Is That Dress White and Gold Or Blue and Black?

HughPickens.com writes Color scientists already have a word for it: Dressgate. Now the Washington Post reports that a puzzling thing happened on Thursday night consuming millions — perhaps tens of millions — across the planet and trending on Twitter ahead of even Jihadi John's identification. The problem was this: Roughly three-fourths of people swore that this dress was white and gold, according to BuzzFeed polling but everyone else said it's dress was blue. Others said the dress could actually change colors. So what's going on? According to the NYT our eyes are able to assign fixed colors to objects under widely different lighting conditions. This ability is called color constancy. But the photograph doesn't give many clues about the ambient light in the room. Is the background bright and the dress in shadow? Or is the whole room bright and all the colors are washed out? If you think the dress is in shadow, your brain may remove the blue cast and perceive the dress as being white and gold. If you think the dress is being washed out by bright light, your brain may perceive the dress as a darker blue and black.

According to Beau Lotto, the brain is doing something remarkable and that's why people are so fascinated by this dress. "It's entertaining two realities that are mutually exclusive. It's seeing one reality, but knowing there's another reality. So you're becoming an observer of yourself. You're having tremendous insight into what it is to be human. And that's the basis of imagination." As usual xkcd has the final word. It would make the comments more informatively scannable if you include your perceived color pair in the title of any comments below.

Comments

White balance and contrast in camera.

[Z00L00K]

It's primarily an effect of the camera fooling around with the white balance and contrast/light to try to get a correct image and failed.

Add to it that the picture looks different depending on which display you have on your computer and you have a nice debate.

Re:White balance and contrast in camera.

[jrumney]

Not depending on which display so much, but with LCD displays, depending more on what angle you are looking at. Look at it straight on, and the dress is white and gold. Ask the person next to you, and they will tell you it is blue and black. Turn your screen towards them and the effect will be reversed.

Re:White balance and contrast in camera.

[drinkypoo]

When I see stuff like that, my brain corrects the image for me.

Are people who are seeing this dress as weird colors just defective?

(I'm defective in other ways, like a lack of modesty, and some respiratory issues, let's not get all twisted here)

Re:White balance and contrast in camera.

[itzly]

That doesn't explain why two people looking at the same photo swear they see totally different colors.

Re: White balance and contrast in camera.

[Hardhead_7]

It's not just that. It flipped for me. I saw it Friday morning abd, saved the pic to my PC so I could open it up and use a color picker to see it was actually blue when it looked white/gold to me. Later that day, I was going to show it to a co-worker so I opened it up again. But now it was blue/black. It'd flipped twice more for me since then.

Curiously, the xkcd comic doesn't fool my eyes at all.

Re:White balance and contrast in camera.

[jrumney]

It depends

1. on the angle you are looking at the screen (on a phone, from the side, or a laptop tilting your screen back further, it definitely looks black and blue, straight on, see the other factors below).
2. whether you see the top of it first and scroll down to the rest of the image, or you see the bottom first and scroll up (I think because the top neck area is definitely gold, but the colors of the dress in the photo definitely have a blue tinge to them)
3. whether you perceive the light source behind as a window or a mirror (your brain adjusts the white balance of the dress accordingly)

Re:White balance and contrast in camera.

[drinkypoo]

The thing is, in the provided picture, the dress actually IS white and gold, or at least grey and gold. Load it up in an editor and snip pieces of it out if you don't believe me, look at them on their own, compare them to color swatches. That doesn't make the dress any particular color. It makes the picture a particular color. The "white"/"black" part is banging right around 50%, which is clearly neither white nor black.

The camera diddled the image, maybe it was diddled even more before we actually saw it. Then we're all amazed that it doesn't look like the thing. But people have accidentally been taking pictures of things which don't look like things since time immemorial. They're called shitty pictures.

Re:White balance and contrast in camera.

[jpapon]

From any angle, it looks blue to me. Very distinctly blue. I'm actually somewhat baffled that anyone (nevermind the majority of people) perceives it as white and gold.

Re:White balance and contrast in camera.

[jpapon]

The thing is, in the provided picture, the dress actually IS white and gold, or at least grey and gold

I'm sorry, but it is definitely not. I just opened it up in GIMP, and the blue areas have hue values between 225 and 230. While yes, the saturation is low (30-40), that definitely still makes it blue, albeit a washed out blue.

Load it up in an editor and snip pieces of it out if you don't believe me, look at them on their own, compare them to color swatches

I did. It's definitely blue. Not highly saturated blue, but blue nonetheless. It's certainly not white/grey.

Re:White balance and contrast in camera.

[markdavis]

>"Not depending on which display so much, but with LCD displays, depending more on what angle you are looking at. Look at it straight on, and the dress is white and gold"

Well, in my case, when I look at the photo in any light, on any monitor, at any angle, at any time, I have and have always seen only light blue and brown/gold. There is no situation where it is either "blue and black" or "white and gold".

The question is what we see in the photo, not what the dress ACTUALLY is- we can't know that because all we are allowed to see is a [poor] PHOTO of the dress, not the actual dress. And it is obvious the camera white balance and exposure is way off, trying to compensate for something, resulting in a photo with a probably very false representation.

Re:White balance and contrast in camera.

[wonkey_monkey]

The thing is, in the provided picture, the dress actually IS white [...] or at least grey [...]

What's your definition of "IS" here? Because the only objective measure I can think of is to look at the RGB values, and pretty much every pixel in the "blue" area has a B value about 20%-30% higher than the R or G value. That, to me, makes it about as objectively blue as it's possible to be.

Re:White balance and contrast in camera.

[caseih]

It's actually gold/black, white/blue. The real dress is dark black in the areas that people, including me, see as gold (and the rgb values are in fact gold), and the areas that people see as white (rgb values are highly saturated blue) are in fact dark blue.

Agreed that the picture is rubbish.

Re:White balance and contrast in camera.

[thegarbz]

No parent is right, it depends on which display. Only one LCD technology has the ability to screw up colours with viewing angles so spectacularly. On any IPS or PVA display the viewing angles won't mess up the colours. My girlfriend said blue and black from her laptop which has a TN display, then changed her mind when she saw it on the desktop with an IPS display.

Re:White balance and contrast in camera.

[jrumney]

The black is the bit that looks "gold". The "white" part is actually blue. Once I realised that the oversaturated light source behind was a reflection of the flash in a mirror (you can see more evidence of flash reflection on the dress itself), and not the sun shining through a window, my brain started processing the image totally differently, and I can no longer see it as white and gold unless I scroll the image so I can only see the top 20%.

Re: White balance and contrast in camera.

[Coisiche]

I'd also heard, that much like the rotating ballerina silhouette where you can change your perception of the direction of rotation, it is possible to change your perception of the colours in that photo. I couldn't do it though, stays white and gold.

Re:White balance and contrast in camera.

[BarbaraHudson]

Doesn't happen to me. It's always blue-purple and a muddy black. Even directly in front of a 26" screen centered at eye level. And even after blocking off the obviously over-exposed background. That's what you get when to take a picture with a crap camera in lousy lighting - something ambiguous.

Re: White balance and contrast in camera.

[jrumney]

Curiously, the xkcd comic doesn't fool my eyes at all.

Which makes me curious - which pair of colours are the two dresses in the xkcd comic to you?

Re:White balance and contrast in camera.

[beuges]

This morning, I saw it on my phone in my darkened bedroom, and it was clearly blue and brown. Just now, I opened the Washington Post link on my 24" screen in a sunlit room, and it was clearly white and gold. I then found the link that I had seen on my phone this morning (not Washington Post, so I wanted to confirm that it just wasn't two different pictures that I was looking at), opened it up, and it was white and gold there too. Went back to my bedroom and closed the curtains, and it remained white and gold for a bit, but after I left the room (after my eyes had adjusted a bit to the darkness), it was blue and brown again. The picture on the Washington Post was also now blue and brown. Now that my eyes have adjusted to the sunlit room again (and the white Slashdot background), I switch back to the Washington Post tab, and it's white and gold again. My wife (who's now gotten fed up with following me around to look at this picture under different lighting conditions) has had pretty much the same experience as me.

