Apparently, I missed the big noise raised about the color of this dress.
Some people see it as blue and black. Others (including me) see it as white and gold.
It didn’t surprise me that the colors in an image could shift — it’s a well-known phenomenon in color perception. I was very surprised at the individual variation. I see it as white and gold, and can’t see it any other way. Others are the reverse.
So far, this is the best explanation I’ve seen.
Light enters the eye through the lens—different wavelengths corresponding to different colors. The light hits the retina in the back of the eye where pigments fire up neural connections to the visual cortex, the part of the brain that processes those signals into an image. Critically, though, that first burst of light is made of whatever wavelengths are illuminating the world, reflecting off whatever you’re looking at. Without you having to worry about it, your brain figures out what color light is bouncing off the thing your eyes are looking at, and essentially subtracts that color from the “real” color of the object. “Our visual system is supposed to throw away information about the illuminant and extract information about the actual reflectance,” says Jay Neitz, a neuroscientist at the University of Washington. “But I’ve studied individual differences in color vision for 30 years, and this is one of the biggest individual differences I’ve ever seen.” (Neitz sees white-and-gold.)
Usually that system works just fine. This image, though, hits some kind of perceptual boundary. That might be because of how people are wired. Human beings evolved to see in daylight, but daylight changes color. That chromatic axis varies from the pinkish red of dawn, up through the blue-white of noontime, and then back down to reddish twilight. “What’s happening here is your visual system is looking at this thing, and you’re trying to discount the chromatic bias of the daylight axis,” says Bevil Conway, a neuroscientist who studies color and vision at Wellesley College. “So people either discount the blue side, in which case they end up seeing white and gold, or discount the gold side, in which case they end up with blue and black.” (Conway sees blue and orange, somehow.)
That’s a good general description of variations in color perception, but the mysterious dress is unaccounted for. Why does my brain say it’s one color, but someone else’s brain says it’s a different color? What’s the cue? I tried cutting out bits of the image and displaying them with different backgrounds in Photoshop, but I couldn’t shake my initial perception.
The neatest thing about it, though, is that it undermines our trust in the accuracy of our color vision. If we’re seeing this one image differently with different brains, what other differences are there that aren’t so clear cut? Forget “What is it like to be a bat?” — what is it to be like that human being right next to you?
By the way, the dress is on Amazon, and it’s definitely blue and black, but you should read the reviews.
P.S. In the real world, you won’t have trouble telling what color the dress is. The ambiguity is a property of the specific image.
Is the trim gold for you? I see a blue dress with black lace stripes.
Those who are seeing it white and gold (I am): Scroll down so you’re only seeing the very bottom of the dress. Also, try backing up and looking at it from afar.
I love the fact that this is showing so many people how eye-witness accounts, including their very own, aren’t intrinsically trustworthy.
I can definitely see how the part that some (including myself) believe to be white could be seen as blue. In my mind, though, the light seems to be coming from the other side and I attribute the bluish appearance to be simply shadows. However, I cannot in any way convince myself the gold part could ever be black.
I see it as a lightish blue with a hint of purple, possibly caused by being in a shaded area with a high contrast background, and gold very much like the center part of the second image (after looking it up, it would appear to be cornflower blue, maybe). How people can see it as the other two is odd but fascinating.
If I didn’t trust pz I would be adamantly denying that there is any remote possibility that dress is blue. The blue dress must be a different dress with the same pattern.
I guess given that the object itself is not confusing what we’re talking about is the perception of that one image, which I have an incredibly hard time believing anyone can see as blue. I mean, if you take it into an image editor and eyedrop the colours you get a bunch of shades of blue grey, and a bunch of orange-brown. I want to see every shade represented in a chart, but I’m far too lazy to actually do it.
Perceived colour is distinct from defined colour though. I need to have a long, acrimonious argument with one of these wrong headed blue-seers.
I would hazard a guess that the reason is everyone’s brain does the colour correction differently. It probably isn’t hardwired in, but develops in response to visual stimulus when we are infants. So for a boundary condition, some people will end up on one side of the boundary and some on the other, and it may depend on a whole host of varying factors such as what kind of background lighting you were exposed to as a child.
Last night, on BuzzFeed via Fark: I clearly saw royal blue with black trim (like on Amazon). I thought the other colors were a hoax and moved on.
