17

Buzzfeed, and other social media, are buzzing about this image: Magic Dress

When I see this image, I can only see a gold and white dress. However, others claim that the dress is actually black and blue in colour. Going by the Buzzfeed survey (as unreliable as it might be), 70% of users viewing the image believe it to be gold and white while 30% believe otherwise. Some people (also see the comments to this question) find that the dress changes colour for them. Is the dress black and blue in colour?


Moderator note: Original research and theoretical answers will be removed according to site policy.

JP Janet
  • 305
  • 1
  • 6
  • 5
    (I can't decide if this counts as a claim or not.) – Oddthinking Feb 27 '15 at 02:25
  • 6
    @Oddthinking Seems to be a fairly big question on social media right now, I'd vote for it being a claim. – rjzii Feb 27 '15 at 02:48
  • 3
    @rjzii: "Is Benedict Cumberbatch hot?" is another big question on social media. But is it a truth claim? This appears to be a "explain human colour perception on uncalibrated monitors" question. – Oddthinking Feb 27 '15 at 03:48
  • @Oddthinking Good point. Maybe we can keep an eye on the topic and if it appears in major news outlets then it would be on topic - if only so an explanation could turn up high in the Google search results? – rjzii Feb 27 '15 at 04:24
  • 1
    I see no claim that it changes color anywhere. – Sklivvz Feb 27 '15 at 08:27
  • 2
    Are you seriosly saying someone sees it as yellow-white? – Cthulhu Feb 27 '15 at 08:31
  • @JanDvorak But the claim all over the internet is exactly that it *does look* white for some people. I asked 3 persons, neither saw it as such (every one on their monitor). – Cthulhu Feb 27 '15 at 08:36
  • @Cthulhu I read the claim as "appears to be", which is somewhat distinct from "looks". The former is a claim about human perception, which is affected by the actual reflected light after white balancing (demonstrably blue), human expectations about most common materials (which suggests it might be white) and most importantly the human brain's own mechanism for performing its own white balance (which seems to be the main part of the question and explains most of the differences in color perception). To me it's surprising how confident people are about a color balance in a photo as bad as this. – John Dvorak Feb 27 '15 at 08:43
  • 1
    @Cthulhu Well, the first time I saw it *on my PC's browser*, I saw it as bright gold-white, *definitely no blue*. So, I closed the image, doing something else, and curious about this. The next time I saw this *on my phone*, it's blue-brown, and it's dark. Perhaps my brain really thought that it's white dress filled with blue light. Just my anecdote. – Andrew T. Feb 27 '15 at 08:45
  • 1
    Actually, OP might want to clarify what the actual _claim_ is. That it actually changes colors? Unlikely. Does it ask for an explanation for differences in color perception? That sounds off-topic for me. That the dress is white or that it is blue? That is on-topic, but I'm pretty sure you can't prove either from this photo alone. Is there a better photo of the dress? Also, in that case, OP might want to indicate his sub-question is just a sub-question. – John Dvorak Feb 27 '15 at 08:49
  • As with @AndrewT., I can confirm that when I first saw this question, the dress was white and gold. It is *now* blue and black. Voting to reopen. –  Feb 27 '15 at 09:31
  • @coleopterist So would you vote to open this, too: http://en.wikipedia.org/wiki/Checker_shadow_illusion What's the actual question in it, and what kind of evidence/data is necessary to answer it? – ChrisW Feb 27 '15 at 11:53
  • @ChrisW If there's a notable claim that can be answered using it, then I would vote to keep it in. The dress in this question has become something of a meme which makes it notable enough to warrant a proper answer. The scepticism here could be on any number of things: Does the dress change colour? Do some people actually see the dress as gold and white? Is it a hoax (i.e., are different images being displayed to different people?, is it a GIF?, etc.)? –  Feb 27 '15 at 12:25
  • @coleopterist Isn't it too subjective i.e. entirely subjective? You said you saw the color change, but it's a jpeg not a gif, so presumably the 'change' is subjective i.e. only in your perception. – ChrisW Feb 27 '15 at 12:34
  • 5
    @ChrisW I confirmed that I experienced the colour *change* as well (as opposed to some people only seeing one set of colours and others another). The fact that you know that this is an optical illusion doesn't make this question incompatible with this site. If this exact optical illusion has been explained in some other question, then this should be marked a dupe of that. Else, it should be reopened and answered. I mentioned GIF vs. JPEG w.r.t the Internet-audience-at-large (who will hopefully benefit from a good answer to this question). –  Feb 27 '15 at 13:33
  • When I opened this post, the dress was white and gold to me. After I read the chain of comments, I went back to the image and now it appears as Blue and Black. I even checked the image source to see if it wasn't a GIF or something like that. What is happening on this image is, certainly, surprising. – T. Sar Feb 27 '15 at 13:49
  • The problem with the "change color" claim is that I see it red and pink. Sometimes it's actually rainbow colored. I am lying, but you can't disprove me. – Sklivvz Feb 27 '15 at 15:00
  • @Sklivvz The OP amended the question with further references and this is indeed a really popular topic on the internet today. This still doesn't count as a notable claim? – T. Sar Feb 27 '15 at 15:41
  • No, but it's clearly unanswerable beyond anecdotal evidence/opinion/hearsay. Even beyond that, the question is clearly about a current unresolved event. – Sklivvz Feb 27 '15 at 15:58
  • Where's the video link I posted? It provides two photos which indicate very clearly that the dress is blue and black - and also how much this photo is oversaturated. – John Dvorak Feb 27 '15 at 16:14
  • @JanDvorak comments are not to provide answers, especially on questions on hold. – Sklivvz Feb 27 '15 at 17:25
  • @Sklivvz I don't see why the question can't be opened to see if anyone can actually provide an answer. Why jump to conclusions? For example, the Wired article (cited in some of the links in the comments) features explanations by neuroscientists on why the differences in perception occur and could be a starting point for a good answer. –  Feb 27 '15 at 19:33
  • @coleopterist I am not jumping to conclusions, the question has problems and it should be fixed to something answerable according to the rules of this site, especially if it's a viral topic which attracts new users. The non-direct/non-factual version of this is already on [cogsci.se]. – Sklivvz Feb 27 '15 at 19:39
  • @Sklivvz Refactored. –  Feb 27 '15 at 20:02
  • I've removed the speculative part. Note that the answer is trivial. The pixels of the image are blue. One just needs to find someone that has done the analysis. – Sklivvz Feb 27 '15 at 20:12
  • @Sklivvz Well, you deemed the non-trivial angle to this question to be anecdotal. And I tried to piggyback the colour change onto this question. Would reworking it further with [this article](http://www.businessinsider.in/People-are-beginning-to-see-the-The-Dress-change-colours-before-their-eyes-160and-they-cant-believe-it/articleshow/46398362.cms) change your mind? –  Feb 27 '15 at 20:23
  • Let us [continue this discussion in chat](http://chat.stackexchange.com/rooms/21509/discussion-between-sklivvz-and-coleopterist). – Sklivvz Feb 27 '15 at 20:27
  • http://xkcd.com/1492/ – vartec Mar 04 '15 at 07:59
  • For what it's worth, I have watched this photo appear to change colour. Yes, I understand that it's "all in the mind", but my experience with this photo is that how I see it depends on what I was looking at previously. When I first saw the photo, I saw white/gold. I looked at it later and saw blue/black. I kept staring at it, and the blue faded gradually to white, via a series of pale purple-blue colours; while the black faded to gold through a series of browns. – Dawood ibn Kareem Nov 04 '16 at 23:19

