[Ffmpeg-devel-irc] ffmpeg-devel.log.20191129

[burek]

[00:06:31 CET] <cone-922> ffmpeg 03Andreas Rheinhardt 07master:ff5c8e57e756: avformat/vividas: Avoid allocation of AVIOContext
[00:06:33 CET] <cone-922> ffmpeg 03Limin Wang 07master:c1ed00fd185e: avfilter/vf_yadif: rename config_props -> config_output, link -> outlink
[00:35:34 CET] <Illya> there's no reduce add instruction on x86 right? if you want SAD then something like psadw xmm0, xmm1; psadw xmm2, xmm3; paddw xmm0, xmm1; paddw xmm2, xmm3; paddw xmm0, xmm2; movhlps xmm1, xmm2; paddw xmm0, xmm1; movd eax, xmm0
[02:23:43 CET] <iive> Illya, I think there is instruction that does this directly check if psadbw is the one.
[02:24:45 CET] <iive> though it is byte. it looks like you want to work with words
[03:08:25 CET] <Gramner> psadbw to sum 8 bytes horizontally as pointed out already, pmaddwd to sum 2 words to a dword horizontally. aside from that you're indeed stuck with shuffle, add, repeat
[03:11:59 CET] <cone-859> ffmpeg 03Jun Zhao 07master:d96558902277: lavf/vividas: check avformat_new_stream() return
[03:11:59 CET] <cone-859> ffmpeg 03Jun Zhao 07master:2952d9c5dce4: lavc/mvha: Check init_get_bits8() for failure
[03:24:06 CET] <Illya> thanks for the info, just trying to do some really basic simd
[10:55:41 CET] <j-b> hello
[10:57:17 CET] <stevenliu> Hi
[10:59:03 CET] <JEEB> arr
[11:00:20 CET] Action: JEEB will attempt to review the MPEG-H thing this week-end, was able to find a free final draft of MPEG-H in ISOBMFF last week or so
[11:01:16 CET] <JEEB> although I'm not 100% liking adding demuxing before we have a decoder :P
[11:01:34 CET] <JEEB> because it means that people will utilize SomethingElse for that part
[11:02:01 CET] <JEEB> *MPEG-H Audio
[11:07:17 CET] <nevcairiel> feel free to submit a decoder by monday :d
[11:07:23 CET] <JEEB> yea, nah :V
[11:08:55 CET] <JEEB> anyways, have a few things I need to finish up / post
[11:08:57 CET] <JEEB> :s
[11:15:54 CET] <Lynne> is the channel count in 3d audio fixed or can it change?
[11:30:36 CET] <JEEB> Lynne: I think they changed the ISOBMFF mapping to have less single extradata due to that
[11:30:50 CET] <JEEB> the initial mapping had the decoder config in a box, but now they just say it's in-band
[11:43:25 CET] <Lynne> actually, they kind of need arbitary number of streams, don't they?
[11:44:17 CET] <Lynne> so how would that work? lavf would create one stream for each element, with its own side data to signal how to compose it
[11:45:11 CET] <JEEB> I would have expected arbitrary number of channels rather than streams? but I have no idea how the ambisonics API would work :P
[11:45:20 CET] <JEEB> I've only seen it in elenril/koda's WIP repo
[11:45:32 CET] <JEEB> and even then only knowing of its existence, not how it works
[11:50:58 CET] <nevcairiel> there will definitely need some API extensions for proper 3D support, nevermind that someone needs to make a mixer to actually be able to play it :p
[11:53:52 CET] <JEEB> yes
[11:56:38 CET] <Lynne> JEEB: codecs don't generally support arbitrary number of channels
[11:59:11 CET] <Lynne> aac could support 48 channels with a weird pce but you'll definitely need side data to make sense of the channel mapping
[17:31:08 CET] <cone-161> ffmpeg 03Paul B Mahol 07master:3a61297a672e: avfilter/vf_hqdn3d: add support for 12bit and 14bit yuv formats
[17:31:09 CET] <cone-161> ffmpeg 03Paul B Mahol 07master:8c2f81a17a21: avfilter/vf_hqdn3d: add support for commands
[20:01:10 CET] <cone-161> ffmpeg 03Andreas Rheinhardt 07master:2205fb281074: avformat/avio: Use ffurl_closep
[20:01:11 CET] <cone-161> ffmpeg 03Zhao Zhili 07master:19956d01ccf1: libavdevice/lavfi: check avfilter_graph_dump return value
[22:25:53 CET] <cone-161> ffmpeg 03James Almer 07master:de557eb19156: fate/lavf-container: add an AV1 mp4 remux test
[22:25:54 CET] <cone-161> ffmpeg 03James Almer 07master:13ed24359400: fate/lavf-container: add an H264 mp4 remux test
[23:01:33 CET] <cone-161> ffmpeg 03James Almer 07master:378065f08fbc: avcodec/av1_parser: export stream dimensions in avctx
[23:01:34 CET] <cone-161> ffmpeg 03James Almer 07master:0d67af34c2b2: fate/demux: add an AV1 Annex B test
[00:00:00 CET] --- Sat Nov 30 2019