Around noon now (nearby window letting in cracks of sunlight): Looked at the same image. White and gold trim, just as obvious. 0.o
Looks like blue cloth and black cloth in warm lighting to me, partially because the area around the dress has yellow shadows (particularly the rack of black and white stuff to the lower left), even the leftmost image has this effect, since the background looks very very yellow to me.
I can kind of make it appear less blue by cranking up the brightness on my screen (I usually have a very low brightness) and messing a bit with the colours, though, so I suspect it might be a combination of the settings on the computer one is using and the variance in colour perception between individuals. We’re all a bit dissimilar, and some may be slightly more sensitive to blue than others. Considering how many cones we have in our eyes, I’d be surprised if there wasn’t a variance that could create disagreement.
Probably also a bit perceptual, though. I said it looked like black cloth in warm light because the exact colour I see on the black is gold and yellow, but in such a way that it would have to be black cloth considering the lighting there seems to be. ^_^’
On further reading it seems that the significant factor is whether you think it’s in light, or shadow.
I see the dress in the shadow of the bright lights behind it, and I see white and gold. What do other people think, shadowed or lit? White or Blue? Is there a correlation?
That’s so weird–the first time I saw it, it was definitely blue and black and I wondered what the white-and-gold people were talking about. I then scrolled down to read some older posts and when I came back up, it had changed to the middle of the three photographs, a darker white and gold. How can that be? “Who’re you gonna believe, me or your lyin’ eyes?”
Confusingly, that’s basically what Nagel was asking about in the first place. It’s even worse than that: Nagel asked what it’s like to be Nagel. Seriously. But bats are weird (no offense, bats) or different from us in ways that are mostly irrelevant, and he wanted it to sound like a harder question than it actually is. You’re expected to go off the deep end, because the title refers to some bizarre alien flying mammal with magical sonar superpowers you don’t comprehend, then you don’t even bother to read what the real paper really says since it has accomplished its mission in life. But he was certainly talking about you and me and himself and every other human being. Of course, this paragraph might have convinced you that we can be pretty strange beasts as well, depending on how you look at us.
If it could get any more confusing than that, he also wasn’t exactly saying science can’t explain color vision or individual differences in perception or whatever. That’s not a big deal, but there’s some other mysterious ineffable thing which is a big deal. It’s still not clear what the issue with qualia is even supposed to be, but what they’re apparently saying can’t be explained is more along the lines of the existence of anybody who is perceiving or experiencing or thinking anything at all.
Couldn’t we do without that? Why aren’t we zombies? Or are we? That is it in a nutshell. Also, if you were a zombie, how would you know? But if you’re asking that … then what’s a zombie and why does anybody care?
It’s the lighting. In a dark room, it would appear as all black.
Yet to me it’s totally unbelievable how you can see white and gold. There are some images where the lightening is really cranked up and then I see it, but before that, no.
Am I weird for seeing it (particularly the main picture at the top) as blue and gold? It seems like everyone else calls it blue/black or white/gold….
So basically this answers the question of ‘does my blue look like your blue?’ Obviously, my blue looks like your white, and my black looks like your gold. Problem solved.
* For a generic value of ‘you’, where ‘you’ is a person seeing the dress in white-and-gold.
Count me as another one whose perception of the colors has changed. When I first saw it, I saw white-and-gold clear as day. Then I looked again and saw…blue-and-brown? And then by scrolling up through the picture starting at the bottom I was able to make myself see the blue-and-black.
And so an explanation of it being a matter of our brains making it up on the fly makes so much more sense to me than the suggestions I’ve seen tossed around about actual physical differences in our ability to see the colors. This certainly seems more like a brain trick and I can make it go three different ways now. Nifty.
freemage
It’s one or the other, THERE IS NO MIDDLE GROUND!!!
You should know that by now.
That’s kinda what I see, but the “blue” looks like the kind of blue that you get when white is covered in shadows. While I see a bluish color, my brain tells me it is actually white. The gold, however, is clearly gold, and I cannot fathom how it could ever be seen as black.
freemage, same here. From afar, I see blue and brown.
I think the reason this is so shocking is because people are talking about it in an either blue/black or white/gold way. But people are actually seeing varying shades (that person in the article says they see orange and blue).
