: [video4linux2,v4l2 @ 0x7fe514000920] buf_len[0] = 385024 < expected frame size 614400
[18:56:17 CET] <michaelni> Chloe, theres still quite some differnces, for example -formats or -h decoder=msmpeg4v3
[19:30:53 CET] <Chloe> michaelni: ah, right. I thought the formats were fine before since oyu just commented on the '-codecs' but I see now
[19:41:37 CET] <fjsanchez> Can anyone tell me where the AVFormatContext is populated with data?
[19:41:53 CET] <fjsanchez> in a decode stream
[19:44:05 CET] <JEEB> it gets populated with data during various stages of demuxing. as in, when you avformat_open_input() it can have some stuff set to it, but as you read more and more stuff the information can get improved
[19:44:23 CET] <JEEB> there's also avformat_find_stream_info() which specifically starts filling up various information
[19:44:37 CET] <JEEB> that said, there's still a possibility that you get rather major updates to the input avformat context
[19:44:42 CET] <JEEB> such as new streams appearing in MPEG-TS
[19:45:20 CET] <fjsanchez> I see, I am trying to find where the frame_size is set for a rawvideo input
[19:45:55 CET] <fjsanchez> I was lookin in avformat_open_input() but I saw that the context was already an argument of the function, so I was unsure
[19:46:10 CET] <JEEB> yes, you allocate a context first
[19:46:24 CET] <fjsanchez> yes I saw that on the main
[19:46:36 CET] <JEEB> that just initializes the structure and doesn't yet know anything of your input
[19:46:43 CET] <JEEB> you can do stuff like override the IO callbacks at that point
[19:47:36 CET] <JEEB> libavformat/rawvideodec.c is rather simple
[19:48:14 CET] <JEEB> it sets things and creates the stream in read_header()
[19:48:45 CET] <JEEB> after that it just calls av_get_packet()
[19:48:57 CET] <JEEB> in read_packet()
[19:48:57 CET] <fjsanchez> That's great! Thank you, I will start from there
[20:57:55 CET] <tmm1> anyone interested in a consulting gig to implement aac ssr ^^
[21:10:52 CET] <rcombs> wtf is SSR
[21:19:08 CET] <tmm1> scalable sample rate
[21:19:14 CET] <tmm1> see ticket above
[21:24:19 CET] <tmm1> used widely in singapore dvb, and also some stations in the UK
[21:27:13 CET] <JEEB> funky
[21:31:01 CET] <atomnuker> never seen it anywhere
[21:54:40 CET] <rcombs> I still dunno what it _is_, though
[22:40:47 CET] <kierank> tmm1: we use SSR?
[22:40:51 CET] <kierank> Not sure about that
[22:59:12 CET] <JEEB> ah
[22:59:46 CET] <JEEB> the zlib check isn't pkg-config so you have to add extra include/lib search paths with cross-compilation (with a prefix that isn't sysroot)
[23:03:46 CET] <feliwir> do you see any special about this file? https://www.dropbox.com/s/u5f1iavdvzuv0vc/vp6log.txt?dl=0
[23:05:26 CET] <feliwir> i found an entire family of vp6 files that does have artifacts with ffmpegs decoder
[23:05:32 CET] <feliwir> in comparison to the On2 decoder
[23:06:36 CET] <tmm1> kierank: are you in uk or singapore?
[23:06:54 CET] <kierank> Uk
[23:07:19 CET] <JEEB> feliwir: I bet that stuff doesn't get too much love so samples are welcome
[23:07:43 CET] <JEEB> (and if there would be someone caring about that format that would be even better)
[23:09:29 CET] <feliwir> JEEB,
[23:09:30 CET] <feliwir> https://imgur.com/a/bD15k
[23:09:54 CET] <feliwir> it's getting even worse
[23:10:01 CET] <feliwir> when more movement happens
[23:12:42 CET] <feliwir> where can i upload a sample file?
[23:18:18 CET] <feliwir> added a couple of sample images
[00:00:00 CET] --- Sun Feb 11 2018
More information about the Ffmpeg-devel-irc
mailing list