Color wheels are wrong? How color vision actually works • Hacker News Board

Re: Color wheels are wrong? How color vision actually works

#31

Post by cynicalkane » Tue, Feb 01, 2011, 9:57 PM UTC

joeld42 wrote: post #9

This article tries to sensationalize and obfuscate something that is pretty simple in reality. RGB are the primaries in light. By mixing these three colors you can create any color the human vision system can perceive (yes, because of tristimulus). When white light hits a material, some of those RGB wavelengths are absorbed (subtracted). RGB - GB(Cyan) = Red, RGB - RG(Yellow) = Blue, RGB - RB(Magenta) = Green. Thus,…

This isn't correct. The color space is horseshoe-shaped and cannot be represented by a triangle of three additive primary colors. In addition, standard color gamuts tend to be quite a bit smaller than theoretically possible. Check out: https://secure.wikimedia.org/wikipedia/en/wiki/Gamut . This is why hi-def TV is defining extensions to the usual colorspace.

Furthermore, the laws of physics don't particularly care that humans perceive three color dimensions, and surfaces and lights are allowed to absorb or emit whatever wavelengths they please. When you have a light source with a funky spectrum, this can cause colors to look very strange, as a surface might not reflect a funny, spiky spectrum the same way (w.r.t. the human 3-dimensional color space) it will reflect a smooth color spectrum of the same color. You've probably seen cars seemingly change color when the light source is primarily those orange-ish streetlights. My parents' red minivan, for instance, turns a red-tinted grey.

Re: Color wheels are wrong? How color vision actually works

#32

Post by slavak » Tue, Feb 01, 2011, 10:01 PM UTC

joeld42 wrote: post #9

This article tries to sensationalize and obfuscate something that is pretty simple in reality. RGB are the primaries in light. By mixing these three colors you can create any color the human vision system can perceive (yes, because of tristimulus). When white light hits a material, some of those RGB wavelengths are absorbed (subtracted). RGB - GB(Cyan) = Red, RGB - RG(Yellow) = Blue, RGB - RB(Magenta) = Green. Thus,…

Sorry to disappoint you, but color theory is in fact anything BUT simple. For starters, RGB are not "THE primaries in light." In fact, primary colors aren't a property of light but of the way you perceive light. You could just as successfully choose a different set of 3 primary colors and still be able to represent [an approximation of] any other color. That is, of course, assuming you're trying to represent color to a human with 3 types of cone cells - otherwise things only get more complicated...

Re: Color wheels are wrong? How color vision actually works

#33

Post by jessriedel » Tue, Feb 01, 2011, 10:11 PM UTC

lutorm wrote: post #18

jessriedel wrote: post #13

EDIT: joeld42 answered my question elsewhere on this page: http://news.ycombinator.com/item?id=2166782 The artists reds and blues are just approximations to the magenta and cyan of printers. ---- So I understand that in order to mimic the electromagnetic stimulation to the cones you only need to control the amount of RGB light hitting the retina. And I understand that propagating light is additive, so that for light…

Maybe you mistyped, but cyan paint absorbs red light, it reflects everything but red, i.e. cyan.

yes I did

Re: Color wheels are wrong? How color vision actually works

#34

Post by jacobolus » Tue, Feb 01, 2011, 10:16 PM UTC

m-photonic wrote: post #29

>And magenta? It comes from full R and B with no G, activating Filter #1 full-positive, Filter #2 at zero. This doesn't seem right to me. If the second filter were at zero, you should have a pure red and not something with blue content in it like magenta clearly has. I think he may be representing the second filter as R+G-B, when R+G-2B would make more sense. The latter system shows FFFFFF as being neutral on the yel…

None of these are straight-forward “linear” sums: there are differing amounts of each type of cone cell in the retina (and the proportions vary from one part of the retina to another), there are several levels of combination of signals which we don’t fully understand currently, the eye/brain adapts to what it’s just been looking at, what else is in the visual field, what it knows the light source to be, what “memory colors” it expects for an object, and so forth.

Thinking of the mechanisms of the eye operating directly on “FFFFFF” is a very imprecise model for what’s happening.

Re: Color wheels are wrong? How color vision actually works

#35

Post by jacobolus » Tue, Feb 01, 2011, 10:18 PM UTC

martian wrote: post #21

In the early 20th century Albert Munsell did a series of empirical studies to determine exactly how people see colors relative to one another. The results are expected, but fascinating. http://en.wikipedia.org/wiki/Munsell_color_system

Why are the results expected?

(Also, I really need to expand the Munsell Wikipedia article, but Albert Munsell was dead before much of the empirical work that went into the 1929 Munsell Book of Color – by that point his son was running the company – or into the 1943 Munsell Renotations, which were based on the work of the Optical Society of America.)

