## How does "The Dress" illusion work on a computer screen?

For the discussion of the sciences. Physics problems, chemistry equations, biology weirdness, it all goes here.

Moderators: gmalivuk, Moderators General, Prelates

King Author
Posts: 736
Joined: Sun Apr 12, 2009 12:30 pm UTC
Location: Pennsylvania, USA

### How does "The Dress" illusion work on a computer screen?

I understand the science behind the optical illusion of The Dress, at least as much as anyone without an optical science background can, but I've never been clear on how the illusion was supposed to have worked on a computer screen. Colors in images are stored as discrete values (lossy format wiggle room notwithstanding). They should appear the same to everyone, regardless.

I just !gimage'd The Dress. Some of the images look black and blue, but most look white and gold. I copied one of each such image into Paint, and used the eyedropper to take a peek at their RGB values.

A random sample of a gold area from a gold-and-white image...
R: 128
G: 111
B: 75

Unarguably a brown/gold-ish sort of color.

A random sample from the "white" area...
R: 130
G: 140
B: 175

Again, unarguably, well, grayish. But clearly not black or blue.

A random sample of blue from the blue-and-black dress...
R: 20
G: 48
B: 118

And black...
R: 30
B: 33
G: 47

Irrefutably "blue" and "black."

Sampling color data from over a dozen images, they're almost all clearly, unarguably gold and white. RGB values don't lie.

I can understand how there might be an optical illusion if you saw it in real life, since color data in real life isn't stored as discrete values (as far as we know >_>), but I never understood how some people supposedly saw the dress as blue and black when looking at an image on their computer or phone, when the RGB values from the vast majority of images of the dress are unarguably white and gold.

Do those people see all whites and golds on computer monitor as blues and blacks? How could they possibly see just this one particular image as blue and black when the RGB value is undeniable? Or were they just lying all along, since the original Tumblr post about The Dress specifically said "my friend and I are arguing whether this dress is gold and white or blue and black," and the people saying it was blue and black were just being silly and saying the wrong thing on purpose? (I really do wonder what people would have said about The Dress if the original post simply said "what color is this dress?" and didn't seed peoples reactions with suggestions.)
I have signitures disabled. If you do, too...you can't read this, so nevermind >_>

Flumble
Yes Man
Posts: 2203
Joined: Sun Aug 05, 2012 9:35 pm UTC

### Re: How does "The Dress" illusion work on a computer screen?

As discussed in here, the discrepance doesn't happen because of the RGB values, but because of the perceived context. If you assume the lighting is blueish (like the left part of the linked comic), the inherent colour of the dress will be assumed to be gold and white. If you assume the lighting is whiteish (like the right part of the linked comic), the inherent colour of the dress will be assumed to be black and blue.

And, yes, people will assume the question is about the inherent colour of the dress, not the RGB values of the picture.

Nyktos
Posts: 138
Joined: Mon Mar 02, 2009 4:02 pm UTC

### Re: How does "The Dress" illusion work on a computer screen?

Right. I could take a picture of a white dress under, say, red light. Undoubtedly if you examined the RGB values of the pixels making up the image of the dress, they would be shades of red or pink. But people would likely have no trouble telling that the dress is actually white, because the human brain is capable of correcting for the lightning. In the famous photo, you can't see much background, which makes it hard to tell what the lightning is actually like, and apparently different people's brains made different guesses.

Sizik
Posts: 1242
Joined: Wed Aug 27, 2008 3:48 am UTC

### Re: How does "The Dress" illusion work on a computer screen?

King Author wrote:Colors in images are stored as discrete values (lossy format wiggle room notwithstanding). They should appear the same to everyone, regardless.

Oh really?
gmalivuk wrote:
King Author wrote:If space (rather, distance) is an illusion, it'd be possible for one meta-me to experience both body's sensory inputs.
Yes. And if wishes were horses, wishing wells would fill up very quickly with drowned horses.

speising
Posts: 2339
Joined: Mon Sep 03, 2012 4:54 pm UTC
Location: wien

### Re: How does "The Dress" illusion work on a computer screen?

That's the definition of an (optical) illusion - that something appears different than it actually is.
I don't get the distinction between real life and screen. I don't see discreet RGB values when i look at my screen, either. I see colours.
Last edited by speising on Tue Aug 25, 2015 10:34 am UTC, edited 1 time in total.