So it appears to be linked to the lighting conditions that your eyes are adjusted to when seeing the image initially... even after they've adjusted to the ambient light, the brain appears to stick to the image it created initially.

Re:White balance and contrast in camera.

[itzly]

I did. It's definitely blue. Not highly saturated blue, but blue nonetheless. It's certainly not white/grey.

Agreed, but white objects in the shade under a blue sky are usually blue. Look at the white snow under the trees, for instance. http://khongthe.com/wallpapers...

Re:White balance and contrast in camera.

[Anonymous Coward]

I saw it as hot pink and emerald green.

Unfortunately, I made the mistake of telling that to my psychiatrist, and that's why I'm typing this from a padded cell.

Re:White balance and contrast in camera.

[jrumney]

You actually need the overexposed background there to fool your brain into thinking the dress is underexposed to see the blue as white, and the black as gold.

Re:White balance and contrast in camera.

[dAzED1]

given that a color sensor on a digital camera is only sensing RGB (sometimes 1 or 2 more...) and that most monitors are only doing the same....just what would you expect white or grey to be, in the end? How would it be "captured" as an image if not by a sensor which can only detect R, G, or B?

Re:White balance and contrast in camera.

[Yomers]

Because what we perceive as the color of some surface is really a reflection of ambient light from that surface. So color of the surface would change depending on color of the light that it is reflecting. But our brain has auto white balance image preprocessing filter that fixes surface color for us based on light color , which is computed partially using our knowledge on what color things should normally have. In this picture dress could be white if it is not in a lighten by yellow colored light behind it, but lighten by blueish light source, and camera white balance is set somewhere in a middle. Otherwise it's blue.

Re:White balance and contrast in camera.

[drinkypoo]

pretty much every pixel in the "blue" area has a B value about 20%-30% higher than the R or G value. That, to me, makes it about as objectively blue as it's possible to be.

Are you kidding? It's possible to be balanced far more towards blue. But also, gold is not a color. I learned back in my Amiga-using, pixel-editing days that there's a lot of blue in most metallics. I don't know if that's an artifact of what happens to light when it bounces off of them, or what, sorry. Not a physicist. But I know that if you're trying to make something look metallic, you're going to be adding some blue to it.

Re:White balance and contrast in camera.

[dywolf]

i suspect a fair bit is related to the color temp setting of their monitors as well.

Re:White balance and contrast in camera.

[Applehu Akbar]

Looks absolutely, positively white/pale gold to me. And that was on three screens with a photographer's eye.

Re:White balance and contrast in camera.

[Fnord666]

So it appears to be linked to the lighting conditions that your eyes are adjusted to when seeing the image initially... even after they've adjusted to the ambient light, the brain appears to stick to the image it created initially.

Here is a pretty good explanation of why this might happen.

Re:White balance and contrast in camera.

[grnbrg]

Put it into to Photoshop and eye-dropper the colours. They are quantitatively light blue and dark brown.

But they can perceived as either blue and black or white and gold.

Re:White balance and contrast in camera.

[Jawnn]

No, you have an utterly mindless collection of noise, from a crowd of people who don't understand the rather simple science involved here. Is it any wonder that we have so many climate change deniers or anti-vaxers in the world?

Re: White balance and contrast in camera.

[Hardhead_7]

Blue and black. I don't know why it doesn't trick me, I usually fall for optical illusions. It's not like I have particularly good color perception or anything.

Re: White balance and contrast in camera.

[Hardhead_7]

I can't control it. It's "flipped" for me three times. I've tried to do different tricks like viewing angles or imagining it one way before I look, but nothing works.

Re:White balance and contrast in camera.

[wiredlogic]

It's definitely that. I first saw it on the news in a hotel lobby and it was clearly blue and black there. Now, viewing the original stills on my monitor it is blue-tinged white (bad white-balance) and gold. I can get it to shift toward darker blue and gold by changing my viewing angle to the LCD.

Re:White balance and contrast in camera.

[wonkey_monkey]

Are you kidding?

No, I'm not kidding, just trying to be objective about it.

It's possible to be balanced far more towards blue.

It's also possible to be get quite a bit closer to grey, so I think to claim that the dress "actually IS white/grey" is pushing it a bit. Compare the disputed "blue/white" parts of the image to an actual grey of the same intensity side-by-side and I don't there's any way someone with normal eyesight could fail to see the difference.

But also, gold is not a color.

I don't remember saying it was...

Re:White balance and contrast in camera.

[Oligonicella]

Yes, the photo was washed out and so looked different than the physical dress. Agreed.

Gold, however, is a color. Just like seafoam green contains blue, so does gold - at least, once you introduce lighting changes. That's how you shade a color, add or subtract amounts of each color (gold is FFD700, defined by HTML standards). It's just one of those many thousands of colors that aren't anywhere near primary or secondary. Brick red, tangerine and slate blue are all like that yet are still considered colors.

Re:White balance and contrast in camera.

[michelcolman]

The definitive clue is in the almost perfectly specular reflection on the draped bits near the shoulders: This is clearly a reflection of the illuminating light, and it's a bluish color in the picture. Pick that as the white point and you get a white dress with golden-brown applications.(...)Then you'll clearly see that it's a picture where the white balance is off (in terms that most people will find on their cameras, the camera balanced for "incandescent light" instead of "fluorescent light"), not a blue dress.

Great theory, except for the fact that the company that makes the dress has confirmed that it's blue and black.

When I first saw it, I was completely convinced it was blue and black. But then later that day, I saw it again and it was definitely white and gold. I couldn't believe I had ever seen it as blue and black. Ambient light and monitor color settings definitely change the perception. When you see the dress as white and gold, looking at the bright spot at the lower right of the picture makes the dress start to turn blue. Dimming the monitor also helps.

Re:White balance and contrast in camera.

[Lonewolf666]

Using a Samsung SyncMaster 204B here. In dim lighting conditions.

Looking at the picture from a "normal" angle (approximately from the front and with the screen slightly below eye level), the dress looks black and blue to me. But if I look down on the screen at an angle of maybe 70 degrees, the blue becomes sort of white and the black lace gets a bit of a golden hue. Still not clearly white and gold, but I see how one could get the idea.

Now the SyncMaster 204B is seven years old and probably not quite state of the art anymore. And it took me a rather unusual viewing angle to get anywhere near "white and gold". The people who saw the dress as white and gold must have pretty shitty screens ;)

Re:White balance and contrast in camera.

[dos1]

You're mistaken. The proper point to look at searching for white reference is in lower left part of the image - there's a part of some other, clearly black and white dress visible, and it has obvious orange - not blue - tint. Also, it's not hard to find other pictures of the same dress. There's also somewhere a confirmation of the taker of that photo saying that this dress is obviously blue and black. https://gs1.wac.edgecastcdn.ne... http://s3-ec.buzzfed.com/stati...

Re:White balance and contrast in camera.

[Bite The Pillow]

Look at a few pictures of white clothing under a black light, and then look at the dress again.

The blue color is very close to black lit white, and you may trick your brain into seeing it as such.

Re: White balance and contrast in camera.

[brianerst]

I can switch the perspectives if I scroll the picture. If the picture comes into frame from the top (shoulders), it's white/gold. If the picture comes into the frame from the bottom (hem), it's blue/black. The top is far more "golden" while the bottom is far more "blue".

Re:White balance and contrast in camera.

[Camel Pilot]

I am fairly convinced that those that claim the dress to be "distinctly blue" are just trolling the rest and extracting some weird sort of pleasure from it. The interesting aspect of this story is that there are sooo many people willing to troll others.

