[MPlayer-dev-eng] Re: [PATCH] Extradata handling for X-QT over rtsp

[Carl Eugen Hoyos]

Nico Sabbi <nicola_sabbi <at> fastwebnet.it> writes:

> > Atm, if, for example, an AVC stream contains an "esds" atom, or SVQ3 an
> > "avcC" atom, the code will (IMO wrongly) insert this incorrect atom behind
> > bih. The change would only copy correct atoms for the respective fourcc's.
> 
> there's a basic misunderstanding on this subject:

It's correct that I never looked what esds means, but I know that avc1 decoder
needs the content of avcC atom, and SVQ3 decoder searches for SEQH which is the
content of "SMI ".

> esds stands for "elementary stream descriptor" in mov parlance,
> so it applies to every possible codec, not only to mp4v
> (and not even only to aac as demux_mov.c seems to believe),
> so an esds may contain extradata for whichever codec used.

So, in your opinion, code in svn is correct, my patch is wrong?

> > Second change is that needVideoFrameRate() would be called for every
> > quicktime rtsp stream, wich is correct: Other fourcc's than mp4v, SVQ3 and
> > avc1 might play without extradata (as mp4v and SVQ3 do at least for some
> > samples).
> 
> I don't see this change

In svn, needVideoFrameRate() is only called for avc1, SVQ3 and mp4v in X-QT 
over rtsp, but every video content of X-QT (which could be any content of a mov
file) needs this function. The patch changes behaviour: needVideoFrameRate() is
called for every X-QT rtsp payload.

Carl Eugen