Colorspace inheritance, Lab, Basic ect

[Boudewijn Rempt]

On Sat, 10 Dec 2005, Casper Boemann wrote:

> Hi
>
> Following a discussion with Bart on IRC we have come up with some thoughts on
> how to design the colorspace inheritance.
>
> The superclass is of course KisColorSpace defining the interface and nothing
> else.

Right.

> Subclass to that we have KisLabColorSpace which uses lcms to implement
> everything from scratch.
>
> And Subclass to that we have KisBasicLcmsColorSpace which provide default
> implementations for every other ColorSpace that is represented in lcms.
>
> Now there are a couple of colorspaces that don't have lcms types to go along.
> They need to implement everything from scratch as well, but they could reuse
> KisLabColorSpace in some cases by first converting to Lab, then call labCS to
> do the work and then convert back.
>

Right, too.

> The u16 etc helper classes we should think about some more. Perhaps
> inheritance isn't the best way to leverage the power. Anyway they are fast
> loosing their contents these days.

The big thing there was the alpha channel.

Boudewijn