I see the colours rendered on the screen – light blue and gold. Try it with the colour picker in paint – that’s what you get. Clearly, my brain refuses to compensate for someone else’s inability to use white balance. :)
I don’t understand how anyone could come up with white and gold, or with blue and black. You’re all nuts.
VERY blue and dark brown….
Is this the same phenomenon behind the old practice of adding bluing agents when laundering white linens or interpreting some older peoples’ white hair as blue?
I wonder how many of us are artists? As someone who eventually wanted to be an illustrator/concept artist for games, do you see it as the uncorrected blue and gold (whether the dress is actually blue or its a white that is currently shaded blue) or if its the other two versions?
Appears gold and slightly blue-ish white to me, as if it was gold and white and in shadow.
I saw it as white and gold at first and couldn’t see it any other way, but then I looked at it a few minutes later, and it’s blue and black now. I can’t get it to switch back.
At any rate, the popularity of this picture is a good chance for some science educators to grab the public’s attention! :)
Somehow I spent most of yesterday being blissfully unaware of the dress controversy. Then I went to Mashable, and it was everywhere.
I’ll take the llamas instead.
No, you’re not weird, that’s the way I see it as well. At least now I know what today’s XKCD is about.
Definitely “white” and gold. I put white in quotation marks because it’s not pure white; it’s blue-gray tinted, which is confirmed with a swatch. But I’d generalize it to white. The gold is nonnegotiable.
In the trio of dress image, the rightmost one is “blue” and gold. I put blue in quotation marks because in absolute color, it would easily fall within what I consider blue range – again confirmed by a swatch – but if I saw that image and hadn’t seen the others, I’d likely rate it as “probably white, with allowance for shades of blue and gray”. The trim looks absolute brown, but I’d probably call it gold or yellow. The shadows are dark gray but enough is still lit that I can tell what the color is.
There also seems to be a lot of ambient yellow light around, so I can see that turning the blue to white. Black to gold would also make sense, if the lace is sufficiently shiny.
Something about the colors on the Amazon dress photo seems off, like I can’t quite look at it, like I’m missing detail. That photo is definitely blue and black, aside from that caveat.
Arawhon, I had career aspirations of making a living in fine arts before the economic reality of that shifted me elsewhere, and I definitely looks like some shade of blue/some shade of brown, and upon reading that people saw it as white/gold, blue/black, & that the actual dress is blue/black, I thought “wow, the color correction in that image is waaay off!”, but it didn’t alter how I saw it.
if the only evidence we have that the color of the dress (image) is seen differently is that from people on the internet we have too many variables for me to say anything other than people on the internet see this image differently nothing more is possible.
There are at least 2 things that are different, the display that it is being seen on and the person see it. There is also the unknown environment that it is being viewed in the ambient light and the reaction of the display to the light, none of that is controlled and are independently variable. How can you tell which is which. The image itself also has something to do with it. the black fabric does not seem to be a flat black with the thread dyed i.e. not woven but a print with some shine to the surface.
well?
uncle frogy
It’s white and gold, taken in low light. The blue and black version is a totally different dress.
A friend of mine showed this to me yesterday. She initially saw it as white and gold. I saw it (see it) as blue and black. Rather I see it as navy over exposed and washed out and a washed out black with some gold because of the yellowish light around it. I looked around and saw a few people had imported the image into photoshop and color matched, showing the colors in the picture (not real world) were shades of blue and brown (by photoshop’s color palette). When I showed my friend the colors in isolation she saw them as blue and brown, then when she looked back at the picture she saw the dress as blue and black (brown).
I remember when I first saw it, it changed from white and gold to blue and black. I thought it was actually a practical joke and the image was a gif.
@unclefroggy
I know of a couple of households that all looked at the same picture on the same screen and different people saw different colors. At least one friend has seen it change. If the magazine accounts are to be believed, several full offices of people had the same experience.
I see blue and brown. Why am I alone in the blue and brown camp?
My perceptions seem to line up with themadtrapper’s: it looks white and gold, with the white blue-tinged as if in shadow or blue light, so I can easily agree that it’s actually a brightly-lit blue.
The “gold” is gold, though, or rather an orangey-brown – by which I mean that, whatever the colour of the real-world physical object pictured, the camera has produced gold-orange-brown pixels. It’s very strange to me that anyone’s eyes would automatically correct for that, to the extent that they can’t even see the colour I do.