Re: Color wheels are wrong? How color vision actually works

#36

Post by jackolas » Tue, Feb 01, 2011, 10:30 PM UTC

armandososa wrote: post #27

Artists get it wrong I kinda get upset about all his use of "artist" as a derogatory term. Have he ever painted anything in real life? Beleive it or not, just a canvas plus red, blue and yellow acrylic paint will be enough. Oh, and some kind of talent.

I'm not sure that understanding color as we experience it is required to be a good artist, but it might help the more engineer-ish of us.

Re: Color wheels are wrong? How color vision actually works

#37

Post by cwp » Tue, Feb 01, 2011, 10:34 PM UTC

steadicat wrote: post #26

Why does he conclude that we need 4 primary colors? He knows that our eyes have three kinds of 'sensors', which roughly correspond to [R, G, B]. How the brain processes the initial perceptions, allegedly [R-G, (R+G)-B, R+G+B], doesn't change the fact that you can approximate all colors by mixing quantities of R, G and B. What I'd find more interesting is a proposal (or a mention) of a color space that's based on what…

Actually, this kind of encoding is quite common. Analog television, digital video and JPEG images are all encoded using brightness, red-green and blue-yellow channels. The details vary (eg. between YUV, YCbCr and YPbPr) but they all take advantage of the fact that our eyes are more sensitive to variations in brightness than to variations in colour. As a result, we can subject the colour data to higher levels of compression without noticing any visual degradation of the image.

Re: Color wheels are wrong? How color vision actually works

#38

Post by podperson » Tue, Feb 01, 2011, 10:46 PM UTC

joeld42 wrote: post #9

This article tries to sensationalize and obfuscate something that is pretty simple in reality. RGB are the primaries in light. By mixing these three colors you can create any color the human vision system can perceive (yes, because of tristimulus). When white light hits a material, some of those RGB wavelengths are absorbed (subtracted). RGB - GB(Cyan) = Red, RGB - RG(Yellow) = Blue, RGB - RB(Magenta) = Green. Thus,…

I think you're right in some respects (e.g. most artists have a much more sophisticated understanding of color than red, yellow, blue -- that's the version you get taught in elementary school).

But you're missing the point in others: half the article is devoted to explaining the inadequacies of the RGB model for handling actual real world color. (This is why photos of sunsets -- digital or film -- never look right.)

A simple example -- color looks weird under "white" LEDs (at least the current ones) because they're actually RGB LEDs balanced to create the illusion of "white" light. Some orange things will look all but black under a white LED while others will look orange. Why? Because orange light can be actually orange, or a mixture of wavelengths that gets a similar response from your eye.

And so on and on. The article is a bit annoying (and it would help if it assumed most readers will know about CMYK and RGB color models already) but the fundamental lesson -- that color is more complex than you think and you need to understand the underlying physics and physiology to really understand color is worth making.

Re: Color wheels are wrong? How color vision actually works

#39

Post by jacobolus » Tue, Feb 01, 2011, 10:56 PM UTC

podperson wrote: post #38

joeld42 wrote: post #9

This article tries to sensationalize and obfuscate something that is pretty simple in reality. RGB are the primaries in light. By mixing these three colors you can create any color the human vision system can perceive (yes, because of tristimulus). When white light hits a material, some of those RGB wavelengths are absorbed (subtracted). RGB - GB(Cyan) = Red, RGB - RG(Yellow) = Blue, RGB - RB(Magenta) = Green. Thus,…

I think you're right in some respects (e.g. most artists have a much more sophisticated understanding of color than red, yellow, blue -- that's the version you get taught in elementary school). But you're missing the point in others: half the article is devoted to explaining the inadequacies of the RGB model for handling actual real world color. (This is why photos of sunsets -- digital or film -- never look right.)…

> This is why photos of sunsets -- digital or film -- never look right

The main reason sunsets don’t look right is that sunsets have a huge dynamic range, beyond the ability of our printed photographs or computer displays to reproduce (no one has the sun in their living room).

Re: Color wheels are wrong? How color vision actually works

#40

Post by sedachv » Tue, Feb 01, 2011, 11:06 PM UTC

diiq wrote: post #2

I'm happy to see someone investigating color, and the way we perceive it. So many color theory books are more mystical than scientific; it's wonderful to see a physiological model. I suspect that we continue to teach the RGB, RBY, and CYMK color wheels because they behave correctly with respect to the behavior of particular physical media. Yellow-looking paint and blue-looking paint do make green-looking paint. Red-l…

It's not really a new thing. Rudolf Arnheim's Art and Visual Perception (published in 1974, widely read in art school from what I understand) had a whole chapter about exactly the same material as in that blog post. But before that it had a chapter on the psychology of color, and why the color wheel is in fact useful.