Sunday, August 10, 2008

Color Wheel Question

Blog reader Emily emailed me a question that I couldn’t answer, so she gave me permission to post it to the group mind.

"I was opening up a new file in Painter today and started thinking about that little color wheel it shows you - the standard computer color wheel with all the colors in the wheel and the brightness and darkness controlled by a slider on the side. I noticed that the blue and red seem to have the illusion of spokes of more intense color poking into the wheel and the teal, purple, and yellow appear to have light spokes poking out of the wheel. Only the green just looks nearly flat all the way across. If you desaturate the whole wheel, all the tones are perfectly even. Any idea why this is?"

24 comments:

SCIBOTIC said...
This comment has been removed by the author.
SCIBOTIC said...
This comment has been removed by the author.
SCIBOTIC said...

Don't trust the desaturation, often the computer will adjust the value to even things out across all the different hues.
Here's a quick demonstration based on GIMP's two options for desaturation, based on Luminosity and Lightness.
http://tinyurl.com/colordesat

Fact is, different hues have different values.

SCIBOTIC said...

What's interesting when you pick GIMP's third option, "Average" is you get a perfect demonstration of the spokes you mentioned.
http://tinyurl.com/colorspokes

Keep in mind that your computer screen operates on additive colour, where the primaries (Red, Green and Blue) go brighter when mixed.
As opposed to subtractive colour in paints, where the primaries (Red, Yellow and Blue) go darker when mixed.

So in this case, the secondary colours (Light Blue, Purple and Yellow) have the brightest values.

The brightest value is the white in the center though and no chromatic colour can compete with that, so the darker primary colours make a more dramatic transition when it moves towards white giving you the inwards spoke, but since all the values average out near the middle as they desaturate and head for white, you get the spokes pointing outwards from that area heading towards what they appear closest to, the brighter secondary colours.

I think... This stuff hurts my fragile little mind.

SCIBOTIC said...

One last thing, want to pimp one of my teachers colour theory site.

http://www.huevaluechroma.com/

Casey Klahn said...

Not the complete answer, because I'm not sure what you see, but here goes.

The visible spectrum peaks on certain colors (hues). They appear more intense at these peaks, and all colors do not peak at the same intensity. The actual wavelengths are different amounts or lengths.

It is a good illustration of the differences between value and intensity.

I see the "spokes" popping out as I review your illustration. That is certainly the intensity attributable to wavelength, IMO. As you go left or right from the spoke, the amount of "other" hue is increased, which de-saturates the hue.

Anyway, I could go on, but let's see if my info makes sense to you first. Great question!

Erik Bongers said...

Yes, I think scibotic is on the right track.

A computerscreen (or TV) has little dots of Red, Green and Blue.
You can actually see them with a magnifying glass when you look at a White area of the screen.

So White = Red + Green + Blue !!!
This is called the Additive Color System, which applies when you create colors by shining colored light to mix colors.

Looking closely at your colorwheel on your screen, you'll see only Red dots in the Red area, Green in the Green area, etc.

In the Yellow area you'll see Red + Green dots.
So your screen 'shines more light' in a yellow area.
The purple (Magenta) area = Red + Blue.
The light blue (Cyan) area = Blue + Green.

Those 'spokes' you see mean that your monitor doesn't compensate strenght of the light properly when combining colors.

A so called 'calibrated' monitor will not show those spokes.

You should be able to calibrate your monitor either via the monitor settings or via the settings of the computer or via settings of your graphical software (in order of pref).

(p.s. I also haven't calibrated my monitor and thus see spokes. And I digitally painted many pages of a children's book without big discrepancies when I got the printed book in my hands.)

Katherine Tyrrell said...

Maybe worth mentioning that Casey uses intensity when other people might say "chroma" - the strength of a colour

Value is a different dimension - that's the lightness or darkness or tone of a colour.

Erik Bongers said...

A thing I have forgotten to mention.
Yellow will always appear brighter to us than all the other colors.
Blame our mind for this: yellow is the color of natural light like the sun or fire, so to our mind yellow = light.
Maybe this is also true for the two other 'spokes', I'm not sure.

Paolo Rivera said...

My answer was going to be along the lines of the above, but I also thought it might be an optical illusion of some sort.
I took the posted image into Photoshop and inverted it. The resulting wheel was simply rotated 180°, albeit with black at the center. Although the additive and subtractive primaries switch places, I still get the same "spokes" in the subtractive primaries, which leads me to believe that the they are a psychological effect, as opposed to physical stimuli.
What do change, however, are the "spokes" in the additive primaries, which demonstrates what "scibiotic" stated about the contrast between white and the darker additive primaries.

I'll admit that Photoshop could conceivably be skewing my results, but based on what it showed me, the values should produce a smooth gradient that only appears to be uneven, i.e. produces "spokes."

As to why this happens, I'm afraid I don't have the answer. Maybe one of us should Ask a Color Scientist.

Casey Klahn said...

But, notice there are no spokes in the value wheel.