Xanthir
My HERO!!!
Posts: 5400
Joined: Tue Feb 20, 2007 12:49 am UTC
Contact:

### Re: How does "The Dress" illusion work on a computer screen?

King Author wrote:I understand the science behind the optical illusion of The Dress, at least as much as anyone without an optical science background can, but I've never been clear on how the illusion was supposed to have worked on a computer screen. Colors in images are stored as discrete values (lossy format wiggle room notwithstanding). They should appear the same to everyone, regardless.

Eyes are not computers. They do not see things the same from person to person. In addition to bio differences, our brains do *significant* context-sensitive alteration to the colors we see to produce the colors we "perceive".

(If you have two books side-by-side, one in shade and one in sun, the black ink on the in-sun book is brighter than the white page of the in-shade book. We still consider it darker because our brains aren't stupid.)

(Also monitors aren't calibrated perfectly, and will display colors slightly differently.)

I just !gimage'd The Dress. Some of the images look black and blue, but most look white and gold. I copied one of each such image into Paint, and used the eyedropper to take a peek at their RGB values.

A random sample of a gold area from a gold-and-white image...
R: 128
G: 111
B: 75

Unarguably a brown/gold-ish sort of color.

A random sample from the "white" area...
R: 130
G: 140
B: 175

Again, unarguably, well, grayish. But clearly not black or blue.

A random sample of blue from the blue-and-black dress...
R: 20
G: 48
B: 118

And black...
R: 30
B: 33
G: 47

Irrefutably "blue" and "black."

Sampling color data from over a dozen images, they're almost all clearly, unarguably gold and white. RGB values don't lie.

Note that the "white" is blueish - its highest channel is blue. Some people are more sensitive to that, other's can't see with sufficient sensitivity.

Black is often displayed gold/brown in washed-out photos; even in real life, shiny black fabric with light shining on it can look gold/brown for some materials. Some people's brains auto-correct the colors one way or the other.
(defun fibs (n &optional (a 1) (b 1)) (take n (unfold '+ a b)))

King Author
Posts: 736
Joined: Sun Apr 12, 2009 12:30 pm UTC
Location: Pennsylvania, USA

### Re: How does "The Dress" illusion work on a computer screen?

Flumble wrote:As discussed in here, the discrepance doesn't happen because of the RGB values, but because of the perceived context. If you assume the lighting is blueish (like the left part of the linked comic), the inherent colour of the dress will be assumed to be gold and white. If you assume the lighting is whiteish (like the right part of the linked comic), the inherent colour of the dress will be assumed to be black and blue.

And, yes, people will assume the question is about the inherent colour of the dress, not the RGB values of the picture.

Does color perception really work that way? It's based on your assumptions about the lighting?

(Also that xkcd comic doesn't mean anything. It's just a colorized version of the checker shadow illusion -- the actual Dress photo's RGB values are objectively white/gray and yellow-ish. And who in the right mind would look at the dress on the right side of the xkcd comic and call the dark part black? Dark yellow or even brown, but surely not the same color as ink.)

(Also also, the lighting in the original photo is clearly bright white. Just look at it...
https://en.wikipedia.org/wiki/The_dress ... nomenon%29
...the top-right corner is a massive blaring white beam, the entire right side of the photo is.)

Nyktos wrote:Right. I could take a picture of a white dress under, say, red light. Undoubtedly if you examined the RGB values of the pixels making up the image of the dress, they would be shades of red or pink. But people would likely have no trouble telling that the dress is actually white, because the human brain is capable of correcting for the lightning. In the famous photo, you can't see much background, which makes it hard to tell what the lightning is actually like, and apparently different people's brains made different guesses.

That metaphor doesn't work. Your photo would obviously appear to everyone to be red. Yes, we could understand that your dress is actually white and merely appears red because red light is being shined on it. But people wouldn't perceive it to be different than the actual RGB values of the image file. No one without a form of colorblindness would look at your photo and see a green dress.

With The Dress, the RGB values are unarguably X, but some people see them as Y.

Sizik wrote:
King Author wrote:Colors in images are stored as discrete values (lossy format wiggle room notwithstanding). They should appear the same to everyone, regardless.

Oh really?

I'm aware of that illusion, but it's not comparable to The Dress because it's grayscale. Everyone agrees that the colors involved are gray and white, and the illusion can be easily revealed by eyedropping color from square A and painting it on square B to see that they're the same. When I took an eyedropper to The Dress, no such illusion was revealed.

