Do photographers see ambiguity in the color of the blue/black (gold/white) dress?

Question

Okay, so, this has taken the Internet by storm today... You've probably seen it and lots of commentary.

Apparently, many people see this as gold and white; to me, it's unambiguously blue. There are a number of articles (for example on Wired) explaining that this is an optical illusion and going into details about what most photographers already know well — the human vision system's mechanism for coping with changing light sources, and white balance and all that.

Try as I might to see it the other way, it just appears to be a blue dress, poorly photographed and with bad attention it the lighting. (And my perception happens to be correct; see this update on the original.) But many of my friends insist that it is either "clearly" white/gold, or at least ambiguous. And many of them are not... crazy people... and many are even artists, but none a serious/enthusiast/expert photographer.

So...

• Is it that my years of experience with digital photography and lighting have trained my brain to the point where I'm seeing it differently from the uninitiated? (See How to recognize different lighting color temperatures? — recognition of the color of light is certainly something that can be learned?)
• Or is it that many people have terribly calibrated monitors, compounding the problem? I know that most consumer monitors come with a very high default color temperature, blue-shifting everything, so I kind of suspect that it is at least a major factor. (Except, I showed my children on my system, and they see it as "white and kind of bronze".)
• Or is it really something that varies from person to person, with a background in photography not having anything to do with it?

I know this is an net meme thing, but I'm specifically interested in the photographer's perspective. I don't need a recap of the Wired article — I know all that. I want to know if it's still true for people with experience looking at photographs and lighting. The dress is blue, and I'm wondering if being used to thinking about the color of light (to the point where it's automatic) made it natural to see it correctly (and basically whether photographers are more likely than the general public to be among those who see it correctly).

---

Or, to come at this from another direction:

1. As a photographer, can you explain a plausible lighting situation where this could be a white dress? The only one that would make sense to me is if the dress were strongly lit by daylight or a daylight-equivalent source, and the background in tungsten and not lit by that same daylight. How could I take a white and gold dress and shoot it this way using standard interior lights (that is, no colored gels) and with global white balance as the only color-tweaking tool?

2. Could you recreate a different scene using either blue and black or gold and white and which would cause the same visual consternation? What elements would be necessary to do so?

If you are able to answer either of those questions, does the fact that you can answer meaningfully play into how you perceive the original?

Answer 1

My monitor is calibrated (less than a month ago).

I see the white/gold dress, but the highlights on the white piping have a blue tinge to me.

However I have seen pics of the (supposedly) original dress, and it is a deep blue and black.

To me, the only way I can reconcile this pic, and the pic of the actual dress is that if this pic was taken with a really bad white-balance and/or horribly overexposed. But that doesn't explain people who see the above pic and state "Blue/black"

---

I am almost of the opinion that this is an amazing marketing campaign. They put out this masterfully shot pic of the dress that was designed to go viral.

---

Update It is now several hours after I made my initial post. The sun has gone down where I am, and now I am relying on a mix of halogen and LED (daylight) lighting in my office, whereas previously I also had indirect sunlight through two windows.

I have now started to see a distinct (dark?) blue sheen in the OP's image - but nothing like the deep blue shown in my pic. However I still see gold, and I still perceive the color of my posted pic as the same.

So I believe that my ambient lighting is messing with my colour perception.

---

Pic of the dress in context:

Answer 2

To me the image appears white with a bluish tint (perhaps even a light baby blue) and the gold. or brown. It just won't read as black no matter how hard I try to convince myself. I think its the black object behind it that makes it never go there for me.

I can't reconcile the deeper blue of the actual dress with the slight blue cast in the image. It reads more as mixed lighting than anything although there are clues that it is not.

It will flip for my wife but not for me. I was wondering the same thing about experience with color being a factor but in the other direction...maybe we all see what we see and we just use our experience as justification.

Answer 3

Viewing these three images side by side from this article makes it fairly obvious what is going on with the viral photo: Choices about exposure and white balance determine how colors in a photo are perceived. Even black objects can be so overexposed as to over-saturate all three channels (RGB) and make black appear to be white. Amplifying the three color channels by different amounts can make any color appear to be near-saturated white.

