[FFmpeg-devel] [PATCH][RFC] variable frame sizes

Michael Niedermayer michaelni
Mon May 25 23:48:39 CEST 2009


On Mon, May 25, 2009 at 11:23:36AM -0700, Eric Buehl wrote:
> I see...
> 
> This brings up an interesting question:  what should cropping do in the
> event of a frame size change?  Here are some options I see:
> 
> 1: Blindly keep cropping the same absolute values.  This will require
> re-checking of crop boundaries to ensure we can't over-crop as is done in
> opt_frame_crop_top, opt_frame_crop_bottom, etc and may result in erroring
> out mid-encode.
> 
> 2: Scale of the crop values based on the amount of change from the previous
> frame size.  This would be as if all frames were first pre-scaled to a
> normalized size, then cropped, then scaled again to their output frame size.
> 
> I would tend to vote for the latter approach since it should never error out
> and will result in visually consistent cropping in the output.  This would
> also require scaling padding in the same way.
> 
> What do you think?

as this stuff will be migrated to libavfilter scale/crop i dont care too much
about what is done now as long as it does not randomly mix variables from 2
streams and then is exploitable ...

In libavfilter it also would be easy to pass a parameter to switch between 
1. and 2. well it would be easy too in ffmpeg.c but i prefer to keep the
amount of code that needs porting to libavfilter small.

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

He who knows, does not speak. He who speaks, does not know. -- Lao Tsu
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090525/b4e42e63/attachment.pgp>



More information about the ffmpeg-devel mailing list