Enough people have asked me to adjudicate this question that I really have to do an emergency post.
Unless you are living in a dark cave you have probably heard the furor. What color is the dress????
It seems that 75% see it as white and gold and 25% see it as blue and gold.
http://mashable.com/2015/02/
https://www.yahoo.com/health/
http://www.people.com/article/
http://petapixel.com/2015/02/
http://gawker.com/what-color-
http://www.cnn.com/2015/02/26/
https://simplifywork.blogs.xerox.com/2015/02/27/gold-or-blue-whats-the-hue/#.VPDgW-E-WDl
https://www.yahoo.com/health/
http://www.people.com/article/
http://petapixel.com/2015/02/
http://gawker.com/what-color-
http://www.cnn.com/2015/02/26/
https://simplifywork.blogs.xerox.com/2015/02/27/gold-or-blue-whats-the-hue/#.VPDgW-E-WDl
I have learned to avoid this type of question. I have learned in my marriage that if I say “Wow! Look at that really gorgeous woman in the turquoise dress! She’s really cute, and I would like to ask her out – maybe take her to Bermuda for the weekend.” this will start an argument with my wife about the color of the dress!
But clearly I need to weigh in on this. It comes down to how we define color.
What is "color"?
Definition #1: Color is something computed from the spectrum of light that comes off an object.
If this were true, a piece of copy paper would be brilliant white in the sunlight and a very very dark brown under incandescent light. This phenomenon is explained by the fact that the cones are autoranging.
Definition #2: Color is defined by the signals collected by the cones after this auto-ranging.
This optical illusion proves that to be wrong: http://en.wikipedia.org/wiki/Checker_shadow_illusion. The cones are collecting the same signals for A and B in that illusion, but the neurons in the eye are comparing the pixels with their neighbors. “This pixel is bluer than the one next to it, so I am going to call it “a little bit blue”.
Definition #1: Color is something computed from the spectrum of light that comes off an object.
If this were true, a piece of copy paper would be brilliant white in the sunlight and a very very dark brown under incandescent light. This phenomenon is explained by the fact that the cones are autoranging.
Definition #2: Color is defined by the signals collected by the cones after this auto-ranging.
This optical illusion proves that to be wrong: http://en.wikipedia.org/wiki/Checker_shadow_illusion. The cones are collecting the same signals for A and B in that illusion, but the neurons in the eye are comparing the pixels with their neighbors. “This pixel is bluer than the one next to it, so I am going to call it “a little bit blue”.
Definition #3: Color is defined by the signals leaving the eye.
This doesn’t quite explain the checker illusion. The lower part of the brain pulls a lot of tricks in an attempt to interpret a visual scene. In the checker example, the lower brain interprets each of the squares of the checkerboard as an object, and simplifies things by saying that the variation in signal intensity is due to shading, and not due to any properties of the dress.
Definition #4: Color is defined by the interpretation provided by the lower brain, after segmenting the scene into distinct objects.
Still not there yet. The thing that is (likely) the point of confusion in the dress image is that the brain actively seeks a white point from which to judge color. Look at a newspaper. What color is it? White. Now lay a piece of white ultra bright copy paper next to it. What color is the newspaper now? It turned dingy. Your brain first used the newspaper as a white point, and made its color assessment based on that. When the copy paper was introduced, your brain picked up a different definition of white to compare things to.
In the dress picture, you will note that the upper right portion is saturated. This is confusing to the brain, since the autorange in the eyeball doesn’t normally allow things to saturate. Our eye would scale this so that we could see the bright area, and we would not be able to tell the color of the dress.
How does the brain interpret the saturation that happened in the camera? Does it see that as the white point and assess from there? Or does it come up with another brilliant (pun intended) explanation, and set it’s white point to something beyond 255, 255, 255? This is a guess, but I think that different brains might set different white points.
Objects don’t have an inherent color. Color is a subtle interplay between the light hitting an object, the light reflected from the object, the spectral response and autoranging of the cones, the low level segmentation into distinct objects, and the interpretation of white point.
So what’s my answer? The question is a silly question. Dresses do not have any inherent color. The better question is “what color do you see when you look at the dress?” That question apparently depends on the viewer. Color is in theeye brain of the beholder.
This doesn’t quite explain the checker illusion. The lower part of the brain pulls a lot of tricks in an attempt to interpret a visual scene. In the checker example, the lower brain interprets each of the squares of the checkerboard as an object, and simplifies things by saying that the variation in signal intensity is due to shading, and not due to any properties of the dress.
Definition #4: Color is defined by the interpretation provided by the lower brain, after segmenting the scene into distinct objects.
Still not there yet. The thing that is (likely) the point of confusion in the dress image is that the brain actively seeks a white point from which to judge color. Look at a newspaper. What color is it? White. Now lay a piece of white ultra bright copy paper next to it. What color is the newspaper now? It turned dingy. Your brain first used the newspaper as a white point, and made its color assessment based on that. When the copy paper was introduced, your brain picked up a different definition of white to compare things to.
In the dress picture, you will note that the upper right portion is saturated. This is confusing to the brain, since the autorange in the eyeball doesn’t normally allow things to saturate. Our eye would scale this so that we could see the bright area, and we would not be able to tell the color of the dress.
How does the brain interpret the saturation that happened in the camera? Does it see that as the white point and assess from there? Or does it come up with another brilliant (pun intended) explanation, and set it’s white point to something beyond 255, 255, 255? This is a guess, but I think that different brains might set different white points.
Objects don’t have an inherent color. Color is a subtle interplay between the light hitting an object, the light reflected from the object, the spectral response and autoranging of the cones, the low level segmentation into distinct objects, and the interpretation of white point.
So what’s my answer? The question is a silly question. Dresses do not have any inherent color. The better question is “what color do you see when you look at the dress?” That question apparently depends on the viewer. Color is in the
Dodging the proverbial color bullet, dude. LOL
ReplyDeleteAh John, your are thinking as a color scientist and not a textile/fashion guy. My wife has me trained to be both, so what I see is a white dress with gold banding and fine pale blue pinstripes. Or at least that's what i see with a fairly inferior image. And as for Bermuda, my wife would probably say 'fine, as long as I get to go and keep all the credit cards.'
ReplyDeleteThis is the best post on this dress thing I've read! Nice job. :)
ReplyDelete"I couldn't stop reading, very captivating!"
ReplyDeleteThank you for the valuable insights!"
ReplyDelete. "This article deserves a wider audience!"
ReplyDelete