Re: [Gimp-developer] Attempt to summarize the discussion of my examples of what doesn't work in unbounded sRGB

[Gez <listas ohweb com ar>]

El lun, 21-04-2014 a las 11:49 +0200, Nicolas Robidoux escribió:
> On 21 April 2014 10:47, Teo Mazars <mazarst ensimag grenoble-inp fr> wrote:
> > ... are you really saying that Gradient should be implemented using a
> > non-perceptual color space?
>
> I am sure this has been discussed in more detail elsewhere (didn't a CSS
> committee discuss this somewhat recently?), but the short answer is that
> most people will be happier if gradient is done in a perceptual color space:
>
> http://www.imagemagick.org/Usage/color_basics/#gamma

As a user, I find both useful.
If I need a gradient for a background, I prefer perceptually uniform
gradients. When I need to mask images, linear gradients make more sense.
But that's probably a matter of taste and not only a "technical"
decision.
For that reason, I think that forcing one or the other is insufficient
for catering users needs.
For gradients it makes a lot of sense to add an option so the users can
choose.

It has been discussed that alpha blending in gamma space should be
provided, although it's not "correct", to keep the appearance of legacy
files.
In that particular case, an option is given for people who choose that
appearance, even though it's technically incorrect.
Why not offering an option for technically *correct* gradients?

Anyway, I think it would be much better to make GIMP work properly in
linear space (the way light behaves in real world) and worry about gamma
correction only in the end, for display and export.
If a specific tool needs a temporary flip to gamma and back for whatever
reason, that can be added on top of that (ideally as an option, so no
technical use is hindered by a personal aesthetic preference).
I'm pretty sure that a true linear workflow with some options to
specific gamma corrections can make all the linear/gamma modes go away
from the precision menu, simplifying the UI and making everything more
consistent.

Gez