LAB Channels type

Adrian Page adrian at
Fri Apr 21 14:27:48 CEST 2006

BERGER Cyrille wrote:
>>I did it like that originally, but adrian changed it, which fixed the
>>display so somehow unsigned is right
> That means there is a bug in the display code then, because lcms uses
> CIELab (the FAQ comfirms it
> And in CIELab (as shown here a*
> and b* are signed. As opposed to ICCLab (mostly used in tiff files)
> where a* and b* are translated to be unsigned. But I will wait for
> Adrian answer :)

a and b are signed, however lcms encodes them as unsigned shorts using 
an offset. We're using these encoded values directly, so we have to 
treat them as unsigned and decode them to signed when presenting them to 
the user.


More information about the kimageshop mailing list