Awesome harmony palette

As painter we use not always all colors, we choose some, for practical and aestehic reasons.
Finally we try to find an nice color association between… Colors and tones.

I think a dynamic palette would be great addition for us all, like this:

A) Pick a foreground color with the eyedropper

B) Get the corresponding complementary, tertiary, tris, analogous colors (3 min, maybe better 5) derived from the tonal value and hue of the foreground color picked before

Of course this color harmony system works only efficiently if we are using the RYB color wheel, and not… the odd TV color system, rgb.

1 Like

Have you tried using Gamut Masks? - I know it’s not exactly what you’re looking for, but it can help.
Also you should be able to set up Advanced Color Selector sliders for those sets of colors as well (Configure krita -> Advanced Color Selector -> Color Sliders).

Yes i use gamut masks ,
Every suggestion is welcome

My point is, to get a dynamic color palette just by selecting a color with the eyedropper

Oops, I can’t upload my gamut

why do you say that about RYB?
if your in digital your probably stuck with a RGB or adobe RGB on your display.

Color settings are computed quite differently, with digital tools

Just pick one of the available wheels in any painting related software,
and see if you get complementary colors,

Following color theory we shall see violet as complementary of the base color yellow.
You won’t see it, in digital.

Rgb was just some spice and my critique as nobody of the artists using digital paintings noticed this and seems to care about color harmonies.

1 Like

I never noticed that swap in colors before. I will keep it in mind.

Are you asking for a color scheme designer like

Very true. Of course, there are more schemes than just RYB. For CMYK, which is also subtractive, yellow is not complementary to violet.

It seems that just having a palette with these colors wouldn’t be enough, because when you blend them on the canvas, wouldn’t you want the colors to interact the same way? Or is your proposal just to have the colors available in a palette to pick from?

I was researching on this topic and it seems RYB and CMY are somehow analogous to each other the sucky thing about RYB is that it seems to go only one way. I can’t find formulas to change it back to RGB for now. I get the feeling that CMY might be better.

Better to start simpler, to get a better understanding what features are really needed, otherwise I agree with you.

My idea was to have a palette with limited colors linked to the color picker. And button to refresh the palette on demand.

Coming from a traditional background I find all those color wheels confusing, and just some useful. I’m wondering to see no RYB wheel at all

This sounds sophisticated, I’m for a simpler approach, and like to mix colors on canvas, the manual way
Maybe I misunderstood, and you meant the better solution is to use gradients instead of a palette?

This color conversion seems to be tricky. I’m not sure but isn’t it possible to translate the color wheel into a matrix?
If I’d be good at math I’d try to come up with more helpful ideas…

I didn’t get any notification. Sorry for my late reply :expressionless:

So i’m asking because coming from a traditional background compared, then the mixing of colors is different. Screens mix the color of light, paints mix the color of pigments. When mixing pigments, Blue+Yellow = Green, when mixing light Blue+Yellow = Gray. What does your RYB wheel look like?

I made 2 RYB gradients (16 bit color)

#1 is using multiply to mix colors (so we get a green for Blue + Yellow) - which is more like what you’d see in traditional (though not the same). #2 is the normal mixing for screens (additive).

I’m guessing that the palette and color wheel that we see are just generated using gradient functions, which reflects the on canvas mixing.

1 Like

Awesome, I love both color schemes, although #1 is my preferred one.

An example of the ideal color wheel :

I suggest you to explore the whole website

I am still not in position to test RYB or CMY correctly since I am implementing another thing, but in RGB color space it should be more like this:

rgb interpolation
ard interpolation

I really don’t know.
I usually go under the assumption that I did not invent the color space and the person that created made the conversion formulas better because they invented it. The problem with RYB is that is comes from paint even if you input that color you need to talk to the computer in a language it understands in the end.

For what I have seen:

  • RGB>RYB (exists)
  • RYB>RGB (none)
  • RGB>CMY (exists)
  • CMY>RGB (exists)

If i make this I will implement CMY instead of RYB if no conversion back to RGB is found.

