[Bf-committers] Color space and alpha issues

Brecht Van Lommel brechtvanlommel at pandora.be
Sun Dec 18 13:14:14 CET 2011


I've added some extra details about the consequences of choosing
either alpha format.

Both have some annoying consequences, I'm still going back and forth on this...

Any computation time advantage that premultiplied alpha has for alpha
over operations is basically negligible, since that's actually a
really simple and fast operation compared to the many others that
happen in compositing/render. Either format is slower for certain
common operations.


On Sun, Dec 18, 2011 at 9:48 AM, James Ruan <ruanbeihong at gmail.com> wrote:
> The use of premultiplied alpha format is to save some computing time.
> The definition of straight alpha format is clearer that stores alpha
> info in a separate channel. No matter which one is used as uniform
> internal representation, it is required that handling conversion
> between them when necessary. The optimal choice could be make by
> evaluating computing time it consumes when using it. So IMHO
> premultiplied alpha seems to win that.
> 2011/12/15 Brecht Van Lommel <brechtvanlommel at pandora.be>:
>> Hi,
>> Color management and alpha issues are a recurring topic, I've written
>> a wiki page explaining issues with the current code and what I think
>> should be done to solve them:
>> http://wiki.blender.org/index.php/Dev:Source/Image/Color
>> You might notice I'm not really mentioning OIIO/OCIO. It's a nice
>> project to integrate those, but it doesn't solve current issues and is
>> fairly orthogonal to the proposed changes.
>> I might not have time to tackle this for 2.62, but it doesn't
>> necessarily need to be done all in one big change. Feedback welcome,
>> about what you think should be done different, and which alpha format
>> convention should be used?
>> Thanks,
>> Brecht.
>> _______________________________________________
>> Bf-committers mailing list
>> Bf-committers at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-committers
> --
> James Ruan
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers

More information about the Bf-committers mailing list