Re:White balance and contrast in camera.

[cerberusti]

My gf and I looked at it on the same screen.

My answer was periwinkle and brown for the image, but I thought the real dress was probably blue and black. She says white and gold.

I had to spend a moment looking at the background to determine that it was probably blue and black.

Re:White balance and contrast in camera.

[TheRaven64]

Zoom right in on the bits that you think are white, so that they fill your entire monitor. They're obviously blue. For a lot of us, that's the colour that we see when we look at it in context as well. I can see how you'd interpret it as being white by overcompensating for the colour in the bottom right, but that doesn't stop you from being wrong. The gold bits are gold when you zoom in (mostly, some are black), but a shiny black often looks yellow-gold in overexposed photos.

Re:White balance and contrast in camera.

[Bite The Pillow]

People seeing it for the first time, on the same device, in the same room for several hours, see different colors.

Your experience is not enough to draw any conclusion from.

Re:White balance and contrast in camera.

[EvilSS]

So it appears to be linked to the lighting conditions that your eyes are adjusted to when seeing the image initially... even after they've adjusted to the ambient light, the brain appears to stick to the image it created initially.

Here is a pretty good explanation of why this might happen.

Something is wrong. You said "pretty good explanation" but you then linked to Gizmodo. These two things are mutually exclusive.

Re:White balance and contrast in camera.

[Applehu Akbar]

In fact I did try to Save As the jpg given in the Tumblr reference, but it was in some weird-ass DRMed format that made it unviewable even in Photoshop.

Re:White balance and contrast in camera.

[Baloroth]

Yeah, and that's kinda the whole point: everyone who looks at the photo is automatically (and completely subconsciously, without realizing it) applying color-correction to the dress, based on the brain's similar experiences with color-correcting and the visual clues in the picture. What makes the picture interesting is that it's so close to the edge between white/gold and blue/black that different people can perceive it differently, even on the exact same screen. Actually, I've seen it both ways, though I believe the picture that I saw as white/gold was ever so slightly lightened (based on a totally not scientific color picking of the blue areas). The picture was also a smaller version, which may have made the difference. The point is, the picture is a fascinating example of how what humans really perceive is not what they're actually seeing, but a heavily interpreted version based on context and various visual clues. In fact, humans would be effectively blind without that processing (imagine face blindness, but for everything you see).

Re:White balance and contrast in camera.

[serviscope_minor]

Well, the best explanation is that our brain does very, very hefty colour correction on its own. The reason is probably so we can ignore the irrelevant information, i.e. the light colour and any sensitivity lag between the different cones and identify objects.

If you've ever tried to do colour computer vision you will be *astonished* at how good we are at it.

Anyway, my guess is that the reason for this one appearing as two different things is it's very close to some threshold. It's possible to interpret it as more or less normal colours under a blueish light (i.e, sunlight), which makes you perceive the blueish grey as white. Or, you perceive it as blue under washed out much more orange light (e.g. incandescent bulbs) in which case you perceive the blueish grey as blue. The "white/blue" colour is objectively blueish grey if you check the actual hex values: it's mostly grey but the blue channel is a bit higher, but not by that much.

Our brains seem to model the complete co varying of the colours so once you decide what the light colour is, this effects the perception of all the colours. Black is nevertruly black and always reflects some light, so once the brain has decided the blue is super washed out by excessive light and that light is orange, that makes bright orangish things perceived as black since that's how black would appear under those circumstances. Given the rather good corrections of the picture, it appears those people are correct (though I'm on the white/gold side).

These two perceptions are very, very different and so it's kind of a binary thing, either you perceive one or the other. Now, I suspect that this photo is very close to the tipping point which is why some people perceive it very differently from others.

Either way, it's a fascinating example of colour interpretation.

This one is also excellent:

http://en.wikipedia.org/wiki/F...

Re:White balance and contrast in camera.

[The Rizz]

I can only see it as blue, period. Not trolling - I really cannot see this as white in any circumstances, even the XKCD "color balanced" bit I still see it as blue (albeit a much lighter blue on the left).

Maybe this has more to do with some effect similar to color-blindness? There are more types of color-blindness than just the standard red-green, and they have different effects on how colors are seen (and some kinds of color blindness can actually make you see other colors better). Maybe this dress has highlighted a previously unknown type of color-blindness?

Re:White balance and contrast in camera.

[drinkypoo]

(gold is FFD700, defined by HTML standards).

HA HA HA! YOU FELL INTO MY TRAP!

Pardon my caps, but I set a lot of these, and they are rarely stepped in so beautifully. That's my fault, of course. But, here we go.

When you hear the name of a color, you think of some color which you associate with that, or a thing which you think of as being that color which actually has a whole texture, reflectivity, depth, etc. But the truth is that the same name is being used right now by a dozen different paint companies to describe a dozen different colors, which are then described in thousands of different ways by the downstream users of the paint. There are some absolute color standards based on elements, oxides of the same and the like, but even those are frequently "abused". The truth, though, is that outside of a small handful of colors, they are not defined by any unifying principle. Your use of the HTML standards is particularly hilariously perfect, I really hope that you meant to put your foot there! Outside of web design, nobody but nobody gives a crap about that. Before that, we had the X rgb.txt file, which nobody outside of X-land cared about, which begat HTML colors by the way, junior. But meanwhile, over in the land of professional color, there were multiple competing color standards including AGFA and Pantone — and there still are.

It's interesting that part of this debate is also over the color "black" because "black" is what you see when you don't see anything, and if any part of the dress were truly black then the photograph would look like an editing mistake. Even the color black is subjective. That should be intuitively obvious to a web designer (who else would even mention HTML colors?) who has to deal with the real-world effects of differing black and white levels all day. I may be sitting at a monitor with 120% color (Adobe gamma) but I don't expect other people to have them. As far as I know, there is no color correction e.g. for Android which is not manual, and then the color adjustments become utterly perceptual.

I suspect that celebs are adept at perceiving the actual color of the dress because they have appeared in so many washed-out photographs.

Re:White balance and contrast in camera.

[pr0fessor]

That's what I saw when I opened it in gimp but imagine it is still white and gold and that the photo appears blue because of reflected light. Which is one of those things that happen with some materials used in womens clothing. It is often an intentional effect like a white cami with a little sheen that will reflect the colors of the other garments they are wearing. My wife has a few of those A.K.A goes with just about anything.

Re:White balance and contrast in camera.

[Ogive17]

Whew, I thought I was the odd ball. That is exactly how I would describe what I see.

Re:White balance and contrast in camera.

[Wycliffe]

So it appears to be linked to the lighting conditions that your eyes are adjusted to when seeing the image initially... even after they've adjusted to the ambient light, the brain appears to stick to the image it created initially.

This isn't what happened to me. The first time I saw it (before I realized the controversy), it was clearly
white/gold but since then no matter how I look at it, it is clearly black/blue and I can't make it change
back to the white/gold. I tried finding the original picture I looked at which was white/gold but everything
in my history was now blue/black.

Re:White balance and contrast in camera.

[serviscope_minor]

Well, I'd hesitate to call you colour blind since you are in fact correct. The dress really is blue and your brain is somehow undoing the mangling that's been done by the camera and lighting to arrive at the correct colour. I can't unsee it as white and gold however.

You can think of it all hinging on the blue/white stripes.

Objectively they're grey with a little blue in (check with a colour picker). All colour interpretation is ambiguous since you have do undo the effect of lighting and uneven colour sensors (eyes included). If your brain decides the light is white with a bit of blue, then that implies the underlying colour must be white and so the other colours must be gold.

If however your brain makes the call that it's strong orange light then that implies the stripes are in fact blue. That further implies the other stripes are dark grey, usually intrepreped as black.

There are other optical illusions designed to trigger this different intrepretation such as this:

http://en.wikipedia.org/wiki/F...

