[FFmpeg-devel] [PATCH] Use an int MotionVector for find_block_motion.

[Reimar Döffinger]

On 7 Jan 2012, at 14:06, Michael Niedermayer <michaelni at gmx.at> wrote:
> On Sat, Jan 07, 2012 at 12:20:38PM +0100, Reimar Döffinger wrote:
>> Using the double variant causes several pointless conversions between
>> double and int.
>> Worse, one of the conversions is in an inner loop together with a
>> function using MMX, resulting in undefined behaviour.
>> In general the typedefs/structs seem overkill and slightly obfuscating
>> to me, but I left that in the assumption that the author/maintainer
>> prefers it.
>> 
>> Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
> 
> LGTM but please mention  Ray Simard in the commit message for finding
> and analyzing the bug 

Sorry, it was not meant for committing, it is completely untested.
I will push once someone (preferably Ray Simard) gives confirmation it works and does not break the filter.