[FFmpeg-devel] [PATCH 2/2] avformat/movenc: Support merging h264 fields into field pairs
Yusuke Nakamura
muken.the.vfrmaniac at gmail.com
Mon Mar 10 21:07:00 CET 2014
2014-03-10 20:15 GMT+09:00 Michael Niedermayer <michaelni at gmx.at>:
> On Mon, Mar 10, 2014 at 01:18:16PM +0900, Yusuke Nakamura wrote:
> > 2014-03-10 12:48 GMT+09:00 Timothy Gu <timothygu99 at gmail.com>:
> >
> > > On Mar 9, 2014 6:59 PM, "Michael Niedermayer" <michaelni at gmx.at>
> wrote:
> > > >
> > > > Fixes Ticket3393
> > > >
> > > > Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
> > > > ---
> > > > libavformat/movenc.c | 12 ++++++++++++
> > > > libavformat/movenc.h | 3 +++
> > > > 2 files changed, 15 insertions(+)
> > >
> > > Missing documentation:
> > >
> > > Set to 1 merge field pairs into one packet. This option is only valid
> for
> > > H.264 codec. If the codec is not H.264, then this option would be
> silently
> > > ignored.
> > >
> > > Also I wonder if it's a good idea to print a warning if this option is
> set
> > > but codec isn't H.264.
> > >
> > > Does this work with stream copying? If so, then more documentation is
> > > needed.
> > >
> > > How about HEVC?
> > >
> > > [...]
> > >
> > > Timothy
> > > _______________________________________________
> > > ffmpeg-devel mailing list
> > > ffmpeg-devel at ffmpeg.org
> > > http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> > >
> >
> > WTF!?
> >
> > This patch allows to generate files which violate 14496-15.
> > According to 14496-15, a sample is an access unit.
> > An access unit can contain only one primary coded picture.
> >
> > What spec allows this?
>
> The quicktime spec probably does allow it. At least i dont see
> anything in it that would disallow it.
>
> And AFAIK some encoders like mainconcept use it, i dont have that
> encoder so my knowledge here is based on people who do ...
>
> Do you object to this patch or have suggestions on how it can be
> improved ?
>
> [...]
> --
> Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
>
> No great genius has ever existed without some touch of madness. --
> Aristotle
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
>
I object to this patch unless any spec of derived file format based on
14496-12 or quicktime allows it.
Do some encoders like Mainconcept write their own brand(s) in ftyp?
If not, it is just a violation of the spec.
If so, the option should be available only under their brand(s).
More information about the ffmpeg-devel
mailing list