At the very least, I can't possibly believe that anyone with normal vision sees this...
https://en.wikipedia.org/wiki/File:The_ ... non%29.png
...and this...
http://www.justjaredjr.com/photo-galler ... lor-is-04/
...as identical.

Xanthir wrote:
King Author wrote:I understand the science behind the optical illusion of The Dress, at least as much as anyone without an optical science background can, but I've never been clear on how the illusion was supposed to have worked on a computer screen. Colors in images are stored as discrete values (lossy format wiggle room notwithstanding). They should appear the same to everyone, regardless.

Eyes are not computers. They do not see things the same from person to person. In addition to bio differences, our brains do *significant* context-sensitive alteration to the colors we see to produce the colors we "perceive".

(If you have two books side-by-side, one in shade and one in sun, the black ink on the in-sun book is brighter than the white page of the in-shade book. We still consider it darker because our brains aren't stupid.)

But that's completely different. Just earlier today I walked into the kitchen to see that someone had spilled sugar. From the hallway, it looked dark gray. Once I got closer and stood above it, though, it looked white. The difference was the lighting relative to my perspective. But my literal physical perspective was completely different between the two viewings. If I took a photo from the hall, the sugar would look gray. From overhead, it'd look white.

The key point there being, the photo is a static snapshot of a moment in time from a single, immutable perspective. If I took the hallway photo and passed it around, no one would think the sugar looked white or orange or purple. How could they? The photo preserves a single set of lighting conditions.

Xanthir wrote:(Also monitors aren't calibrated perfectly, and will display colors slightly differently.)

...

...

I can't believe that didn't occur to me. That probably accounts for the vast majority of the whole thing.

Xanthir wrote:
I just !gimage'd The Dress. Some of the images look black and blue, but most look white and gold. I copied one of each such image into Paint, and used the eyedropper to take a peek at their RGB values.

A random sample of a gold area from a gold-and-white image...
R: 128
G: 111
B: 75

Unarguably a brown/gold-ish sort of color.

A random sample from the "white" area...
R: 130
G: 140
B: 175

Again, unarguably, well, grayish. But clearly not black or blue.

A random sample of blue from the blue-and-black dress...
R: 20
G: 48
B: 118

And black...
R: 30
B: 33
G: 47

Irrefutably "blue" and "black."

Sampling color data from over a dozen images, they're almost all clearly, unarguably gold and white. RGB values don't lie.

Note that the "white" is blueish - its highest channel is blue. Some people are more sensitive to that, other's can't see with sufficient sensitivity.

Black is often displayed gold/brown in washed-out photos; even in real life, shiny black fabric with light shining on it can look gold/brown for some materials. Some people's brains auto-correct the colors one way or the other.

Ah, good points.
I have signitures disabled. If you do, too...you can't read this, so nevermind >_>

Nyktos
Posts: 138
Joined: Mon Mar 02, 2009 4:02 pm UTC

### Re: How does "The Dress" illusion work on a computer screen?

King Author wrote:That metaphor doesn't work. Your photo would obviously appear to everyone to be red. Yes, we could understand that your dress is actually white and merely appears red because red light is being shined on it. But people wouldn't perceive it to be different than the actual RGB values of the image file. No one without a form of colorblindness would look at your photo and see a green dress.
It's not a metaphor, it's the same effect modulo the disagreement on what's being depicted. You seem to think that people were disagreeing over what colours appear in the image file, when the actual disagreement was about what colour the dress is. Under normal circumstances, nobody is actually thinking about what colour the pixels are.

And no, of course nobody would see a green dress, but if I edited it just right I might be able to trick someone into seeing a pink dress.

King Author wrote:
Xanthir wrote:(Also monitors aren't calibrated perfectly, and will display colors slightly differently.)

...

...

I can't believe that didn't occur to me. That probably accounts for the vast majority of the whole thing.
Unlikely. I consistently saw white-and-gold on various devices, and my dad saw it differently on the same screen I was looking at. (Interestingly, he claimed it was purple, but that's probably more to do with differing definitions of "blue" than anything to do with the specific image.)

You definitely can screw around with the display settings on your monitor to make it appear one way or the other, but if it's calibrated "normally" that's not the main factor.

Sizik
Posts: 1242
Joined: Wed Aug 27, 2008 3:48 am UTC

### Re: How does "The Dress" illusion work on a computer screen?

King Author wrote:At the very least, I can't possibly believe that anyone with normal vision sees this...
https://en.wikipedia.org/wiki/File:The_ ... non%29.png
...and this...
http://www.justjaredjr.com/photo-galler ... lor-is-04/
...as identical.