What's unique about this is it's not made to trigger two different intrepretations in one person, but instead by chance triggeres quite evenly the different intrepretations in different people, leading to big debates.

For you, the XKCD one isn't extreme enough to push you in different directions. Interestingly high level intrepretation has a big part. I can't cite anything but I can give a personal anecdote. The other day I saw an odd green thing on the floor in my hotel room. It was actually my backpack lit with green tinged light, but was crumpled in an odd shape so I couldn't tell what it was. When I figured it was my backpack, I could no longer see it as green (the actual colour is black). I probably spent 5 minutes staring at it trying to re-see the green with no success. It's a good illustration that the brain uses information from all levels including high level information such as recognising a specific object that I know is in the room to intrepret light colour.

Re:White balance and contrast in camera.

[beelsebob]

Nope, just different lighting conditions to you. The effect can be reproduced on absolutely perfect IPS displays, if you set up the lighting correctly.

Re:White balance and contrast in camera.

[The Rizz]

Well, I'd hesitate to call you colour blind since you are in fact correct. The dress really is blue and your brain is somehow undoing the mangling that's been done by the camera and lighting to arrive at the correct colour. I can't unsee it as white and gold however.

I'd actually think it was the other way around - I'm less blue colorblind which is why I always see blue, regardless of surrounding context. You've got slight blue colorblindness which is why context causes you to interpret the blue as a different color.

The other day I saw an odd green thing on the floor in my hotel room. It was actually my backpack lit with green tinged light, but was crumpled in an odd shape so I couldn't tell what it was. When I figured it was my backpack, I could no longer see it as green (the actual colour is black).

This could indicate another possible explanation; that in many people the color you first see it as becomes the "right" color, and it becomes difficult/impossible to see it otherwise, as your brain has already interpreted it, and is remembering the "right" color even if you're trying to see it "wrong".

Re:White balance and contrast in camera.

[rastos1]

Put it into to Photoshop and eye-dropper the colours.

Excuse me. Is GIMP good enough for that functionality?

--
On the internet nobody can hear you being subtle

Re:White balance and contrast in camera.

[Solandri]

I've been doing white balancing and level adjustments on my photos for over 20 years now. The dress as shown in the photo is white/gold (actually more of a light-blue/gold). You can confirm this with eyedropper measurements in Photoshop.

However, if you look at the sliver of background which appears to the right, you can tell the photo is badly overexposed. If your eye spotted this and your brain compensated for it by interpreting the pic as what might see if you stepped out into bright sunlight after being in a darkened room, the dress will appear blue/black.

Re:White balance and contrast in camera.

[fatwilbur]

That's it, no more using words to describe the colors a person sees, only wavelength numbers allowed.

I don't know about all you other crazy people out there, but I clearly see 582 nanometers.

Re: White balance and contrast in camera.

[Cochonou]

Yes, it's about the same for me. I see it predominantly gold and white. However, I can see it black and blue if I first look at this white and gold dress, and then I scroll the image from the bottom. I guess it's because most of the orange illumination "hints" are in the lower part of the image.

Re:White balance and contrast in camera.

[Solandri]

This dress thing is actually a simplified version of a really powerful and freaky illusion which unfortunately only works in a darkened room. You take a quilt with different color squares. You can verify in white light that the different colors span the rainbow.

On top of this quilt, you lay a black piece of cardboard which has a cutout allowing you to only see one color patch. Say it's a greenish patch. In the darkened room with the single white light, you can confirm that this single visible patch is green.

Then you put colored filters in front of the light until the patch appears to be red/orange. The light is red/orange, so the green patch now appears red/orange. Makes sense right?

Then you remove the black cardboard so you can see all the color patches. And suddenly that red/orange patch appears green again. Cover it up with the black cardboard again and it'll appear red/orange. Remove the cardboard and it appears green.

What's going on (and also what's going on in the dress photo) is that your brain is using the information of surrounding objects (in the dress pic, it's the background visible to the right) to gauge the amount of light and the color of the light. It then corrects for the color of the light to white balance what you see, and for the brightness to compensate for over/under exposure. So the colors and intensity your eyes see remain the same, just how your brain interprets them changes.

Re:White balance and contrast in camera.

[Anubis IV]

This is one of the key aspects of the case. The color of the pixels in the image is clearly blue. That's irrefutable. We can do a quick color analysis to figure that out, as you did.

But, unfortunately, that doesn't answer the question of what color the dress itself is. Just because the picture shows it as blue doesn't actually mean it is blue, and there are numerous illusions, color correction issues, or optical afterimage effects that could cause something that actually is color A to appear as color B, either to the camera or to us.

I'm firmly in the "the dress appears blue and brown/black but is actually white and gold" camp. There are numerous highlights in the image that cause the true color to pop out from under the blueish pall that is over the whole thing. You can see a more true gold up on the right side of the collar and along the top edges of some of the ridges. Likewise, you can see a truer white in various places where the light hits it more directly. Whether it looks blue and brown because there's something blue behind the camera that is reflecting a blue pall over the dress, or it looks blue and brown because of the white balance being out of whack on account of the massive backlighting going on, I don't know, but there's plenty of evidence in the image to suggest the dress itself is white and gold, even if it doesn't immediately appear that way in the image.

Re:White balance and contrast in camera.

[q4Fry]

My wife asked me, and my answer was "black and .... blue, I guess?" I thought it was a trick question what shade of blue one would call it. Since hearing the point of the question, I can reliably identify the "black" as "muddy bronze," but the only way to trick my brain into thinking the dress is white (still with a blue cast) and gold is to take my wife's monitor and look at it from a 45 deg angle from the top. My monitor still looks blue from any angle.

Re:White balance and contrast in camera.

[hey!]

I've sat right next to people who see the dress differently than me. It's *the same image* on *the same monitor* at *the same time*. So it's not a case of the monitor calibration or the camera white balance that creates the discrepancy, although obviously manipulating those things will change our individual perceptions of the dress. What's interesting here is the differences between people presented with an identical image.

Color doesn't exist in the external world. "Purple" isn't a wavelength of light, it's a kind of "additional data" tag which our brains add to parts of an image that allows us to extract more information from it. Consider the famous "Rubik's Cube" optical illusion where the same square looks either orange or brown based on whether contextual cues make us think it is in shadow or not. There's an illustration here.

The only difference between the Rubik's Cube illusion and The Dress That Broke The Internet is that practically *everyone* experiences the paradoxical sensations of the Rubik's Cube Illusion; in the case of the dress the paradox is in how sensations *differ between people*. The dress image is a kind of borderline case where our brains can "tag" the "pixels" of the image in one of two possible ways depending on what it thinks the context is. Different brains are trained by different experiences to expect different contexts. If we saw the dress being worn and in person, chances are with all that context there'd be less disagreement.

Re:White balance and contrast in camera.

[Anonymous Coward]

I've seen it on a number of different displays under different conditions, not because I'm particularly interested in the stupid dress, but because a number of different people wanted my opinion. Well, it was that it's blue and black, and that looking at the background it's probably even bluer and blacker in real life. They agreed but told me everyone on the internet told them otherwise. I told them the internet is full of trolls.
Anyway, I sort of hoped the thing would have blown over by now, but since it's on Slashdot, especially for you I looked for a better picture of the dress and found one: http://www.bbc.com/news/uk-scotland-highlands-islands-31656935 End of discussion.

Re:White balance and contrast in camera.

[loufoque]

The color of the pixels is white and gold. The white is blue-ish of course, but the people who see the dress as white and gold can clearly see that.
People who see it as white and gold see the real colors of the pixels. People who see it as blue and black have their brain doing color compensation automatically.

I'd say the normal people are the one who see it white and gold...

Re:White balance and contrast in camera.

[brausch]

Basically, me too. I've seen it both ways. I actually saved the file on my desktop so I can be sure I'm looking at the same thing each time and it has flipped. I mostly see blue and black/brown, sometimes light blue and light brown, and rarely the pure white and gold. Pretty weird, but also pretty cool.