I saw (and continue to see) blue and dark olive-greenish-gold (brown gold is a close enough description, but I perceive some green in those stripes.)
@27:
You refer, of course, to the blue llama and the gold llama?
(As to the dress…I’m still waiting to see any part of it as black.)
Okay. It’s really not individual variation. I have seen many, many different versions of this picture now. The first one I saw clearly looked blue and black to me. This one looks white/light blue and gold to me. It’s the different images that cause the confusion, right?
if I take that image, and look at a bit of the color in a graphics program, the gold some varient of this: R=25, G=109 and B= 75 If I take a bit of the white/gray the RGB is 131,137,171
now if black’s rgb is 0,0,0 and blue is 0,0,255, I think those of us who can see gold and white are seeing correctly :)
I’m seeing it as light blue that could be a shadowy white, and gold. It might not be the one linked above; it could be this: http://www.amazon.co.uk/Roman-Womens-Detail-Bodycon-Dress/dp/B00SJETV4U/
Same style, different colourway. (There’s a couple of ‘reviews’ there that indicate I’m not the only person who thinks so.) I’m still puzzled by the black appearing brown though.
@40
I saved an image (to make sure it wasn’t a GIF and wasn’t changing on me) and kept it on one monitor, checking on it.
It changed colors dramatically several times.
Never trust your eyes again. :P
@saganite #40
Nope. I first saw it last night, when a woman at our weekly Team Trivia game was showing everybody the image on her phone and asking what they saw. Everybody was seeing the same image. She said she clearly saw white, with maybe the tiniest bit of bluish tinge in the shadows, but she was baffled that anyone saw it as blue. I clearly and unequivocally saw blue. Pure medium blue. Periwinkle. With goldish-greenish-brownish trim.
This whole thing was a case study in why “eye witness accounts” are often useless.
I would be curious to know the kinds of screens people are viewing the photo on, since computer screens have unique display capabilities and color profiles. A difference between warm and cold white balance might be enough to throw a person’s perception of the image in another direction.
@Howard Bannister #43
@Lady Mondegreen #44
Uhm, well. Considering when I just came back to this website to check the comments, it now flipped for me to blue and gold, I suppose you’re right. Weird. I have now seen this image both ways. I’m still not sure about the individual variation thing, though, if somebody like myself can see it one way first and the other way next. Or maybe there are people who only ever see it one way?
…all of this is assuming PZ didn’t switch out the images while I wasn’t looking to mess with us… :-D
Even in the left hand picture, the one with the light/brightness jacked up, I still see (light) blue and lighted black.
I’m wondering if “having taken a shitload of pictures with a digital camera where the colours are totally off” plays a role in mentally adjusting to blue and black.
What I find amusing are the comments by people who are certain they are right and the other side is wrong and attempt to enforce that belief by various version of ad hominem. It is apparently very important for them to correct no matter what. Kind of reminds me of another arena of debate.
Oh, and my two kids and my husband viewed it on my screen, one of the kids saw white and gold, the other one didn’t want to comment and my husband said he saw white and gold earlier but blue and black now.
Oh heck i just scrolled back up and now i’m seeing it as mid-blue and black. Colours, how do they work?
http://www.bbc.co.uk/news/uk-scotland-highlands-islands-31656935 — it was definitely the blue and black version.
I can see it both ways! With my contacts in I see white/gold and with my contacts out and glasses on, I see blue/black – really disconcerting to go back to the same pic and see it differently. I’m guessing the two different lens media are filtering slightly different wavelengths, and based on the explanation above, flicking my brain to interpret it one way or the other. It still does my head in though, even with the explanation. At least I don’t think people are lying about what they are seeing.
I was reading this in Feedly and saw the dress and thought, “Well, it’s obviously off-white and gold: what else could it be?” (And I’m good with colours: I work with them for a living.) And then my husband came home and I went to greet him and after about ten minutes I came back to my computer, and there on the screen was a light-blue-and-black dress and now I am FREAKING THE FUCK OUT. I thought it was an animated gif that had changed colours or something. I can’t see it as anything except blue-and-black no matter what I do. I’m still not absolutely convinced that the picture wasn’t somehow changed when I was out of the room.
