[FFmpeg-devel] About guess_correct_pts / AVFrame.best_effort_timestamp

[Reimar Döffinger]

On Fri, Feb 18, 2011 at 07:24:30PM +0100, Nicolas George wrote:
> A parser is a partial decoder, that interprets the bitstream just as much as
> necessary to extract some information about frame boundaries and reordering,
> without doing any actual pixel decoding.

Currently only the first, with what is wished here it would also
have to implement the reordering algorithm (which in turn
requires more parsing than currently done).

> Is that right? Is there any piece of information that a parser
> computes/returns that a decoder does not compute/return?

IIRC our decoders can't really do the splitting to frame
boundaries (should be trivial to make them return any
information necessary though), and the parsers also
sometimes create the extradata in a format that the
decoder needs.