Hi Stefan,
I'm not an expert in color spaces, but I suppose it's perceptual uniformity
of CIELAB that makes it a good choice for editing photos. CIELUV is
probably another good choice. The relations between color variables a*, b*
and the tristimulus values X, Y, Z are nonlinear to account for our
perception of color differences for various colors. While I suppose it is
possible to design other perceptually uniform colorspaces based on XYZ, it
sounds more like a research topic and is probably beyond the scope of the
darktable project. Simply rotating a* and b* axes is not going to give a
perceptually uniform colorspace (the rotated colorspace will not be as good
for editing as Lab/Luv are), nor rotating the original chromaticity
variables x, y is going to help.
When you refer to nonlinear adjustments you have probably addressed the
curves' shape. Color contrast module was written when Lab curves were not
yet available, and it's equivalent to applying straight-line a/b curves.
However even straight-line curves transform nonlinear color functions a*,
b*, and their observed effect is by definition nonlinear.
That said I'd like to note that Lab curves are not the only tool to adjust
color in darktable. Color correction module is capable of split-toning
(different color adjustments in highlights and shadows) remains a great
tool. Color zones module allows to adjust saturation, color and lightness
based on hue. It's probably just what you need to adjust greens. I'm a big
fan of both.
On 11 October 2013 10:32, "Stefan Böttner" <virtual...@gmx.de> wrote:
> Hello,
>
> I've had this thought in my mind for a while. I'm referring to adjusting
> color (such as saturation and color contrast), in particular with the
> ab-curves in the tone curve module. When I change these, they will in
> general act nonlinearly on the a- and b-channels.
>
> Now what I'm wondering is this: The a- and b-axes are somewhat arbitrary
> within the chrominance place (I'm sure they have been chosen for a reason,
> though). In other words, I could have different axes u and v obtained by
> rotating a and b around the origin (i.e. gray) by some angle phi (this
> should be equivalent to converting to LCh and shifting h by phi), and have
> my nonlinear adjustment curves act on u and v values. Note that due to the
> nonlinearity I cannot achieve the same transformation on a and b directly
> with appropriately different curves. Even with just a linear scaling (color
> contrast module?), it is not as general since arbitrary nonuniform scalings
> in 2d-space require a third parameter as they are described by symmetric
> 2x2-matrices.
>
> Thus I'm effectively proposing a family of colorspaces Luv parameterized
> by phi, of which Lab is a special case (phi=0).
>
> The reason why this came to my mind is that with the a/b-curves it's not
> always easy to get the colors the way I want them. If for example I
> emphasize green (by boosting negative a's), I just don't get the right tone
> for foliage.
>
> Now I'm wondering if this consideration makes sense at all. Maybe it won't
> help at all. Just wondering what y'all think.
>
> Stefan
>
>
>
>
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most
> from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
> _______________________________________________
> darktable-devel mailing list
> darktable-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/darktable-devel
>
>
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
darktable-devel mailing list
darktable-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/darktable-devel