Re:White balance and contrast in camera.

[Goragoth]

I'm still befuddled by what the deal with this picture is. All I can see are the colours that you can confirm by going into any image editing program and checking the pixels: i.e. a light blue and a golden/brownish colour. Now some people are saying you should make some sort of judgement as to what colour the dress really is, when compensating for the bad exposure, to which I can just say: I don't know. There isn't actually enough information in the picture to make that call. I wonder why this isn't a more common answer; when you can't be sure of something "I don't know" is perfectly acceptable as an answer.

Re:White balance and contrast in camera.

[TMB]

You know it really is blue, right?

(the reviews, incidentally, are hysterical)

[TMB]

Re: White balance and contrast in camera.

[shahlak]

It cannot be the display because I and my husband saw the pic on my mobile screen, his mobile screen as well as our laptop. No matter where, at what angle, light or time we see it, it always appears golden brown & white with blue tinge to me. My husband sees blue & black.

Re:White balance and contrast in camera.

[neurocutie]

"Put it into to Photoshop and eye-dropper the colours. They are quantitatively light blue and dark brown." NO. As several of the neuroscientists interviewed have tried to explain, there is NO such thing as "quantitative color"". Color is a PERCEPTUAL phenomenon that is INTERNAL to the individual perceiver (human). Physics has nothing to say about color, and color is not a function of any simple physical or quantitative system. Color is certainly tied to wavelength or any other similar quantification of physical properties. You can't just make measurements using Photoshop and determine color (nevermind the FACT that there is no dependable mapping between RGB GUN values and actual photon/physical output of the display device you are using to look at the image).

Re:White balance and contrast in camera.

[kenshin33]

20 minutes ago I was white and gold (pic in 1st link). Now,it's blue and black.
My GF was next to me, and saw the opposite of what I saw.

Re:White balance and contrast in camera.

[markdavis]

WTF? You are the one insane...

Re:White balance and contrast in camera.

[Baloroth]

Everyone who looks at the photo will see the dress as *some* color. That's the point. It's not a conscious judgment, you see what your eyes/brain are telling you is there. Some people see it as white, some people see it as blue (and some people have seen it as both).

There isn't actually enough information in the picture to make that call.

But your brain does it anyways, because that's what your brain does.

Re:White balance and contrast in camera.

[SeaFox]

The camera diddled the image, maybe it was diddled even more before we actually saw it. Then we're all amazed that it doesn't look like the thing. But people have accidentally been taking pictures of things which don't look like things since time immemorial. They're called shitty pictures.

This is what happens when publications fire their professional photographers and give their "reporters" (who skill-wise are glorified bloggers now) smartphones and tell them to take their own pictures, because that's much cheaper than hiring real pros. The sad part is they probably think this is all great. After all, they're getting a lot more page views over this stupid photo controversy than a properly taken picture of the dress would have attracted to their site.

Re:White balance and contrast in camera.

[BarbaraHudson]

Maybe the ones who see the dress white and gold are the same ones who's optic responses have been damaged by too much J. J. Abrams lens flare.

Re:White balance and contrast in camera.

[hucker75]

Stick it in a photo editor and ask it what colour it is. It's white (with a faint blue tint) and pale gold. If the dress is blue and black, the photo is wrong.

Re:White balance and contrast in camera.

[gzuckier]

Not depending on which display so much, but with LCD displays, depending more on what angle you are looking at. Look at it straight on, and the dress is white and gold. Ask the person next to you, and they will tell you it is blue and black. Turn your screen towards them and the effect will be reversed.

I had it absolutely perfectly, laptop display, partially dimmed room lights, so that when you scrolled down so that only the last couple of bands on the dress were visible it was obviously black and blue, but as you scrolled up it became blue and gold. Haven't been able to reproduce the effect since, but it is possible! Don't stop trying!

Re:White balance and contrast in camera.

[gzuckier]

This dress thing is actually a simplified version of a really powerful and freaky illusion which unfortunately only works in a darkened room. You take a quilt with different color squares. You can verify in white light that the different colors span the rainbow. On top of this quilt, you lay a black piece of cardboard which has a cutout allowing you to only see one color patch. Say it's a greenish patch. In the darkened room with the single white light, you can confirm that this single visible patch is green. Then you put colored filters in front of the light until the patch appears to be red/orange. The light is red/orange, so the green patch now appears red/orange. Makes sense right? Then you remove the black cardboard so you can see all the color patches. And suddenly that red/orange patch appears green again. Cover it up with the black cardboard again and it'll appear red/orange. Remove the cardboard and it appears green. What's going on (and also what's going on in the dress photo) is that your brain is using the information of surrounding objects (in the dress pic, it's the background visible to the right) to gauge the amount of light and the color of the light. It then corrects for the color of the light to white balance what you see, and for the brightness to compensate for over/under exposure. So the colors and intensity your eyes see remain the same, just how your brain interprets them changes.

Edwin Land (of Polaroid fame) did a lot of research on this. The best was his demonstration that you could take a set of like 5 blocks of colors like 10 nm apart arranged in ascending or descending wavelength and get an obvious set of slightly different colors going from yellow to greenish yellow. But if you then made a matrix of like 100 blocks with each chosen randomly from one of those yellow blocks, it demonstrated the entire spectrum.

Re:White balance and contrast in camera.

[gzuckier]

>"Not depending on which display so much, but with LCD displays, depending more on what angle you are looking at. Look at it straight on, and the dress is white and gold"

Well, in my case, when I look at the photo in any light, on any monitor, at any angle, at any time, I have and have always seen only light blue and brown/gold. There is no situation where it is either "blue and black" or "white and gold".

The question is what we see in the photo, not what the dress ACTUALLY is- we can't know that because all we are allowed to see is a [poor] PHOTO of the dress, not the actual dress. And it is obvious the camera white balance and exposure is way off, trying to compensate for something, resulting in a photo with a probably very false representation.

In my "research" (i.e. I had a lucky configuration looking at the pic) described further up the comments, the black/blue impression of the dress comes when isolating just the lower few bands of the dress, the blue/gold comes from the upper region. Note that the background light at the bottom is goldish, the background light higher up is bluish. (OK, baby Boomers all together now, "That's funny, you don't look Bluish").

Re:White balance and contrast in camera.

[gzuckier]

Because what we perceive as the color of some surface is really a reflection of ambient light from that surface. So color of the surface would change depending on color of the light that it is reflecting. But our brain has auto white balance image preprocessing filter that fixes surface color for us based on light color , which is computed partially using our knowledge on what color things should normally have. In this picture dress could be white if it is not in a lighten by yellow colored light behind it, but lighten by blueish light source, and camera white balance is set somewhere in a middle. Otherwise it's blue.

https://en.wikipedia.org/wiki/...

Re:White balance and contrast in camera.

[gzuckier]

Well, the best explanation is that our brain does very, very hefty colour correction on its own. The reason is probably so we can ignore the irrelevant information, i.e. the light colour and any sensitivity lag between the different cones and identify objects.

If you've ever tried to do colour computer vision you will be *astonished* at how good we are at it.

Anyway, my guess is that the reason for this one appearing as two different things is it's very close to some threshold. It's possible to interpret it as more or less normal colours under a blueish light (i.e, sunlight), which makes you perceive the blueish grey as white. Or, you perceive it as blue under washed out much more orange light (e.g. incandescent bulbs) in which case you perceive the blueish grey as blue. The "white/blue" colour is objectively blueish grey if you check the actual hex values: it's mostly grey but the blue channel is a bit higher, but not by that much.

Our brains seem to model the complete co varying of the colours so once you decide what the light colour is, this effects the perception of all the colours. Black is nevertruly black and always reflects some light, so once the brain has decided the blue is super washed out by excessive light and that light is orange, that makes bright orangish things perceived as black since that's how black would appear under those circumstances. Given the rather good corrections of the picture, it appears those people are correct (though I'm on the white/gold side).