Value v. Chroma.

Here is a reference about Color Vision and the metric differences in your eye's perception. I think the monitor being equaled, you may agree that colors have "spokes" of intensity when understood as cones receiving the different hues differently.

link

The specific illustration I want you to see is this one.

I may not be correct in all of this, but I just wanted to illustrate what I mean by wavelengths of light producing the intense "spokes".

Cheers!

Justin said...

I didn't read all the replies, just breezed over most of them, but from what I can tell, Photoshop (and painter and others) give every possible "color" a hue value, a value value, and a saturation value. The fact when you desaturate it is actually WRONG. To calculate the desaturation, it uses the hue's stored value, not their REAL value. There are ways around this, but they are awfully inconvenient.

Justin said...

typo

*The fact is when you desaturate it, the result is actually "wrong"

Speck said...

On Emily's color wheel the primary colors aren't equidistant. Blue and yellow are directly opposite. If yellow were moved counterclockwise to the 11:00 position the green would be more intense and create a poke-out spoke illusion. The primary colors would all poke in. I'm sure there's some mathematical equation to explain that but it's way beyond me.

I think the in/out spoke illusion is created from trying to represent a three-dimensional concept in two dimensions. If a color *sphere* were created with the primary colors equidistant, the directional spoke effect would disappear after viewing from several different perspectives.

Oy, I feel like I'm in a quantum physics class. My brain hurts.

Tom Scholes said...

I can only assume that the spokes are noticeable due to the contrasts of the inherent value and perhaps intensities each color naturally exhibits.

When you want to see the values of colors digitally you can not simply desaturate. You can either convert an images color space to LAB color and then desaturate or you can simply convert to grayscale. I don't believe it's perfect, but it's much closer to what we see.

Tometheus said...
This comment has been removed by the author.
Tometheus said...
This comment has been removed by the author.
Emily said...

Wow, so many interesting answers!

Michael Pieczonka said...

have a strong feeling this one has to do with a limitation of the computer monitor, and the fact that it uses additive colour based on RGB pixels. Interesting that the green part of the wheel does not have not have a chroma spike like the other R & B areas, but also just as interesting seeing as Green is one of the 3 pixels categories.. but also the only one that is not a true primary.

Charley Parker said...

As several people have pointed out, different hues have different intrinsic values at their highest degree of saturation, which is why Munsell's three dimensional color model is distinctly assymetrical.

In desaturating Painter's color wheel, you've removed the intrinsic value differences along with the colors, leaving only a grayscale value scale unrelated to hue.

If instead you take a screen capture of the color wheel and desaturate that by converting the image to grayscale, you'll see the radial lines of brightest and darkest color values remain visible.

Sample here.

James Gurney said...

Thanks again to all of you for clarifying this for Emily, me, and hopefully lots of other people. The knowledge bank out there is very impressive. Charley, your screen capture comparison really drives the point home!

Seneca the Younger said...

I suspect you're also seeing some Mach banding

Ally said...

Hmm I do not have a calibrated monitor. Whenever I squint at the image, the spokes are not as intense, so I believe it's a value vs. intensity thing and not a monitor thing. Even when I took a screen capture and desaturated it, I still did not see the spokes in the gray scale image.

For an art example, you can take a still life of Chardin's, The Silver Goblet, and desaturate it in photoshop. The bright red of those apples seems to be very dark in the colored image, but when desaturated, it's actually a much lighter value than it seems. Just a step or two darker in value than the golden yellow! What seems to be a much lighter green is actually darker than the red.

briggsy said...

1. The yellow, cyan and magenta sectors are each lighter than the adjoining R,G and B sectors for the reason Erik gave: in each of them lights of both of the adjacent colours are glowing. This however has nothing to do with monitor calibration: no amount of calibration could or should change this relationship.

2. The apparent linear spokes in the yellow, cyan and magenta sectors are a kind of optical illusion, as Seneca suggested. Colours increase steadily in lightness on either side towards these spokes, so that the spokes are lighter than the immediately adjoining colours on either side, but not by a sudden step. You can confirm this by greatly magnifying the image so that the individual pixels show.

2. The peaks themselves are created by point 1 above rather than by different visual responses to light of different hues, although the latter does explain why the yellow peak is lighter than the cyan and magenta peaks, and why the green valley is lighter than the red and blue valleys.

3. To amplify SCIBOTIC's first comment, the desaturate command transforms all colours to a grey of the same "L" value in HLS space. L in HLS is not true lightness, but is a rather artificial parameter defined as 0 for black, 1 for white, and 0.5 for all pure colours, no matter whether they look light (e.g. yellow) or dark (e.g.blue):
http://www.huevaluechroma.com/094.php

4. Converting to greyscale (or desaturating in Lab mode) transforms each colour to a grey of the same "L" value in Lab space, which as Tom Scholes states is not perfect but much closer to how we see. So Ally should use the greyscale conversion, not the desaturate command, to reveal true lightness relationships.