Seeing it on different screens doesn’t answer the question – I showed a roomful of students the image (although it’s hard to know which is the “right” image now, so since there are so many manipulated versions out there now, I picked one that seemed the most obviously white and gold to me), and 2/3 saw blue/black, 1/3 saw white/gold. Not all of them had encountered it already.
(sorry, the important point being that it was on a single screen, and it was via a classroom projector)
(side note: it was relevant to the class, because we talked about how perception varies and can introduce a level of bias to observational experiments a couple of weeks ago)
I can’t see the blue and black version and it is frustrating! I think I wanted this to be one of those optical illusion where you can change something dramatically as a perceptual trick if you stare it too long, or stare elsewhere and look it again, or one of those things that could be interpreted differently based on how you are thinking. But just inherently seeing colors differently? Somehow that manages to be both obvious to me (psychology and neuroscience background, of course there are differences in people’s perception) and depressing to me (gut level feeling that I am missing out on something).
If I look at the whole dress, it’d white and gold to me. If I look at the bottom third alone, it’d blue and gold. If I take my glasses off, my 20/1500 vision tells me the whole dress is a blue and gold blur.
What’s weirdest to me is that the 3 small pictures side by side clearly look like 3 totally different dresses (white/gold, blue-gold, blue-black) even though it’s the same dress in different lighting. The big picture is the middle of the small pictures, right? In which case it’s weird that the same picture looks white/gold alone, but blue/gold if it’s flanked with other dresses.
I teach a 100 person class this afternoon and will perform the experiment.
“The ambiguity is a property of the specific image.” Bingo! Give that man a cigar…sadly, the corollary to that thought is “If it’s on the Internet, it must be true!” Who knows how much the original image was manipulated? Orwell is laughing quietly to himself somewhere…
@anteprepro #58
Try looking at the colors in isolation, then looking back at the image. If you isolate the lighter areas and just look at that everyone I have showed it to sees blue. Isolate the darker areas and depending on which stripe you look at it will look varying shades of washed out black brown to more gold brown. The part at the neck is the most goldy brown. After looking at the colors in isolation, keep them on one side of your screen and look at the dress on the other. If your screen isn’t big enough scroll back quickly. This worked of a couple of friends to get them to go from white/gold to blue/brown (washed out and yellow light reflecting black).
I do think that taking a lot of digital pictures and seeing how colors get washed out in that kind of light affected my perception of it.
It’s shot against a very bright background. If I thought it were lit like the background is, I’d think blue and brown. If I thought it were in the shade compared to the background, I’d say white and gold (just walked in off the street, for example, although yes, this dress is on a hanger or mannequin.)
Rereading The Master and his Emissary by Iain McGilchrist, about brain hemisphere differences, now in the monumental chapter 2. He would have it that contextual differences, tracked by the right hemisphere, have a big efffect on interpretations.
I can easily believe that different people’s eyes perceive colors very differently, because I’ve experienced it. I had cataracts, and although I didn’t know it because it happened very gradually, I was seeing the world dimly and with a yellowish tint. When I had the cataracts removed and replaced with artificial lenses, the difference was amazing. The most striking change was that shades of blue appeared much brighter and more saturated than before. All of this may be irrelevant to the discussion at hand, but I think it’s interesting that blue is one of the colors people are arguing about. And by the way, I say the dress is light blue and gold.
I suspect that the blue/white thing results from the human visual system having evolved in a world where shadows mostly have a blueish hue. So, a crappy photo of an unfamiliar blueish object, against an ambiguous overexposed background, could either be a bluish object that is well-lit, or a white object in the shade in an otherwise brightly-lit environment.
I perceive the dress as blue and a murky dark brown color, and can’t get it to switch to white and gold, for what its worth.
I can’t see the original image. which is the one in the middle of the three, as other than blue and brown. The other two are the alternatives as described. A trick of the light rather than a trick of the brain (see Goethe’s angry rejection of Newton).
There are reports that some people, looking at the same image, on the same monitor, at the same time, still see the colours differently. If these reports are true, that can’t be just a trick of the light, because you have two observers looking at it in identical conditions and seeing it differently.
If the reports mentioned above are true, it doesn’t matter one bit how much the original image was manipulated. The interesting phenomenon here is how two people, looking at the same image, in identical conditions (so they both are exposed to the same amount of manipulation at the same time) still see different things.
I see bone & tan. But that may just be my video card.