These two perceptions are very, very different and so it's kind of a binary thing, either you perceive one or the other. Now, I suspect that this photo is very close to the tipping point which is why some people perceive it very differently from others.

Either way, it's a fascinating example of colour interpretation.

This one is also excellent:

http://en.wikipedia.org/wiki/F...

http://www.michaelbach.de/ot/c...

Re:White balance and contrast in camera.

[jrumney]

I've noticed this too. Many of the photos being spammed on my Facebook feed are cropped showing only the top 25% of the photo. In those photos, the light blue/gold shows up (with the light blue looking like bad white-balancing, so could easily be seen as white). But whenever the full picture is visible, the blue/black is more apparent, and when the bottom half of the dress is all that is visible, it definitely looks blue/black. Some of the people that have responded here saying they never see white/gold also stated that they have large monitors, so maybe they have never seen the image cropped, as it first appears on my relatively small laptop display on the buzzfeed site and others where I have to scroll down to see the full image.

Re:White balance and contrast in camera.

[RespekMyAthorati]

Oh, so RGB={0,0,255} isn't "quantitatively blue", huh?
Horseshit. If there is no red, no green, and %100 blue, then the color is quantitatively blue.

Re:White balance and contrast in camera.

[RespekMyAthorati]

Or, you could just use the eyedropper tool in photoshop, which clearly indicates that the colors are light blue and medium brown.
If these aren't the actual colors of the dress, then the picture is improperly exposed.

Re:White balance and contrast in camera.

[stealth_finger]

That's what me and my brother thought, dependant on a number of factors. But when it was on the news we were both seeing the different colours, regardless of viewing angle or distance or any other factors. The effect was even carried through to a photo on my phone of the paused tv screen. It's the most bizarre thing.

Re:White balance and contrast in camera.

[KozmoStevnNaut]

No, the highlights are simply blowout because the picture was taken on a phone camera. Give me a bright enough light, a piece of dark fabric and a phone camera, and I can easily make white or near-white highlights appear on a picture of said dark fabric.

Brown highlights are common on "black" fabrics, since they are often extremely dark brown (or blue), not true black. So the dark parts are definitely black.

If the blue parts of the dress were actually white, that would require a light source with a lot of blue light, as well as a camera with very misconfigured white balance. You would have to deliberately do this to mislead. It's also impossible to do this on an iPhone (which the picture was taken with).

Lastly, we know for a fact that the dress is blue and black, because the manufacturer doesn't even sell a white and gold version.

Re:White balance and contrast in camera.

[jpapon]

White/grey have nearly equal levels of R,G,B. This color clearly has a higher intensity in the blue channel.

Re:White balance and contrast in camera.

[jez9999]

I see it as white and gold, every time. I just about get how the dark blue could change to light blue, and my brain perceives it as white given the dark contrast with the stripes. What I don't get is how black can turn to brown or gold or something. Surely lighter black should just be grey.

Re:White balance and contrast in camera.

[jez9999]

I'd actually think it was the other way around - I'm less blue colorblind which is why I always see blue, regardless of surrounding context. You've got slight blue colorblindness which is why context causes you to interpret the blue as a different color.

This is probably the best explanation I've seen. I *cannot* see the dress as anything but white and gold, no matter what lighting conditions, etc. I'm not totally colourblind to blue, and when I grab some of the "white" pixels in GIMP and blow them up they look light blue; but in context they look white (maybe a "dull" white).

My mum, however, usually sees it as blue and black. She often sees things as dark blue that I would call "grey". So maybe my eyes just have fewer blue cones.

Re:White balance and contrast in camera.

[jez9999]

But they can perceived as either blue and black or white and gold.

Can someone explain to me how a black garment could change to a gold colour instead of grey? It would virtually take a red/orange streetlight to achieve that effect.

Re:White balance and contrast in camera.

[tehcyder]

People seeing it for the first time, on the same device, in the same room for several hours, see different colors.

Your experience is not enough to draw any conclusion from.

If you don't see it as blue and black, you are clearly some sort of alien lizard-being.

Worryingly, that means that only about a quarter of the people around me are actually human.

Re:White balance and contrast in camera.

[tehcyder]

I'm firmly in the "the dress appears blue and brown/black but is actually white and gold" camp.

I bet you and the rest of the white/gold fascists are feeling pretty stupid now you've seen a picture of the (undeniably) blue and black dress?

I knew I was right, I just fucking knew it. Finally, a victory for humanity against the powers of darkness!

Re:White balance and contrast in camera.

[KozmoStevnNaut]

No, it's distinctly a grayish blue. The photo is perfectly fine showing which colors the dress is, it's just a bit overexposed so the colors doesn't show through as strongly as the official model photo on Amazon.

And shiny black often appears golden or brownish under strong direct light which is actually a really big clue. Since the "gold" parts fade way too much to black in the dark parts, it's obviously now brown/gold, but black with a strong light shining on it. And no light source can simultaneously put warm highlights on black, while at the same time casting a cool blue tint on white, especially not one as strong as in the picture. Not in this universe, at least.

Re: White balance and contrast in camera.

[KozmoStevnNaut]

But how can those factors simultaneously cause a warm highlight on black and a cool blue tint on white?

The answer is that they cannot do that.

Re:White balance and contrast in camera.

[tehcyder]

I'd say the normal people are the one who see it white and gold...

Literally Hitler.

Re:White balance and contrast in camera.

[tehcyder]

I am fairly convinced that those that claim the dress to be "distinctly blue" are just trolling the rest and extracting some weird sort of pleasure from it. The interesting aspect of this story is that there are sooo many people willing to troll others.

That would be a strong possibility, except for the inconvenient fact that the dress IS blue.

Us "blueys" are like Neo in The Matrix, piercing through the deliberate veils of convention to see the truth.

Re:White balance and contrast in camera.

[tehcyder]

Stick it in a photo editor and ask it what colour it is. It's white (with a faint blue tint) and pale gold. If the dress is blue and black, the photo is wrong.

You'd better tell the manufacturers about your interesting discovery, since from their point they only sell a black and navy blue version. But what would they know?

Re:White balance and contrast in camera.

[tehcyder]

Surely lighter black should just be grey.

What we see as black is often very dark blue or green or brown in reality.

Re:White balance and contrast in camera.

[tehcyder]

when you can't be sure of something "I don't know" is perfectly acceptable as an answer.

No one's asking you to stake your mother's life on it, it's just your impression of a dress colour.

If I say "that bird looks black" no one's going to arrest me if it turns out to be dark brown with some dark blue flashes and a white tail feather.

Re:White balance and contrast in camera.

[tehcyder]

gold is not a color

Gold is just shiny yellow, like silver is shiny grey.

And irony is colour blind people getting the real colours right more than those with normal vision.

Re:White balance and contrast in camera.

[tehcyder]

That doesn't explain why two people looking at the same photo swear they see totally different colors.

Because one is a human being, and the other is a man-shaped robot from the future posing as a human as he plots to destroy the one man who can save the human race from enslavement?

Re:White balance and contrast in camera.

[jrumney]

The dress is overexposed which moves the black to grey, and the blue has been desaturated, which leaves mostly yellowish hues like brown and gold behind.

Re:White balance and contrast in camera.

[Capt.Albatross]

Put it into to Photoshop and eye-dropper the colours. They are quantitatively light blue and dark brown.

But they can perceived as either blue and black or white and gold.

Thank you. I was prepared to believe that the dress was blue and black, but not that the image of it contained black (FWIW, I have always seen it as unambiguously blue/brown from the start, before I knew there was a question.)

The explanation I read (New Scientist) completely overlooked the significant issue of accuracy in color reproduction, both in the camera and in the display.
 

