Casper Boemann cbr at
Tue Jul 13 02:07:40 CEST 2004

> Casper Boemann wrote:
> I think we are on to something here. I think that we should discuss the
> pixel representation and how it relates to channels, channelinfo and the
> datatype of channel.
> If we aggree on this (even though we may not implement it straight away)
> then we would be able so understand better the iterators.

First the subject of this thread is KisPixel, simply because
KisPixelRepresentation is such a long word. I don't put anything into the
change of word.

Do we agree that a channel is one component of a pixel - like red in RGB, or
imaginary in a complex, or hue in HSV.

Do we agree that we call those colorchannels

Do we agree that alpha is also a channel. How many alphachannels can we
have - more than one per colorchannel  (possible a single one for all
colorchannels). More than that ?

Do we agree that we call alphachannels well alphachannels and that they have
a slightly different meaning than colorchannels.

Do we agree that each channel (in principle) could be of a different type.
Say in HSV the H is int16, but S and V are both int8. And that we don't know
what the future may bring of formats but that we should be prepared.

Do we agree that all of this should be hidden in a colorstrategy, possible
(even likely) in collaboration with KisPixel.

Do we agree that ChannelInfo tells us something about the channels for
userinterface purposes. Like the name and more.

Do we agree that it should be possible to turn on and off individual
channels for viewing purposes ?

Do we agree that this is best handled by the color strategy as it might
involve some math to show the channel in a manner that is suitable for a rgb
monitor, and that this math could be colormodel specific. (How would you
show HSV with the H channel turned off - only the colorstrategy would know)

ok that will do for now - everybody please feel free to answer

best regards
Casper Boemann

More information about the kimageshop mailing list