Related to:
I’m wondering if a RYB vectorscope (in addition to the RGB one) would not help to adjust image colors harmoniously.
Complementary colors also explained here.
What do you think ?
The vectorscope is not RGB anyway, it’s either CIE Luv or JzAzBz rewritten in polar coordinates hue/chroma. So I’m not sure how 3D primaries (or more D) intersect with this at all. From where I sit, it’s a completely different formalism and mindset.
Thanks for your answer.
CIE Luv or JzAzBz are converted to a RGB color wheel (hue difference of 120° between the three RGB colors).
Picked from PR #9209 (example is Adobe RGB in JzAzBz, with hue ring overlaid):
A RYB color wheel (hue difference of 120° between the three RYB colors) would give a feedback about the colors complementarity of the image (art view).
It’s clear that RYB is not a colorspace, and this cannot be compared to CIE Luv or JzAzBz colorspaces. This is not the goal.
But it could provide a quick feedback about the images in terms which are well known in the art domain.
EDIT Krita doesn’t propose the RYB wheel. My mistake. The Krita plugin Pigment.O plugin does.
There is no such thing as an RGB color wheel. From what I read in the code, the hue ring is an image drawn from a hue gradient in either Luv or JzAzBz. Then, the image buffer is converted to RGB.
But that’s already what JzAzBz in polar coordinates is supposed to do, in a perceptually-even way.
In the current vectorscope I see more or less 120° between red, green and blue hue.
This doesn’t give the same color split / repartition than 120° between red, yellow and blue.
If the RYB color wheel makes sense for art (color harmony), the current vectorscope doesn’t cover that subject.
What I need to check to go any further, here, is how the graph is handling out-of-gamut pixels, because clipping could favour or disfavour some hues.
Then, I need to check how Munsell hues are distributed perceptually. Luv and JzAzBz are optimized for evenness of the delta E (color difference), but it’s highly possible that Munsell hues and other artistic frameworks don’t care about that at all and use more arbitrary/cultural hue splitting.
Attempt to provide a RYB vectorscope in this PR.
It must be highlighted that RYB is not distributed perceptually. It is just a twisted sRGB …
Its intent is just to show the colours of your images in a said “art” color wheel to help the analysis of colour harmony.
Comments are welcome.
@s7habo, you have already used https://paletton.com to make some color grading with darktable.
The live color pickers are now displayed on the vectorscopes.
The PR above introduces an RYB vectorscope which tries to be more consistent with the art colour pallets. Otherwise it follows the same presentation logic as the other vectorscopes (hue, chroma, density), which may not be the best one for that usage.
Do you think such a tool can help ?
I would love to read your comments, critics and/or suggestions to improve it. Thanks in advance.
Absolutely!
As a good control, to what extent one follows a color harmony that makes sense.
But, this makes sense only as a check possibility. So that you can return to the standard setting when needed, because otherwise it gets confusing, especially if you then treat the colors over different modules, which currently use different color spaces anyway, which is already confusing enough.
Also, the watermark module now support png files, so you can load palettes from Paletton or Adobe Color, without coverting to svg and sample them with live pickers, together with the areas of interest in the image.
Then you can match one with the other by using DT powerful color grading tools and masking features.
Agreed.
The workflow you describe works whatever the type of vectorscope, as you have already made your harmony choice on Paletton.
The RYB vectorscope can (should) help analysing the colors present in the image, which is a step before choosing the color harmony.