Answer 4

As a photographer, I understand both what I see (blue) and the likelihood that others don't "see" exactly what I see, for any number of reasons -- especially if you allow for different photos of the same subject taken under different lighting conditions and/or different white balance settings.

If anything, I have a (completely unsubstantiated) belief that photographers and other "skilled viewers" probably have an unfair advantage in discerning the true colors in a photo like this because we've trained our brains to look for warm lighting, cool lighting, color casts, etc., and thus, I believe we're probably picking up visual cues in the photo that other people are just interpreting differently -- perhaps in the same way that a musician might pick instruments out of a recording with greater accuracy than someone without that sort of background.

Ultimately, I think this little viral exercise is a great reminder of how important interpretation is in the consumption of visual media, and that it's one part of photography we can never control completely.

Answer 5

The image has an obvious yellow colour cast. If i wanted to correct it, i'd put the eyedropper on the white flecks on the fabric in the lower left, which results in a blue/black dress.

If we wanted to pull the blue tinge to a shade of white, we'd have to increase the yellow, and the image would look completely unnatural and clipped.

So, no, there is no ambiguity for me, and my experience with recognizing white balance problems may have part in that.

Answer 6

Photographers are probably better trained to see colors. In this documentary:

http://www.dailymotion.com/video/xl7cgq_horizon-do-you-see-what-i-see-part-2-4_shortfilms

a completely colorblind photographer who can only see black and white explains how she can still perceive colors.

Also as pointed out in part 3:

http://www.dailymotion.com/video/xl7czm_horizon-do-you-see-what-i-see-part-3-4_shortfilms

the way the brain adjusts the white balance depends on the object we look at. So, under different lighting conditions, a yellow colored square will be perceived to change its color while a banana with exactly the same color does not change its color, because the brain knows it should look yellow. So, in this case, people may expect the dress to look white and that then makes the dress actually look white.

In part 4, they show how the perception of color is affected by language and culture:

http://www.dailymotion.com/video/xl7eh1_horizon-do-you-see-what-i-see-part-4-4_shortfilms

Answer 7

I see the colors as blue-gray with a bit of magenta and a sort of khaki.

If I should guess what is the original color of the dress, my answer would be that it is more likely gold-(off)white than black-blue due to the presence of darker color at the bottom left and some other clues.

Is it that my years of experience with digital photography and lighting have trained my brain to the point where I'm seeing it differently from the uninitiated?

Most likely. I think it is about the memory colors and the ability of human brain to take fairly bad image and make something familiar from it. We've seen women with and gold dresses before (some cultures probably more often than others), we've seen images where shadows were blue... An eye trained in color correction is less likely to get fooled.

I am wondering though what was the actual question asked - what is the color you see in the picture or what is the color of the dress?

Answer 8

I don't see any banana in the photo. With (1) a close crop so no expanse to judge color differences against, ans (2) no object with an overwhelming known color to calebrate against (the proverbial banana), how can you color-correct in your perception?

Perhaps some people see something in the photo that is well known to them.

In fact, I'm supposing that what little is visible in the background on the right is in different light. Hmm, maybe some people tigger off an indoor/outdoor mixed lighting situation.

---

Later: on a larger view, I notice the cow-spotted fabric on the left, in a narrow slice. The light areas seem over-yellow as opposed to ivoy color, perhaps because of experience with photos or how the saturation relates to intensity as some areas are reflecting more light. And the black looks chacteristicly like over-yellow, too.

So if that is lit with the subject (unlike the right side) I say "too yellow in the picture; make colors bluer". If yellow is white, then blue is really blue.

BTW, never saw it untill your post.

Answer 9

Do I see the ambiguity? I can understand it, having been exposed to many images where white objects are rendered with a blue cast, just like everyone else... but I can't honestly say that I can see the blue of the dress as being caused by that, even if I try. There are just too many other visual cues in the image that contradict that impression, I guess.

Coming back to this a day later, I would guess maybe one significant cue is that the temperature of the light in the background is still very warm, and not strongly corrected in a way that would be likely to render shaded whites a deep blue color?