@63
I have had the lens replacement cataract surgery, but only on my left eye. I, too, noticed the effect on blue, and what’s remarkable is that I believe I actually see a little ways into the UV part of the spectrum. I first noticed it at a concert several years after he surgery, where there were big UV lights to either side of the stage. I was seeing a kind of purple glow emanating from them, not just the flourescent effect of the reflected UV from objects under them. When I closed my right eye, the effect was enhanced, and when I closed the left it went away.
I think it must be that the fluid of the natural lens of the eye (the lens is a fluid-filled sac) absorbs UV.
It’s been 20+ years, and the cataract in my right eye has advanced to the point of needing surgery soon. Really not looking forward to it. Recovery from eye surgery is not my idea of fun. But I suppose there have been some advances in the intervening two decades, so hopefully it won’t be too bad.
On my iPad, with the brightness turned all the way up, the top image looks bluish-white and gold; with the brightness all the way down, it’s unambiguously blue and black. Adjusting the screen brightness or the room lighting allows me to alter my perception almost anywhere in between.
XKCD provides a pretty succinct visual explanation of the illusion at work.
On my work monitor it was white and gold. At my home monitor the “white” parts are completely blue and the gold are dark brown.
My tablet won’t show the whole image at once. I get white and gold at the top, but the bottom looks blue and black. If I hold a perception and scroll carefully, I can go down okay, but going up always breaks on the lit-up bit. I think some folks may be looking at the top or bottom first, deciding from there, and making the rest fit. (Hmmm. Maybe a sexual/psychological test there – do you look first at the hips or the bosom?)
That different people looking at the same image respond differently isn’t unsurprising. We can expect that the threshold levels applied by different observers’ visual evaluators will respond differently, and that this will be a combination of physiological differences and divergent experiences. Like a green-and-red striped rugby jersey seen in green light will be black and white, seen in red light will be black and white (but with the black and white inverted), and seen in sodium light will be….?????
By the way, tell Freethoughtblogs that the advertising is not only distasteful, but so instrusive as to be nearly intolerable.
I’ve been a professional photographer and printer for over 50 years. Currently I photograph art and create reproductions on canvas. Color perception is extremely important to me. I saw the dress as blue and black, never as white and gold. Why? First, the two pictures on the left and middle are overexposed. This washes out colors, reduces their saturation. It is a lot easier to see a light blue as white if the colors aren’t saturated. Second, the background is an obvious giveaway. The light is yellow, which indicates the color balance of the original photograph is incorrect. Our eyes quickly see a 100 watt bulb as white (although it is only about 1800Kelvin, which is quite red) and equally adapt to seeing sunlight as the same white (12000Kelvin at noon) which is extremely blue. Light reflects off the fabrics and the color of the originating light will change the perception of the color of the object.
If I had this picture given to me, I would immediately have corrected both the exposure and the color balance (based on the background color). Even the dress on the right hand side is too yellow and overexposed (although better than the other two). Yellow light also tends to flatten contrast. This is why so many people see brown rather than black in the dress. It’s the fault of the photograph.
I have one artist who totally sees a different blue than what is actually in her painting. Even when the painting is held next to the monitor where she can compare two blues in snow, for instance, she insists the snow needs more magenta to match the print. I’ve given up, and just add 20% magenta to the snow.
So, the problem with the images are exposure, color of lighting, color inbalance. Also, different monitors can have decidedly different color balance. I’ll give you this though, PZ. If I couldn’t see the background to know that it was off color, I could make the dress any color you want. But also, things like exposure are glaringly wrong.
@68
Re: UV. I’ve heard of that happening to people. I haven’t experienced it myself.
Re: surgery. I had the surgery about 3 years ago, & it wasn’t bad at all. Right eye first, left eye about a month later. The surgeon only had to make very small incisions in my eyes. I had very little pain, & recovery was relatively easy.
This is a pretty great piece on the sociological factors surrounding the dress argument.
I see the original image as a light blue/lavender and dull gold. Opening the image in GIMP (make sure you set the color profile to sRGB!) and using the eyedropper, I see that the cloth is indeed various shades of desaturated blue and lavender, and the trim is composed of greys, browns, oranges and yellows. So, yeah, the colors in the image are light lavender and dull gold.
Of course, none of that tells us what color the dress appears to be under white (well, actually, as near as one can get to D65) lighting.