If I enlarge the Wikipedia one to the same size as the second one, they look the same to me.
gmalivuk wrote:
King Author wrote:If space (rather, distance) is an illusion, it'd be possible for one meta-me to experience both body's sensory inputs.
Yes. And if wishes were horses, wishing wells would fill up very quickly with drowned horses.

Angua
Don't call her Delphine.
Posts: 5917
Joined: Tue Sep 16, 2008 12:42 pm UTC
Location: UK/[St. Kitts and] Nevis Occasionally, I migrate to the US for a bit

### Re: How does "The Dress" illusion work on a computer screen?

Yeah, the whole dress thing came about because two people on the same computer saw it as different colours...

I would read up on all the processing that goes on in the eye itself before it even reaches the brain. Like, seriously, the retina is complicated.
Crabtree's bludgeon: “no set of mutually inconsistent observations can exist for which some human intellect cannot conceive a coherent explanation, however complicated”
GNU Terry Pratchett

gmalivuk
GNU Terry Pratchett
Posts: 26666
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

### Re: How does "The Dress" illusion work on a computer screen?

King Author wrote:Does color perception really work that way? It's based on your assumptions about the lighting?
Of course it is. That's why the checker shadow illusion works at all, because you assume one area is in shadow and adjust your mental model of the scene accordingly.

With The Dress, the RGB values are unarguably X, but some people see them as Y.
Yes, just like a red-lit scene has red-dominated RGB values but people can see things as not-red colors.

Sizik wrote:
King Author wrote:Colors in images are stored as discrete values (lossy format wiggle room notwithstanding). They should appear the same to everyone, regardless.
Oh really?

I'm aware of that illusion, but it's not comparable to The Dress because it's grayscale.
Being grayscale has nothing to do with it. The same illusion can be done with colors. The difference between the checkerboard illusion and the dress illusion isn't the color of the objects, but the assumed color of the light source. The colored checkerboard squares look to be the same hue, but different brightness.

To get the different hue illusion, you need situations that look like there's a color to the light itself. Here's one that was posted in the comic thread (which you apparently never bothered to read for some reason?). And another one I found:

The center cross-piece is the same RGB color on each side of the image, but "looks like" a totally different color "in the real world" because of the apparent color of the lighting.

The dress illusion is more like these latter ones than like the grayscale checkerboard illusion, due to the combined effects of shadow and non-white light.

At the very least, I can't possibly believe that anyone with normal vision sees this...
https://en.wikipedia.org/wiki/File:The_ ... non%29.png
...and this...
http://www.justjaredjr.com/photo-galler ... lor-is-04/
...as identical.
Which picture on the second page are you talking about? The big one is the same image as the wikipedia link.

Xanthir wrote:(Also monitors aren't calibrated perfectly, and will display colors slightly differently.)

...

...

I can't believe that didn't occur to me. That probably accounts for the vast majority of the whole thing.
No, it most certainly doesn't, because the illusion persists even with people who are looking at the same monitor, where one sees it as white and gold and the other sees it as blue and black.

Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

Decker
Posts: 2071
Joined: Tue Jan 09, 2007 4:22 pm UTC
Location: Western N.Y.

### Re: How does "The Dress" illusion work on a computer screen?

Sizik wrote:
King Author wrote:At the very least, I can't possibly believe that anyone with normal vision sees this...
https://en.wikipedia.org/wiki/File:The_ ... non%29.png
...and this...
http://www.justjaredjr.com/photo-galler ... lor-is-04/
...as identical.

If I enlarge the Wikipedia one to the same size as the second one, they look the same to me.

Oddly enough, yes, they look exactly the same color to me.
I was angry with my friend. I told my wrath. My wrath did end.
I was angry with my foe. I told it not. My wrath did grow.

SDK
Posts: 698
Joined: Thu May 22, 2014 7:40 pm UTC

### Re: How does "The Dress" illusion work on a computer screen?

Decker wrote:Oddly enough, yes, they look exactly the same color to me.

They're the same image, assuming he's talking about the first, larger image on justjaredjr.

I like those two crosses you posted there, gmalivuk. Never looked into this enough at the time to find a good example of the phenomenon.
The biggest number (63 quintillion googols in debt)