Hacker News Reader

I completely agree that's what's going on. But what's fascinating here (to me at least) is how individual our interpretations are.

Can you see what assumptions the gold/white crowd is making in their interpretation? Even with the slider in the OP I can't see it. :) And even worse: I can't understand what assumption I need to make (about the color of objects in the background or shadows or something) to be able to see a gold/white dress.

Phrased differently: I can't imagine how I would construct a scene with a gold/white dress and then take a picture that looks like the one in the OP (regardless of the position of the slider). Scary! :)

do you really see "black" here ?

https://cdn0.vox-cdn.com/uploads/chorus_asset/file/3452082/n...

Objectively, there's only "yellow-brown" (gold) and "light blue" which most people "mentally" correct to "white"

I'm one of the light blue/gold people and couldn't image seeing black anywhere. But when I look at the picture upside down I see it as blue/gold and when I mask the lower half, as blue/black with a gold touch.

http://s4.photobucket.com/user/ometauru/media/YNOPKWp_zpstcv...

Looking at the different things around and behind the dress, everything dark has a similar golden color. Maybe some people correct that to black in the same way most do with the light blue to white

I see blue/black, but blue/gold makes sense to me too. It's more white/gold that seems improbable to me.

I am the same, I cant see either white/gold or blue/black, and I am surprised that the options for most people seem to be white/gold or blue/black.

light blue/gold barely gets a mention in most articles.

local colour vs perceived colour LOCAL COLOUR VS PERCEIVED COLOUR LOCAL COLOUR VS PERCEIVED COLOUR

nobody is arguing that the perceived colour of the dress is black.

It's easier to understand the gold part than the white part to me. But I think I assume that the goldish hue comes more from the light, since the further down on the dress I look the more black it seems.

Phrased slightly differently: The yellow-brown (gold) color is most visible on the upper torso where the fabric (to me) is likely to reflect a light source specularly. On the lower part of the dress (where I assume) light is more ambiantly reflected I see black.

Looking at your color squares to the right it's quite clear to me what my visual cortex is doing: It's assuming that specular reflection expains the difference in color in the left column. So each color there is a superposition of two: the ambient color of the dress and the specular reflection of the light source. Separating those two gives me black dress and yellowish lighting.

UPDATE After writing the above I think I understand why it seems so weird to me to see white: I parse the light source as yellowish, so a white dress would never look blue in that light. To me seeing a white/gold dress requires contradictory hypothesis: the light source needs to be blue to explain bluish tint on white stripes; but it also needs to be yellowish to explain the "gradient" in gold/black stripes from top to bottom of the dress.

I'm the same as your update. I tend to enjoy these sorts of illusions and normally can see them "both ways" for some specific illusion. Not so on this one. The rest of the image (i.e. beyond the dress) leads me to also perceive the light being yellow-ish, in which case a white dress would certainly not appear blue.

Ha! I just had a moment where I could perceive it as white/gold: It was a cropped version showing only the belly portion of the dress with no background, displayed in isolation as a thumbnail on a news site. When cropped like that it made perfect sense. :)

I think people correct the blue to white, since for some reason they mentally assume the picture is taken outside illuminated by daylight where shadows can have a bluish tinge. http://en.wikipedia.org/wiki/Shadow#Color_of_shadow_on_Earth

When i turned the picture upside down with the light source in the back now at the bottom, which couldn't be the sun anymore, my perceived color changed from very very light blue to blue instantly. And now I can't unsee it:)

People are not correcting blue to white. The image is measurably white and not blue. Some people may be correcting back to blue but regardless of whether or not the original dress is blue there is no deep blue of the original dress in the image being displayed. We can measure the RGB or HSV or whatever of every pixel and see that objectively

I see overexposed black. The background is clearly blown out.

Why the condescension? Did you even read the top comment in this thread? The whole point is yes, some people do in fact perceive black in that picture. Because color perception is relative.

I find it funny that the HN crowd is arguing over this, given how in our line of work, I'm sure many of us have looked at an image on several different devices and have seen different shades and colors. Someone's monitor might have a higher contrast setting than your smartphone. And you're condescending to someone for perceiving a color that isn't objectively there.

>Why the condescension? Was not my intention. I'm not a native speaker. Honest question, to learn: How would you rephrase the post to remove "condescension"?

The words 'really' and 'objectively' in your post give the reader the sense that you think what you see is the correct way of seeing things and anyone who perceives it differently is wrong and somehow lesser.

Hope this helps.

Exactly. For most optical illusions you can switch between interpretations, but in this case I only see black and blue (yes the black is gold-ish). But the frustrating thing is that I saw it as white and gold for a moment yesterday, but I can't anymore. I'm starting to question my sanity.

I'm in the same boat as you. I saw it white and gold for about 5 minutes yesterday, and then it gradually transitioned to black and blue, in a "well, I guess that's blue" sort of way, until it was "no way that's white and gold".

I saw it as white and gold originally, but later when I looked at it I had scrolled down the page so that the top part of the picture was hidden, and I saw the dress as blue and black.

Now I see blue and black if I view the whole image or bottom 2/3rds of the image, and white and gold when viewing the top 1/3rd of the image.

I'm the opposite way. Even if I push the slider all the way to the left, I still see it as a white & gold dress, but in a nearly monochromatic blue light.

