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

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

#41

Post by jacobolus » Tue, Feb 01, 2011, 11:29 PM UTC

knutae wrote: post #25

I find the CIE Standard Observer graph enlightening: http://en.wikipedia.org/wiki/CIE_1931_color_space#Color_matc... As far as I understand this, the red function, which corresponds to the red receptors in a typical human eye, reacts mostly to high wavelengths, but also has a small spike in the lower end of the spectrum. This explains why the violet end of the spectrum looks similar to red to human eyes, and this is…

No, the CIE standard observer is not the same as the responsivities of cone cells in the eye. That article you linked describes how it was arrived at:

1. Match monochromatic (narrow wavelength band) light sources of each wavelength with three particular monochromatic lights (called “R”, “G”, “B”), in some cases needing to add some of one of those three to the monochromatic source in order to get the two sides to match – this is a “negative” component. This process results in three (r(λ), g(λ), b(λ)) functions of wavelength.

2. Take three linear combinations of those three functions and call them (x(λ), y(λ), z(λ)), such that (a) the y(λ) function is approximately the same as the photopic luminance function as best it could be computed at the time, (b) all three of x, y, and z had only positive values, and (c) integrating each of the three yields the same value, so that an equal-energy stimulus will have the same X, Y, and Z values.

The red receptors in the human eye do not really have this kind of spike in response in the lower end of the spectrum.

* * *

However, “red” the “psychological primary” or “unique hue” does have some violet/blue in it. Any monochromatic red light source is a bit on the orange side.

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

#42

Post by apl » Tue, Feb 01, 2011, 11:30 PM UTC

  > But I digress, and besides I did promise to be all gross
  > and irresponsible, so I’ll stick with that.
  > So there are R, G, and B cones.

Cute and all, but so very wrong. There are acceptable simplifications; this is not one of them.

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

#43

Post by diiq » Tue, Feb 01, 2011, 11:30 PM UTC

sedachv wrote: post #40

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.

Right! I used "proposed" to mean "argued for" rather than "set forth for the first time".

Arnheim is kind enough to distinguish between "generative complements" and what he calls "fundamental complements", (though I would prefer they be called "perceptual"). He warns specifically against confusing perceptual color and color-in-the-world. By making that distinction, Arnheim validates BOTH systems, rather than attempting to invalidate three-primary systems (as this article appears to do).

The article calls RGB and RBY color "wrong" and unable to "stand up to even minor scrutiny".

Arnheim calls them "generative".

I claim their generative nature (that they inform how to create color) is why they are taught in schools. I'm not denying the four-color idea; I'm just saying it is based on a different definition of color, and different intended use of color.

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

#44

Post by icegreentea » Tue, Feb 01, 2011, 11:55 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.

Agreed. The fact that artists make stuff that 'works' at all should be hint enough that they aren't completely wrong. Just working with a useful approximation/abstraction layer.

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

#45

Post by barrkel » Wed, Feb 02, 2011, 2:16 AM UTC

sfphotoarts wrote: post #12

michael_dorfman wrote: post #5

Earlier quoted context omitted.

You talk about what the article "looks like", but have you actually read it? I found it to be a well-written, well-researched introduction to some of the problems in color theory. If that's what you mean by "a mashup", then I guess all secondary literature is. As for the site, asmartbear is no stranger to these parts, and Jason's posts are well-regarded.

possibly my British grammar is stumping you, but yes, to read something you have to look at it. This article might appear well researched here, but this is the hello-world equivalent on an art forum. Maybe if Jason believes he has something meaningful to contribute to color theory, he should consider updating the Wikipedia page rather than promoting his own self interest by advertising his book & blog on HN.

> to read something you have to look at it.

It seems basic logic is stumping you: when A implies B, and B is true, it does not follow that A is true. You assert that you've looked at the article, and certainly one needs to look at the article to read it, but that doesn't imply that you've read it.

> this is the hello-world equivalent on an art forum

I fear you might be at the wrong place - Hacker News isn't an art forum.

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

#46

Post by mambodog » Wed, Feb 02, 2011, 2:17 AM UTC

A while ago I submitted an article[1] about the YUV colour system, which takes into account human perception of brightness/luminance, and how it differs between colours. It's also an interesting colour system to work with, as the 'Y' channel just stores luminance, so discarding the other two channels gives you a greyscale image.

I recommend reading it if you haven't already.

[1] http://nreynolds.co.uk/blog/hsv-is-dead/

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

#47

Post by podperson » Wed, Feb 02, 2011, 2:46 AM UTC

jacobolus wrote: post #39

podperson wrote: post #38

Earlier quoted context omitted.

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).

No, the main reason is that RGB is an approximation of color not true color. Light in the real world isn't RGB. Human beings do not see in RGB. Read the damn article. (There's also a lengthy post above which tries to make the points over again.)

Do you think CMYK will reproduce sunsets if you shine a bright enough light on a piece of paper and use black enough ink? Same argument.

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

#48

Post by blahblahblah » Wed, Feb 02, 2011, 5:20 AM UTC

So there’s no such thing as "red with a little green" -- there’s just a less intense red. The brain physically cannot see "greenish-red" because the filter removes that information.

Under ordinary circumstances this is true. However, it has been shown that the human brain can be made to see reddish-green, yellowish-blue, etc. Scientific American published an article about this strange hack of human visual perception just last year. http://www.scientificamerican.com/article.cfm?id=seeing-forb...

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

#49

Post by jacobolus » Wed, Feb 02, 2011, 5:46 AM UTC

podperson wrote: post #47

jacobolus wrote: post #39

Earlier quoted context omitted.

> 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).

No, the main reason is that RGB is an approximation of color not true color. Light in the real world isn't RGB. Human beings do not see in RGB. Read the damn article. (There's also a lengthy post above which tries to make the points over again.) Do you think CMYK will reproduce sunsets if you shine a bright enough light on a piece of paper and use black enough ink? Same argument.

(yeah, I wrote that lengthy post)

I understand the concept of metamerism. You are quite right about objects changing appearance from one light source to the next, etc. It’s also true that sunsets often have colors which are more colorful than can be produced by computer displays or 4-color-process prints. However, I stand by my assertion that the main reason that sunsets don’t look right on screen or in print has to do with the lack of dynamic range.

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

#50

Post by jacobolus » Wed, Feb 02, 2011, 5:53 AM UTC

sfphotoarts wrote: post #12

michael_dorfman wrote: post #5

Earlier quoted context omitted.

You talk about what the article "looks like", but have you actually read it? I found it to be a well-written, well-researched introduction to some of the problems in color theory. If that's what you mean by "a mashup", then I guess all secondary literature is. As for the site, asmartbear is no stranger to these parts, and Jason's posts are well-regarded.

possibly my British grammar is stumping you, but yes, to read something you have to look at it. This article might appear well researched here, but this is the hello-world equivalent on an art forum. Maybe if Jason believes he has something meaningful to contribute to color theory, he should consider updating the Wikipedia page rather than promoting his own self interest by advertising his book & blog on HN.

> he should consider updating the Wikipedia page

Please note that anyone enthusiastic to learn and contribute is extremely welcome on the color-related Wikipedia articles; many of them are in a pretty sorry state.