I learned all (well, not everything, but enough) about how lighting and background colors affect color perception learning lighting theory for theater, and I’ve also read articles like this, so the disagreement isn’t at all surprising. In my final exam for my lighting theory class, one of our “questions” was instruction to make a single piece of cloth draped over a chair appear to be a number of different colors by altering the color of the light. This is usually used to direct costuming and lighting decisions so one doesn’t wind up with costumes warping between different colors between scenes or even as actors move around the stage in a single scene (and in live theater, we don’t have color correction available as a postprocessing effect!), but it can also be used for very cool practical effects in things like dance numbers or magical or otherwise out-of-narrative transformations of the stage space where one can make it appear that the dancers’/actors’ costumes completely change in an instant by altering the perceived color or revealing a pattern that was previously obscured.
My class voted 75% blue-black and 25% blue-gold, barely anyone white-gold. But I suspect that’s because we were in a fairly dim lit lecture hall where the colors on the projected image doesn’t always match the colors on my monitor.
So that’s what this is about
http://m.xkcd.com/1492/
ahem
Just to reiterate.
24 ways to view the dress.
I initially saw white-and-gold. Then I clicked through and got stuck on blue and black, even after going back to the main page.
Looking down at an acute angle from above it appears white/gold. Looking up from below it appears black/blue. Looking straight at it from a normal viewing angle it looks brown/blue. This is using a Mac book pro.
Nogbert @ 82 – The reason the color changes when you change your viewing position is because you have an old style screen. If you move to the right or left, the color will shift towards red. The color isn’t actually changing when you go from top to bottom, but will change from light to dark. Newer TFT screens don’t have that problem.
If you look at the picture on the left, it is about four stops overexposed. The center image is two stops overexposed. The first thing any printer does is get the exposure right because you can’t make a color correction until you do. The reason black looks gold in the left picture is because warm light is illuminating the black, and when it is lightened to a light color it turns brown (or gold, as some see it).
I found this really fascinating. If you crop the photo showing only the top 1/4, the dress looks white and gold, but if you do the same for the bottom 1/4, it looks blue and black. When viewed in total, because it’s such a subtle change from top to bottom, it’s like the brain can’t figure out how to interpret the colours. Very cool.
I posted the contrasting images here.
I can see it both ways by tilting my laptop screen. Straight on I see white/gold, but a slight tilt changes it to blue/black.
Interestingly, I initially saw it as white and gold. I was viewing via my laptop.
About halfway through the post, I had to get up and go do something, so I snagged my tablet and loaded this up so I could read it in the other room. Lo and behold, when I loaded it up, I saw it as blue and black.
I thought to myself “Wow, is the color balance *that* off on my laptop screen? I’ll have to look into that.”
I came back, sat down, and now I can *only* see it as blue and black. I cannot, even through creative cropping or otherwise, make that perception change. But, when I first looked at it, I was 100% certain it was white and gold, and could not make my perception change then, either. Strange.
Initially it was dark blue and black. The blue has lightened to something I would term light purple, doesn’t really look blue any more. The black morphed into a mangy black or dark brown, with uneven splotches of green or gold. The mange spread while I watched. It seems to have stabilized now. Can’t imagine it getting to white and gold.
If everyone is judging this from laptops and such, then this phenomenon is clearly more about LCD screens and their polarizing filters (they have them) than it is about eyeballs.
Print it out on a properly calibrated printer and then I’ll believe it’s a visual phenomenon.
Until then I’ll consider to be merely an artifact of a technological limitation that has been widely known since forever.
Seriously, doesn’t anyone remember when they came out with “supertwist” LCD displays that were supposed to help minimize this stuff? Active-matrix LCDs replacing passive-matrix, etc.?
There is just no way the photo at the link is of the same dress as the photo in the post. Same cut yes, but different fabric. Also note the dress up top is in the shade with a very bright, over-saturated sunny background – easily enough to make the white object blue-grey. The dress in the amazon link would look even darker than it does if placed against the same over-ssaturated background while in the shade.
I cannot, for the life of me, see it as white and gold. No matter how hard I try. It’s actually driving me a little batty right now.
I only saw violet blue and brown until I scrolled down not paying much attention and saw a white and gold dress. :o
It really does just depend on whether your brain assumes the room is bright or dim.
That is so nifty.
