Re: [Gimp-developer] GIMP useability - choosing linear vs perceptually uniform RGB
- From: Elle Stone <ellestone ninedegreesbelow com>
- To: gimp-developer-list gnome org
- Subject: Re: [Gimp-developer] GIMP useability - choosing linear vs perceptually uniform RGB
- Date: Fri, 01 May 2015 11:26:08 -0400
On 05/01/2015 09:25 AM, Ofnuts wrote:
It depends a bit what the workflow is. If users need to switch all tools
at random between linear and perceptual, this will be complicated.
Complicated or not, it really is needed.
But
if you can categorize workflows into those that are always perceptual,
those that are always linear,
If you have already have in place a framework that allows the *user* to
choose for all operations whether the operation should be done using
linear or perceptually uniform RGB, then you can use this same framework
to set up different presets for specific workflows.
This would be a great help to beginning GIMP users, especially as they
tiptoe into editing HDR scene-referred images. And it would be a great
convenience for advanced workflows, as long as the user always has the
option to override the presets.
and those that are "let the tool choose
what is more appropriate",
Display-referred editing doesn't mean "alway use perceptual",
scene-referred editing doesn't mean "always use linear", and the tool
itself doesn't dictate anything at all. It always depends on the user's
intentions:
When doing scene-referred editing, linear gamma is required for many
operations because light in the real world combines linearly (apparently
in certain dispersal/scattering situations other considerations apply
when emulating real world colors, but I might have misunderstood the
discussion). But what about noise removal, which most of the time is
best done on perceptually uniform RGB? Isn't noise removal part of
scene-referred editing? The real world doesn't have noise speckles.
When drawing a gradient from red to green or from black to white,
considerations of how light and color combine in the real world dictate
"use linear RGB". But artistic considerations often dictate "use
perceptual RGB".
Usually my conversions to black and white begin as radiometrically
correct luminance conversions done in a linear gamma color space. But
sometimes I convert an image to a color space with the lstar TRC (true
perceptually uniform) before converting to "luminance" because I know
the result will be closer to what I want for the particular image.
I use linear gamma editing a lot, not because someone told me it was
radiometrically correct (though that is what caught my attention
initially) but because I found that linear gamma editing OFTEN produces
better results, judged according to my own artistic intentions.
Artistic considerations trump radiometric correctness, and for some
operations there is no "radiometrically correct" way to perform the
operation. "Scene-referred" isn't the end of image editing. It's just a
stage along the way, and nothing says the artist always has to travel
through scene-referred to get to the goal.
But most of the time the journey is quicker if the artist makes
appropriate, *artist-dictated* use of linear gamma color spaces. Of
course this means the artist needs to educate herself before she can
make informed choices. Having good GIMP presets PLUS user ability to
override the preset will make educating oneself much easier.
On 04/30/2015 10:00 PM, Gez wrote:
You know I'm with you regarding giving users more control over how
operations are performed, but tossing buttons for toggling between
linear and perceptual everywhere in the UI is not a proper solution.
It would be extremely confusing, and people would start toggling them
randomly without knowing what exactly they are for, and only a few
people would benefit from it.
If GIMP users choose to randomly change presets just because they can,
they are still learning something because they can see the resulting
changes. Experimenting with the tools is part of mastering the craft.
Best,
Elle
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]