Color source for paintops

[Matthew Woehlke]

Cyrille Berger wrote:
> As discussed with Lukas on IRC last week, here is a patch that allow more 
> flexibility on the color source used to colorize the dabs.
> 
> It introduce a new base class KisColorSource, which abstract color sources. 
> There are two types of color source, uniform color sources, where the color is 
> the same on the whole dab (but might be different at each stroke), and those 
> who can have different color for each pixel of the dab. And we want to apply 
> transformation to those color source (mostly color transformation).

Well... I haven't been much for KDE work lately, and I'm not likely to 
be poking this until it is in trunk, at best. However, it reminds me of 
an idea I've had around for some time.

Long time ago I had developed some PS hacks for painting decent-looking 
fur without a huge lot of effort. Partly this involved applying color 
jitter to multi-count brushes (not sure what the correct term is for 
each paint operation, i.e. what happens every <spacing> px, paints the 
brush shape multiple times, with jitter factors applied separately each 
time).

Anyway, what I would really like to do is have a 2D-gradient color 
source, where I can assign various parameters to either axis. (I imagine 
this as a series of normal 1D gradients that act as stops in another 1D 
gradient, if that makes sense; if not, holler, and I can explain 
better.) Parameters would be e.g. time, distance along stroke, pressure, 
stroke angle (stroke angle is actually the one I care about)... and also 
something like 'brush part'. For multi-count brushes, divide the axis 
evenly by the count to get the color for each part.

You could also assign an axis spatially, which would be like the above, 
but with a second axis of variation.

However, the real gold is doing this with bristle brushes, where each 
bristle gets assigned an axis value /randomly/. So it is like dipping 
your brush in a paint swatch that is poorly mixed. But better, it gives 
you instant peppered fur :-).

(Then if you want to get *really* fancy, you add 'cross-contamination', 
such that over time, the paint from one bristle mixes with that of the 
other bristles it touches...)

I do have to wonder if this (stick-with-a-bristle) is how this patch is 
supposed to work, though... and if not, if maybe that should be an 
option :-).

-- 
Matthew
Please do not quote my e-mail address unobfuscated in message bodies.
-- 
Ok, so the quotes aren't entirely original.