It looks gold and white or gold and very light blue. Even with the images cut ala normolsen @84 it looks white-blue/gold for me.
Same here. In the middle picture (the one that was making the rounds) I see it as a muted blue and some kind of dark gold or dark bronze colour.
In the left picture, it’s basically the same but much lighter and I could see how some people perceive that blue as white.
In the right picture it looks a fairly deep blue and black to me.
@88 if it is just color balance then why can one person see the dress change color and how can two people looking at the same image on the same screen see two vastly different colors? It is not screen settings.
@antepropro #58
I saw it as pale blue and gold (although I could understand others seeing it as white and gold). Once I found the three versions of the image together, something changed. I can see it as white-and-gold, make it switch to blue-and-black (without looking away, just staring and forcing my brain to perceive it differently), then back again. It is one of the most bizarre experiences.
I research human vision. It’s really great to see this kind of effect get popularized! Vision is fundamentally an inference problem. Your retina passes luminance data to your brain, and your brain has to infer the best explanation for the data. It makes many assumption about light level, wavelength distributions, lighting direction, object shapes, sizes, and so on. Most of the time ambiguous data from one source can be resolved by another cue, such as motion or touch. In the absence of a disambiguating signal, though, the brain basically uses a Bayesian estimate with previous experience as a prior to produce a percept.
Visual perception amounts to the same methods used in science.
BTW, proper display calibration is extremely important to study vision properly, so I was suspicious that internet arguments were confounded by people using different displays. My wife sees white and gold no matter what display we use. I see blue and black. Even in the most overexposed image in the Wired article, it’s clearly a blue and black dress under extremely yellow lighting. I cannot make the material of the dress appear to be white or gold.
Hmm. The XKCD dresses look more similar to me that would expected under the different “lighting”. The blue of the dress, actually, looks exactly identical.
Anyway, I suspect most of you are thinking of the wrong coping mechanism. Yes, the human brain has elaborate mechanisms to detect object color from visual color given information about lighting (and these can go wrong even in pure nature) … but what is involved here is a related, but different, and completely learned, coping mechanism: inferring “natural” visual effect from pixels (or photographs, before computers). That’s also connected to inferring depth from these two-dimensional representations, and inferring orientation.
We know the difference between a picture and what we see directly, but there is no simple rule to reconvert from the picture to reality. The information is truly lost. That is even more context-dependent than the original color correction – and there is no reason to expect different people to apply the same rulebook.
What the photo doesn’t show is why the picture is overexposed.
You see, it is so bright because THERE ARE FOUR LIGHTS!
OK, clearly I am not normal. I see (lightish) blue and gold in any unalduterated version of the image, and I’ve seen it on a dozen different sites under different lighting conditions. My computer tells me that the colors I’m seeing are correct. Even today’s XKCD does nothing. I see the same colors on either side.
What bothers me is that hardly anyone makes a distinctions between what the image depicts and what it actually shows. Sure, what is depicted is a dress that in the real world may well be blue and black, but what the image actually shows is a dress in blue and gold (this is a fact, check via any color picker around).
I guess what is so fascinating and bewildering about this is that everyone is wrong to some degree:
– White/Gold people cannot perceive the picture’s true colors.
And happen to be UTTERLY WRONG!– Blue/Black people cannot perceive the picture’s true colors, but happen to be right about the real-world dress.
– Blue/Gold people do perceive the picture’s true colors, but cannot automagically infer the real-world ones.
tl;dr
YouWe are all wrong!Right, as someone who’s looked into human colour perception in his spare time, I couldn’t sit on my hands over this one. Sadly, it takes 1,500 words just to share my primitive understanding of the situation, but by the end you’ll understand why my first response was “I don’t know what colour it is.”
Things I’ve learned: 1) the strangest things can become internet sensations for the strangest reasons, 2) people need to get their brains checked, it’s obviously blue and black.
Anyone seen this yet:
http://ruinedchildhood.tumblr.com/post/112491912464
That’s the first picture where I saw a dress of this cut in white and gold (left one).
The right one looks blue and black (very close to the Amazon link colors to me).
I refuse to acknowledge this as a big deal. That person who took that photo of the dress probably doesn’t even computer! Shallow people don’t GET to impress me! I’m assuming they are shallow based on a photo of a dress they looked at and nothing else about them.
I don’t like it when people like things I didn’t approve first.