Re:White balance and contrast in camera.

[Capt.Albatross]

Put it into to Photoshop and eye-dropper the colours. They are quantitatively light blue and dark brown.

But they can perceived as either blue and black or white and gold.

You're assuming proper white balance at the time the photo was shot for the lighting conditions of the environment in which the photo was shot for your "quantitatively" assertion. Unless you were there or took the photo, you cannot say with any certainty.

No, gmbrg is making a statement about the colors recorded in the image file, not about the colors of the dress. The values in the image file are precise quantities.

Re:White balance and contrast in camera.

[hucker75]

Greyish blue and blueish white are the same thing. The photo does not show the colours at all, it's washed out, you have lost the colours. In real life, your eyes know what the surrounding light is and subtract that from the dress, so you can tell the colour. It's called "white balance" and your brain is very good at it. But we have no idea what the light source is in that room.

Re:White balance and contrast in camera.

[hucker75]

What have the manufacturers of the dress to do with anything? It's the CAMERA that got the colours wrong, and it's common knowledge that cameras are not as good as humans at white balance.

Re:White balance and contrast in camera.

[neurocutie]

"Oh, so RGB={0,0,255} isn't "quantitatively blue", huh? Horseshit. If there is no red, no green, and %100 blue, then the color is quantitatively blue."

You've missed the point... the question is NOT "Give me an example of something that is liikely to be blue". Sure RGB=[0,0,255] on a standard commercial display device, viewing with a wide range of background conditions and lighting is very likely to appear blue to a normal observer.

But that is not real life and RGB only exists in the confines of a display device, not the real world where can have almost limitless levels of light energies of all different wavelengths and spectral signatures. Even in RGB space, could you reliably tell me whether a pixel in a visual scene that has values [100,100,120] will appear to be blue or bluish? NO YOU CANNOT.

The question of "is there a quantitative BLUE" is instead the question of "Is there a quantitative description of what about the physics of a visual scene (or light stimulus) that will ALWAYS appear to be BLUE, and DEFINES what blue is"... there IS NOT, certainly not a simple one, though we can come up with approximations based on physiology and psychphysics.

Re:White balance and contrast in camera.

[torkus]

It's similar to the trick where you stare at a B&W pattern on a page then look at the wall to see the inverse of it.

Of if you wear tinted sunglasses the world looks (insert color) at first then a while later you don't notice. When you take the glasses off things look...well brighter...but the colors are a bit odd for a few minutes there too.

Or the bunch of other optical illusions where you surround a color with others and it looks different.

It's not some great big mystery for anyone who works with photography (or anything similar that deals with lighting and color). That's why monitors are calibrated to certain colors and brightness otherwise you'd never get consistency. It's why artists draw/paint/etc. under specific lighting. It's why you put makeup on in front of lit mirror with a certain color bulb. It's why home depot sells warm, natural, and cool colored lights. etc etc etc

I can't believe people are actually making a big deal about this.

Re:White balance and contrast in camera.

[mattack2]

Interesting, yours is the first time I've seen someone say specifically that the absolute same image appeared differently.

Even on TV coverage, I've seen them show "white and gold" and "black and blue" next to each other or right after each other..

Re:White balance and contrast in camera.

[KozmoStevnNaut]

Based on the adjectives, one is more blue, the other is more white.

The photo shows plenty of color. From looking at the picture, the background in particular, it is very clear that the image is quite over-exposed, which will significantly lighten colors. Obviously, the camera based its light metering on the black parts of the overall rather dark dress, which caused the blue parts to become overexposed and the the background to blow out.

It's completely ridiculously obvious, to the point where I theorize that anyone seeing white instead of blue actually has a mild form of brain damage.

Re:White balance and contrast in camera.

[hucker75]

It's not obviously over exposed at all. The camera could have had spot metering and exposed the dress correctly, while overexposing the background. This is the right way to take a photo, so you see the subject properly.

Re:White balance and contrast in camera.

[Capt.Albatross]

You are entitled to your opinion, but your attempt to establish it as the sole objective point of view is doomed to failure, on account of its subjective basis.

Re:White balance and contrast in camera.

[dAzED1]

there are no "white" pixels on your screen. There's a backlight, sure, but that's applied to the entire screen, to adjust brightness. Color saturation at extremely high levels is black. At extremely low levels, it's white. That there is blue, is immaterial - give it extreme low color saturation with even just blue, it becomes white. There is an extreme difference between seeing something in person, and seeing something on a computer monitor where two degrees of artificial filtration are occurring due to technological limitations. And for the record, I saw blue and the brown/orange color used in the old cammo. I didn't see white :P

Re:White balance and contrast in camera.

[Capt.Albatross]

"Oh, so RGB={0,0,255} isn't "quantitatively blue", huh?
Horseshit. If there is no red, no green, and %100 blue, then the color is quantitatively blue."

actually no. ask anyone who has only blue cones... they will have no idea what blue is...

So if a tree falls in a forest, and someone, somewhere is deaf, then it doesn't make a sound?

Re:White balance and contrast in camera.

[KozmoStevnNaut]

And yet the dress is definitely over-exposed when we compare with model pictures of the very same dress from Amazon etc.

Re:White balance and contrast in camera.

[hucker75]

Maybe so, but that wasn't obvious from looking at the photo. To me the background was too bright, an indication that the camera has metered on the dress and left the background to get too bright.

Re:White balance and contrast in camera.

[KozmoStevnNaut]

To me it looks like it metered almost exclusively on a black part of the dress, blowing out both the background and the blue bits of the dress.

Re:White balance and contrast in camera.

[hucker75]

Cameras don't meter on a very small spot like that. Usually it's the centre 9th (one third width, one third height) of the whole photo. Either that or it takes account of the whole photo in a clever way.

Re:White balance and contrast in camera.

[KozmoStevnNaut]

The centre 9th of the photo in question is quite dark, too.

Re:White balance and contrast in camera.

[hucker75]

No, the centre 9th of the photo contains only dress, and both colours of it. So I'd expect it to balance this correctly. If I take a photo of a dark object in a brightly lit room like that, I either get the object looking correct, or too DARK because the camera is trying to get the bright room to look right. I can only surmise that the photographer used an Iphone or some such crap. I've never seen a camera make such an arse of a photo.

Re:White balance and contrast in camera.

[KozmoStevnNaut]

The picture was indeed taken with an iPhone :-)

We can definitely agree on the crappiness of the picture.

Frankly, my dear, I don't give a damn

[Anonymous Coward]

/thread

Re:Frankly, my dear, I don't give a damn

[serviscope_minor]

Then please hand in your nerd card on the way out along with the idiots who marked this "insightful".

How the brain perceives colour well is fascinating, and our subconscious is amazingly good at it. The "gold" colour is objectively gold (check in the GIMP), and yet some people's brains manage to correctly interpret it as black.

If you're not interested in that and the underlying algorithms behind how our brain works, then frankly I wonder what the hell you're doing here.

Or possibly you just think you're too cool to be interested in popular stuff, in which case, grow up.

Not a unique photo

[Teun]

The GF is regularly checking out the Dutch version of Ebay (Marktplaats) for clothing and these kind of problems are usually caused by poor quality phone camera's.

Re:Not a unique photo

[Coisiche]

This is slashdot; how are we supposed to know what GF stands for?

Re:Not a unique photo

[worf_mo]

Goldfish, d'oh!

Re:Not a unique photo

[serviscope_minor]

The photo may not be unique but it is interesting in that it seems very close to some threshold in the brain such that the population is more or less split into two very, very different interpretations. That is interesting.

Re: Typical nerd

[Anonymous Coward]

Worse, it isn't even being occupied by a pretty girl

XKCD shows you

[gbjbaanb]

http://xkcd.com/1492/

It is blue and black, but if you up the lighting, and/or display it against a white background the black lace part looks golden.