1 Answers1

19

Is the dress black and blue in colour?

TL:DR; Colour-Constancy illusion.

The dress in real life

Yes. The original dress is black and blue. The dress has been identified and there are many professional photographs which show its colours unambiguously. See reports referenced below.

The dress in the photograph

The photograph being circulated is a very poor snapshot which appears to be overexposed and has incorrect colour-balance. The pixel data values are mostly described as predominantly blue and brown. But this is not as relevant as many of us might at first suppose. See references below.

The dress on your computer screen

The hues on your computer screen will vary considerably, more so for the probable majority of people who don't perform regular colour-calibration of their computers and who don't concern themselves with limited gamuts and other technical issues.

However this answer will not take this factor into account, there is evidence of other factors that are probably adequate in themselves to explain what is happening.

The dress in your brain's visual system

The colour perceived by a human visual system depends to a surprisingly large extent on context and in particular about what clues exist concerning ambient lighting, shadows and other circumstances. This is the subject of colour-constancy in human vision

The processing in the brain attempts to maintain colour constancy regardless of the actual frequency of light falling on the retina from an object. In other words, regardless of any actual RGB values of an image on a screen. The human visual system (which is mostly in the brain, not the eye) does not faithfully measure incident RGB values and present those to conscious levels of thought.

Ambiguous context can cause the brain's visual system to "incorrectly" interpret the colour. A better way to express this would be that there are two sets of different real-life physical dresses and lighting conditions that will produce the same photographic (or retinal) image. Your brain has too few contextual cues to disambiguate the situation and therefore settles on an arbitrary interpretation that depends mostly on your viewing circumstances and physiology.


BBC reports

The BBC reproduced the conditions of the photo and interviewed a Professor John Barbur of the Applied Vision Research Centre, City University London. BBC report

Essentially it is claimed that this overexposed photograph represents an edge-case in human visual perception, our visual systems vary considerably from person to person and our brain's interpretation of the image can vary greatly depending on context - background in the photograph, lighting in our viewing position and so on.

