[FFmpeg-devel] [Bounty] Impl. check if input h264 P/B frame ref's frame prior to last I

[Andrey Utkin]

On Mon, 22 Feb 2016 20:22:30 +0000 (UTC)
Carl Eugen Hoyos <cehoyos at ag.or.at> wrote:

> Andrey Utkin <andrey_utkin <at> fastmail.com> writes:
> 
> > 
> > On Mon, 22 Feb 2016 17:23:06 +0000 (UTC)
> > Carl Eugen Hoyos <cehoyos <at> ag.or.at> wrote:
> >   
> > > Andrey Utkin <andrey_utkin <at> fastmail.com> writes:
> > >   
> > > > As it was discussed today on #ffmpeg and #ffmpeg-devel, 
> > > > only IDR frames are considered safe seek points by design 
> > > > of H.264.    
> > > 
> > > Since valid H264 streams without IDR (and without I) frames 
> > > exist, this cannot be true.  
> > 
> > Haven't heard of such streams, but let's take it granted. Why does
> > it mean that above statement is false? If there's no IDR frames,
> > then it is safe to assume that this stream has no seek points
> > guaranteed to be valid (unless we introspect the references between
> > frames). I don't see anything wrong with such statement.  
> 
> Sorry for being unclear:

> Since valid, useful H264 streams with many ("safe") seek points 
> but without I or IDR frames exist, the statement "only IDR frames 
> are safe seek points" can't be true.

What it takes on libavformat API application level to figure out which
non-I/non-IDR frame is safe to seek onto?

See quoted above my phrase about introspection of references. Is it
involved in what you are talking about?