[FFmpeg-devel] [PATCH] mov: support for Quicktime CLEF/TAPT atoms

Nicolas George nicolas.george at normalesup.org
Tue Jul 3 17:16:35 CEST 2012

Le sextidi 16 messidor, an CCXX, Tim Nicholson a écrit :
> So if there is only one pair of width, height fields, then the equation
> will always have to hold true. It sounds like what is needed are two
> pairs, one the actual frame size, the other the pixel rectangle for
> which the DAR is relevant, which is roughly what quicktime provides. By
> default these two pairs would be the same but could diverge if required,
> allowing one pair to maintain the defined relationship while the other
> was decoupled from this, but providing the necessary information as to
> the actual pixel dimensions of the coded frame.

And that is exactly why the API usually deals with SAR and not DAR: SAR is
intensive while DAR is extensive.

In other words, with an example, if you have a 720×576 video with DAR=16/9
and a 720×554 region inside it with DAR=1.85, you can compute SAR for both
and find 64/45 or 1.42, which is the same except for rounding errors.

Of course, if the various (size,DAR) pairs you have give different results
when you compute the corresponding SAR, there is something wrong in the
input file, and lavc has to decode, or maybe guess, which one is correct.


  Nicolas George
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20120703/e8e1de43/attachment.asc>

More information about the ffmpeg-devel mailing list