Beau Lotto, a professor of neuroscience at University College London, says "the brain has evolved not to see absolutes, but to see the difference between things." Because colours that appear in sunlight look different from those that appear under streetlights, for example, our brains have to focus on the relationship between colours, not the colours themselves. BBC report

Newspaper reports

The Guardian newspaper has a report where Marie Rogers, a PhD student with the Sussex Colour Group discusses colour-constancy.

In our everyday lives, there are many changes in the colour of the light illuminating our surroundings. For example, the yellow glow of an incandescent light bulb versus the blue-ish hue of a fluorescent light. The light that an object reflects to the eye is a combination of both the colour of the object itself and the spectrum of the light source, which may vary. The brain is able to disentangle these two things and decide what colour the object is. Simply put, objects appear the same colour even if the light illuminating them changes – a concept known as colour constancy.

So, how does the brain keep colours constant? One way is by using reference points. For example, say you know your mug is white, but the light being reflected from the mug is slightly red. The brain can then discount a certain amount of red tint from the rest of the scene you are seeing. Other contextual knowledge may come into play, for example you are drinking coffee by the window at dawn. It makes sense for the light to be red-tinted as the illumination source is the sunrise. This is known as top-down processing. All of our perceptual experiences are informed by this kind of processing, resulting from context and previous knowledge.

The New York Times has a report that illustrates this cognitive process with respect to the dress in the photograph. Their sources are given as "Laurence T. Maloney, New York University; Eugene Switkes, U.C. Santa Cruz; Qasim Zaidi, SUNY College of Optometry; Journal of the Optical Society of America A."

Research

Journal of Vision - Surface color perception under two illuminants: The second illuminant reduces color constancy - describes how the presence of two sources of illumination can reduce colour-constancy.

It is likely the troublesome photograph included both ambient lighting and a photographic flashgun. Viewing conditions will vary for people vieing images on the Internet. The image lacks clear/unambiguous colour cues in it's background

Summary

So it's a kind of chromatic visual illusion, in some ways similar to the figure-ground illusion, where the brain struggles to settle on one of two possible interpretations of an image.

Other References

RedGrittyBrick
  • 24,895
  • 3
  • 100
  • 111
  • 1
    But what colors does the one from the photo have? You didn't answer the most important thing. The debate is on the photo, not on the real life dress – Alex Feb 28 '15 at 22:18
  • 10
    @Alex The colors in the photo are particular RGB values. There is no debate on what the colors of the photo itself are; the question is how different people's visual systems interpret the photo, which (in humans, at least) is based on trying to separate out the color of the object from the conditions of the photo. What color people see is what color their brains think the dress is, not the color of the actual pixels. – cpast Feb 28 '15 at 22:57
  • But it does matter because everything revolves around the Internet photo, not the physical dress. People on youtube call those who see the the same colors as the RGB values retarded. Well they would certainly have better carriers in photo editing than the non-"retarded" ones – Alex Mar 01 '15 at 00:14
  • 1
    @Alex No, because no one is saying "it's RGB value such-and-such," because humans do not naturally see RGB values. People are calling it "white" or "blue" or "gold" or "black," which is based on how their brain interprets the pixel colors. Your eyes interpret differing signals on rods and cones to try to assign a color to it (for instance, if you look at a monitor under bluish lighting, and the same monitor under yellowish lighting, you will see the same colors even though the actual light entering your eye is different -- you mentally filter out the difference). – cpast Mar 01 '15 at 21:19
  • @cpast: In fact, humans DO see RGB values. The 3 types of cones in the eye respond to red, green, and blue light. (Which is why we can make pretty accurate RGB images.) So we have (at least) 4 realities here: 1) actual dress; 2) RGB values captured by the camera; 3) color reproduced by your display; and 4) your personal color perception. I suggest that 2 is most important; if what I see as very light blue is in fact 0X59597B, then my hardware & eyes are accurately interpreting the image data. And we can then ask why other people see it differently. – jamesqf Mar 02 '15 at 22:28
  • 1
    @jamesqf, the [cones in the human eye](https://en.wikipedia.org/wiki/Color_vision#Cone_cells_in_the_human_eye) respond to (roughly) "a bell curve centered on yellow-green", "a bell curve centered on green", and "a triangle centered in the deep blue". RGB is a simplification of this system. – Mark Mar 02 '15 at 23:45
  • @Mark: Agreed. There's only so much detail that can be fitted in a comment, especially since I have to write without paragraphing. Still, I think I have a valid point: if the areas I perceive as light blue have RGB values corresponding to light blue, I'm in a very real sense perceiving the image 'correctly'. Those who perceive it differently are presumably experiencing some sort of optical illusion. Now we can simplify the question to "Why do some people see the optical illusion?" – jamesqf Mar 03 '15 at 03:56
  • 3
    @Alex People on YouTube call everyone retarded. They are not a reliable source. – OrangeDog Mar 03 '15 at 15:57