Answer 10

The closest I can get to explaining what happened is to look at how a projector used for a presentation system handles black in projected images.

These projectors typically work by throwing light onto a white screen. Colors like red, blue, green, and everything in between are handled by filtering the light to project the desired color. To get black, however, you have to filter out everything. You effectively project nothing for that portion of the image, and you're doing it to a white screen... yet the result is perceived as black. This works because the white and colored parts nearby be so much brighter than the black parts that our eyes take cues from the contrast, and we know that the area of the image in question is black. But take away that context — walk up close the screen and look at just that point — and it again becomes apparent that the screen is actually white.

I think there is something similar going on here. If you look along the right side of the image you can see the bright light coming in from behind the dress. It giving our eyes cues about contrast that our brains aren't quite sure what to do with.

Answer 11

I honestly can't say if my training and experience as a photographer impacted my way of viewing the dress initially. I think it did, as I immediately saw accurate color (even down to tone levels) of the dress when I first saw it and tend to correct white balance in images in my head very well, but I can't tell for certain this has been impacted by having done fine tune color correction on tens of thousands of images (if not hundreds).

What I CAN say definitively is that my confirmation of my initial impression came down entirely to training and experience. Taking a closer look at the dress, my brain quickly registered the color mismatch between the slightly yellow and slightly blue lighting which is a typical white balance issue in photos shot under mixed lighting.

I then quickly noticed the black and white spotted dress on the rack behind. It extends far enough back to get both the yellow light and the light that balanced out as white in the image, thus it confirms the black and white nature of the dress and the position of the lights in the image.

From there, it is pretty trivial to work backwards to the lighting on the dress and confirm it as blue and black, just like the actual dress turned out to be.

Answer 12

Probably nothing. A large proportion of people wasting their time on this nonsense see black and blue. So I see no reason to assume that your ability to see the black and blue is down to skills, talent or training. There is no statistical evidence to suggest that your time as a photographer has anything to do with it. I don't see why a background of taking photos of things would retrain your visual cortex in any meaningful way.

However, what your background will imbue you with is the knowledge of the processes at work with the illusion which, if you'd instead said "I briefly saw white and gold but I knew intuitively that this was an illusion", might have been relevant here.

Answer 13

I will explain a little further theese questions:

Is the green I see the same as the green you see?

Eyes see things in diferent ways, it is not a mechanical or universal process. Our eyes recalibrate depending on lighting conditions. The white balance process ocurs in our retina all the time.

If you want a live experiment on how the retina change this calibration, cover one eye and cick on this link: http://www.otake.com.mx/Apuntes/ColorCalibration/Red.phtml

There are also several kinds of colour blindness for example.

Are the shadows I see at the end of the cavern the reality?

People see what they know, imagine, deduce from the education he or she has. The calibration I mentioned earlier also includes what the brain asume its true. You know a tungsten light is orange, but you don't think on that when you are reading in such a condition. As a photographer we are more aware of the huge diference in colour temperatures in diferent light conditions, but when you think as a photographer.

Do I need a machine to tell me what I need to see?

Do you want to read the rgb values of the raster file? Do you want all the people to use a hardware to calibrate their monitors? Should their monitors be of some characteristics? Or it is a psicological question... What do you see?

Do I want you to see the same as I am seeing?

I want to belive that that photo is taken in verey bad lighting conditions or camera quality.

There is chroma aberration, overexposed, the lens looks dirty or made of plastic, etc. I want to belive that the person was careless on white balance too. Do you want to see the case in different terms? Do you want to see the same as I see?

We will never know. Probably we can do a survey, that will say what the majority thinks. Beyond that we will never know.

In a bad photo like this... does it matter?

I'm imagining a case where the color of this image matter. "My girlfriend liked really much this dress and took a photo". That is it.

This would not be a product shoot, and if this were industrial espionage case, the colours would not matter.

---

From a commercial photography point of view, you need to have a colour calibrated process, a custom white balance, color calibrated camera, raw files, a colour reference photo on the light situation, good exposure, colour calibrated monitors, controlled ambient illumination on the retouching studio, a standardized press system... I would say to that photo as currently is... YOU SHALL NOT PASS!