I put an orange light in my bathroom so that night-time visits won't affect my sleep more than necessary, and notice that the cap on my antifungal powder bottle, which I know to be orange, reflects that light as a similar color to the bathtub, which I know to be white. If I put the bottle on the tub, I can flip-flop between deciding that the bottle-cap is white, and that the tub is orange. When doing so, my perception of the colors change. The orangeness appears to fade in and out.

Not so with this picture. I can't push my brain to accept any other conclusion than the slider is changing the color of the illumination from blue to yellow, and the dress color is constant.

As a photographer I'm seeing the middle version as if it's lit with daylight blue, the yellow version as if it's overexposed with indoor tungsten, and the blue version as a cold underexposed flash shot.

What people see depends on monitor quality. Most monitors are nowhere close to neutral, and cheaper monitors still aren't neutral even after calibration.

But on top of that, colour perception is notoriously individual. Men and women have different colour perception, and many men are somewhat red/green colour blind.

FWIW, personally I have trouble seeing this as a white/gold dress.

I don't think that your monitor determines whether you see a white/gold dress or not. At first the picture looked blue to me, no matter where I placed the fader. I left the fader on the right, continued browsing and somehow I forgot to close the tab. When I eventually came back to this tab I asked myself: "Why did I open a picture of a white/gold dress? Oooh..." However, when I then moved the fader to the left it quickly became blue again.

I could imagine that there are two effects at work.

1. My brain has some kind of caching-error when I slowly slide the fader to the right as it refuses to update all information necessary to see the real colors. (not white and gold but the actual blue tone)

2. The bright white spot on the right hand side is sort of a focal point. When you look at it first it alters how you perceive the rest of the picture. (causing an afterimage-like effect)

if 1 doesn't happen (i.e you see the photo of the right hand side first) and 2 does happen, then the picture looks white/gold (to me)

White snow looks blue at dusk

White snow looks gold under the sunlight

Your brain controls for these effects (mostly) and you still think 'white snow'

> I can't imagine how I would construct a scene with a gold/white dress and then take a picture that looks like the one in the OP

I can. To me, this initially looked like a situation where the background is lit with incandescent bulbs with the dress being in sunlight. In a photo where you have parts of the scene that are lit with incandescent bulbs and other parts are lit with sunlight, correcting white balance for the incandescent part of the scene (yellow -> white) will make anything that's white in the sunlit part of the scene very blue. That's what I was seeing here.

I first saw it on a friends laptop which has a shitty screen with the brightness cranked too high and it looked gold/white to me. When I looked at it on my desktop (which I have set so that blacks are as black as the monitor gets) it doesn't look anywhere near white/gold.

I saw it as white/gold until I saw one of articles showing better pics of it in blue/black. Now I can only see it in blue/black, even in the original photo I saw that I swore was white/gold.

I couldn't see the white/gold until I saw a much smaller and heavily cropped version of the picture. After it enlarged some, it just became the blue/black dress again even in the smaller form.

Likewise, I cannot see a white dress in that. Even with the slider all the way to the right, it is still a light blue in the image, and makes me think it's a blue-of-some-shade dress in real life.

My perception of the dress's color has flipped from white/gold to blue/black and now back to white/gold again over the course of the last day.

I can say that, in my case, there's been a fairly strong correlation between my own ambient lighting conditions. I've only seen white/gold in well-lit settings and I've only seen blue/black when I was laying in bed with all the lights off. Whether that's coincidence or not I cannot say, but I felt it was an interesting data point.

I think it's reasonable that we would have a default light-color model for illumination by natural sunlight, and that our brains would use that model to subtract the color of natural direct and diffuse sunlight from an objective color measurement upon the retina to determine a perceived inherent color for an object, which is then held as a constant. This would allow the same rock or tree to be recognized as a landmark at both noon and civil twilight.

Since the invention of artificial illumination (starting with fire), humans have had access to multiple light sources, which may be of different colors and coming from different directions. The default illumination model is insufficient to handle this without ambiguity.

This phenomenon is also used for dramatic effect in The Abyss, wherein the only light source available in the bottom of an oceanic trench is a monochromatic glowstick, and to disarm the nuclear bomb, Bud has to cut the blue wire with a white stripe and NOT the black wire with a yellow stripe. The actual wires used in filming were both black with a white stripe.

I really agree with the essentials of what you write:

- "surprised that this whole thing is confusing and dividing so many people"

- "the dress could be either and the photo doesn't really give enough context to say anything definitive"

I have used the same two arguments to explain why this hype appeared in the first place, and why the "scientific" analysis many people try to follow is conceptually wrong. Blog post and corresponding discussion can be found here:

https://news.ycombinator.com/item?id=9120940

Just a silly anecdote but a friend of mine had a room painted "this exact same pearl grey" used in another room. After having it grey, he realized it did not look grey but blue (I know why it was, because the only window of that room got the light reflected from a large bronze-green roof). He had it repainted a different tint. That was amazing.

Hey, I'm currently researching color theory. If you don't mind: Except Realist painter by Gurney, what are some other great resources about color theory that you would recommend? My main paper will be about color in marketing. Cheers, and good to see another scholar of colors!

Once I got to the word "Constancy" I started reading this in Reverend Lovejoy's voice.

Sweet constancy...