I know color theory is very cool and all with RYB but CMY is very similar and probably a better version of it too.

I know she is a very long to talk but she did the research historically too:

But I guess I could just use RGB and force feed a color to a certain angle too. Would not be much of a conversion but a real fake. No interpolation methods would be available just the wheel which sounds awfully sucky.

1 Like

Oof that video is indeed very long but touches a lot of important points…

I’m still not sure what Venn really wants to achieve.

RGB and CMY simply give the biggest possible gamut in additive and subtractive color mixing respectively with only 3 “primaries”, but they don’t model how humans perceive color difference (which is quite non-uniform and hence tricky to model).

It’s hard to implement “the” RYB color model, it’s an old empirical model that has been shoehorned to fit a theory more than once.

So question is, what should that “harmony palette” really give?

My feeling is that this is once more about overcoming the unnatural way colors (don’t) mix in simple RGB blending in comparison to real pigment based colors.

There are some RYB models out there like this paper but the results don’t look very convincing to me.

More realistic pigment mixing is something that can’t be done with just 3 primaries, it requires some spectral model like MyPaint features since 2.0, I think it uses 10 spectral samples (?) to simulate the light absorption.


Exactly. It’s not about additive or subtractive color mixing.
If the shoe doesn’t fit, then the math behind color computation is outdated or needs some updates.

The concept isn’t to fit colors into math but put math in service of colors. Of course, we need to get it working so the math is somewhat important :joy:

Empirical or not, it includes the colors of pigments as they were used in old times.
Let’s look at the genius of the past, (e. G. famous European painters) rather than look at abstract math behind Rgb color tubes.

This answers also the question you posted immediately after. Color harmony in itself defines good taste and this theory of the pigments had been studied and perfected thoroughly the ages.

It’s another color wheel, an artistic color wheel, if we wanna be pragmatic only, if we ignore the spirit of the past. Since we’re artists color harmony is not just a detail but fundamental knowledge.

Fake or not, if the angles are right, nobody cares about the math behind. No worries. The result counts :+1:t2::+1:t2::+1:t2:

Yeah indeed nice video. Nice speak and speaker :joy:

Vermillion Red, Gold Yellow and Ultramarine Blue are used because 1. these pigments were at the time the most colorful and lightfast pigments available, and 2. therefore often prescribed for religious artwork. If you look at the pigments used by the 17th century Dutch masters, they didn’t limit themselves to these three. Rembrandt’s known palette seems to have all the available reds at the time, two of which are more on the magenta side than the vermillion side.

Printers are using CMY because that’s the most efficient. The reason schools still teach RYB is because they are teaching Ittens’ color theory, which is a step backward in this regard. Computers are not pigment based, and there are ways of conceptualizing color besides Ittens, so maybe we should just treat computers as their own medium with it’s own ways of doing things. We after all are not angry with pastels for having different colors than acrylics.

You would remove a consolidated method that worked excellently. A chance less to express ourselves.

Computer or not, the relationship of color counts.

I could argument old Renaissance masters didn’t know Bauhaus and Itten,
The religion doesn’t matter either or all the variations of colors. This isn’t the topic, neither the color wheel itself.
But having the right color wheel is the key to get good colors, or at least a proven tool to operate with nice color combinations

Yeah, I would, because it just doesn’t work that way in a light-based medium. It’s not the computer’s fault that physics just doesn’t work the way Ittens thought it did. Most painting books even suggest having two sets of RYB to get a proper mixture of colors because everyone whose mixed vermillionish red and blue paint has seen that it leads to brownish goop instead of bright purple.

Even if we were to focus on solely human color vision instead of what physics is doing, we’d still have more luck explaining things like complementary color and color blindness using the opponent process and by extension, describing it with a oppent based color model (LAB, YUV and YCrCb are such models).

There are so many different ways to compute color wheels. I feel overwhelmed by this math games, if I can say mine.
But the one I need isn’t there.
As said we don’t talk about colors wheels. It’s the combination of colors.

Take a look at the code of gmic ryb2rgb and rgb2ryb if interested in converting back to rgb.