[FFmpeg-devel] [PATCH v6 4/9] avcodec: add cbs for h266/vvc

[Nuo Mi]

On Sat, Feb 20, 2021 at 2:17 AM Jan Ekström <jeebjp at gmail.com> wrote:

> On Fri, Feb 19, 2021 at 3:52 PM Nuo Mi <nuomi2021 at gmail.com> wrote:
> > > >>
> > > >> The current logic is that we are writing an AU, so the first NAL
> unit in
> > > >> it is necessarily an AU start and subsequent NAL units are not?
> > > >>
> > > > H.266 AU contains one or more PU(3.105). One PU contains one coded
> > > picture.
> > > > Only the first NAL unit of an AU needs the zero_byte(B.2.2)
> > > > H.265 has no PU, each AU has one coded picture(3.1)
> > > > H.266's PU is the H.265's AU. In vvc_parser, we split bitstream to
> PU.
> > >
> > > Er, I don't think this is right: an AVPacket should contain all of the
> > > pictures for a single timestamp - a decoder can then output at most one
> > > picture from it, for the selected spatial layer.
> > >
> > This will make parser code very complicate :).
> >
> >
> > > Though I don't know how the container formats are defined here, for
> other
> > > codecs having multiple packets with the same timestamps causes ugly
> > > problems in muxing/demuxing which we avoid by saying you aren't
> allowed to
> > > do that.
> > >
> > We do not have container format spec for vvc yet. The codec is not
> register
> > on http://mp4ra.org/#/codecs. How we handle svc in H.265? It will have
> same
> > timestamp for diffrent AU.
> >
>
> FYI, there are already drafts for 14496-15 (Carriage of NAL unit based
> video in ISOBMFF aka MP4) with VVC such as document N18825 .
>
> https://isotc.iso.org/livelink/livelink?func=ll&objId=21196449&objAction=Open&viewType=1

thanks, Jan, it's very informative.

Hi Mark,
You are right, let me spend time adjusting the code.
>each sample in the track contains an access unit. The access unit contains
NAL units from all the layers and sub-layers that are part of the operating
point.(11.3.4)

>
>
> Jan
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".