Some programmes on TV over here (including the excellent Last Leg - see it on C4 player) had it on the show, it really is blue and black.

Re:XKCD shows you

[Sesostris III]

The BBC also show the actual dress:

Optical illusion: Dress colour debate goes global

I see white and gold, although the actual dress is blue and black.

Re:XKCD shows you

[cascadingstylesheet]

The photo at BBC does look blue and black. The photo at PBS looks white and gold.

Back-end image file manipulation?

[swb]

I'd swear I saw completely different images of these dresses posted, at extremes of the color controversy and neither was at all ambigous as to what color it was.

I wonder what the likelihood is that two or more images were served to clients, either at random or by some algorithm, to further the controversy? I can see one single ambiguous image that could go either way, but most of the examples I saw looked to be tweaked for maximum color association.

If you served tweaked images to clients so that "everyone" saw a different image, including people who saw different images at different times muddying their memory of what they saw over time, you could really amplify the controversy since people would actually be seeing a different image.

Re:Back-end image file manipulation?

[werepants]

Of course you could, but what's more likely - some elaborate scheme to create a viral controversy that's tied to no obvious material benefit, or a picture that just happened to be taken with a shitty cellphone that gets interpreted differently by different viewers? What's more, lots of people looking at exactly the same image, at the same time, on the same device (for instance, my wife and I) came to opposite conclusions.

The summary and linked xkcd comic do a completely accurate job explaining the phenomena, no conspiracy theories required.

Re:Back-end image file manipulation?

[jason-999-]

That's what I originally wondered, so I downloaded the image. I can see it as white/gold, close/cover my eyes so it's dark, then look and it'll be black/blue.

Re:Back-end image file manipulation?

[swb]

Yes, I realize that the sort of Occam's Razor kind of explanation is that it is a result of a bad picture that exposes some kind of color processing ambiguity and not a result of some kind of manipulation.

That being said, I think it's not unrealistic at all in era of clickbaiting and relentless social media trolling that someone would want to experiment with a scheme for manipulating social media memes or figuring out a way to amplify their views. As far as I know, money can still be made on web advertising, drive-by downloads, tracking, etc.

And as the AC poster who also replied said, there ARE organizations with a vested interest in manipulating socal media, whether its "merely" for advertising purposes or for more nefarious reasons.

Color means many things

[ganv]

This reminds me of the great entries from the competitaion to explain 'What is Color':

https://vimeo.com/87968614 http://www.centerforcommunicat...

By the way, I see white/lavender and brown. It would be very interesting to know what lighting/image manipulation was done to get those colors out of a dark blue and black dress.

Re:Color means many things

[BarbaraHudson]

Load it in the gimp, increase the brightness (a lot) and the contrast (again, a lot). Without that, the color picker says blue and brown/grey. What color is the dress? Who knows. What color is the dress as shown in the picture? Blue/brown/grey.

That being said, if you look at the one place in the picture that doesn't appear to suffer as much from color imbalance / over-exposure (the item of clothing hanging on the back of the chair, on the left side at about waist height), It's a cow pattern print.

Re:Color means many things

[karnal]

That's my thinking exactly. Who gives a shit what the color of the dress actually is; it's obvious from the image displayed that you're not really going to "see" the correct colors anyways.

Re:Color means many things

[werepants]

The GIMP doesn't really mean anything, because what's at play here is our mental perception of color. White snow in the shade has a distinct blue tone if you look at it in a photo editor, but that doesn't mean that it is blue. Really, we've got that exact phenomena going on here - the colors could be adequately described two different ways, white and gold dress in the shade (blue-ish light) or blue and black dress in incandescent light (gold-ish). It's really a matter of interpretation.

For another great example of just how confounding this effect can be: http://upload.wikimedia.org/wi...

Re:Color means many things

[BarbaraHudson]

I was not posting about the color of the real / original dress, and I made that clear. The color of the pixels in the picture is something that is easy to determine. They are blue/dark grey. That people might perceive them as something else doesn't take away from the color value of each individual pixel.

If snow looks a little blue son-screen, you should try adjusting the color temperature of your screen.

Re:Color means many things

[werepants]

I use a color sensor to calibrate my screen for any production work, which I do occasionally (although not as much as I once did) as a professional photographer. Our brains lie to us, and the "actual" color displayed on screen is next to meaningless. That's what this whole illusion is about - regardless of the color on the screen, we can interpret it to be dramatically different "real" colors, based an the assumptions we make subconsciously about the context of the image.

By the way, I can see it both ways. Just look at it and imagine the dress is in a shaded alcove with incandescent lighting. Then, imagine that it is a shiny dress with bright yellow light on it. All I have to do is tell myself one or the other of these scenarios and I can see it blue/black or black/gold.

Really, the whole point here is that efforts to "analyze" this with photo editors misses the point - this is an optical illusion borne out of artifacts in human vision processing. It isn't a physics or technology problem.

Re:Color means many things

[werepants]

The PICTURE of the dress is in fact a very pale blue color, with a brown/dark gold color.
The PICTURE itself (at least on CNN) is not blue/black.

The point is, this doesn't help us solve the problem. Because very pale blue color (on the monitor) could be gotten by capturing an image of white in the shade, or washed out dark blue. Or lots of other more obscure and less likely ways.

Trying to "analyze" this with photo editors misses the point - this is an optical illusion borne out of artifacts in human vision processing. It isn't a physics or technology problem.

Changed for me

[skirmish666]

The first time I saw the picture I could swear it was white / gold. I could see a slight blue hue to the white part but it was more or less white with gold.

After I read another article and saw the dress in a catalogue I read the first article again and it appeared blue / black. I couldn't believe it appeared so differently and had to check I was reading the same article with the same photo again.

Re:Changed for me

[drinkypoo]

Is it in fact the same photo again? Or was it messed with again?

Re:Changed for me

[EvilSS]

The first time I saw the picture I could swear it was white / gold. I could see a slight blue hue to the white part but it was more or less white with gold.

After I read another article and saw the dress in a catalogue I read the first article again and it appeared blue / black. I couldn't believe it appeared so differently and had to check I was reading the same article with the same photo again.

I had the same experience. Kind of jarring isn't it? I had even saved a copy of the image when I first saw it (to play with in photoshop) and checked that to be sure someone wasn't messing with me and everyone else and swapping out the image.

We as a species always seem to be of the "believe it when I see it" persuasion but something like this happens and it is a pretty in-your-face example of just how much our own brains manipulate our sensory input before presenting it to us a reality.

Perception

[Dan East]

First off, the picture is crap. It's overexposed and the white balance is off by a mile. My 10 year old Razr flip phone took better pictures than that.

However, there's still a human perception factor going on. I had looked at the picture on my laptop, and it was clearly white and gold. Then later I pulled the exact same picture up on my iPhone to show it to someone, and it looked black and blue. I then concluded that the picture looked different on my laptop than my phone due to differences in the display. When I got back home I pulled the picture up on both my phone and laptop to do a direct comparison, and both, including on my phone, looked white and gold again.

So I think it depends on whether your eyes are currently adapted to dim indoor lighting or bright outdoor lighting, in addition to the backlight on your device also changing the hue depending on if it's automatically full bright for outdoors or dim for indoors.

Re:Perception

[qubezz]

I think that understanding photography and exposure is the key to recognizing the color in the picture - if you are familiar with photography, you can see that the background is light and almost blown out, and can use that as a reference for how the entire photo was lit when it was taken. However, if your brain doesn't process the context of the photo, and you evaluate it based on the blue background of a web site or a dark room around you, maybe you have the optical illusion that it is white. I can not unsee it as blue because I recognize the photo's lighting.

Re:Perception

[itzly]

However, if your brain doesn't process the context of the photo

I did notice the overexposed background, and I interpreted as a picture taken of a white/gold dress in the shade against a sunny background with a blue sky.

Re:Perception

[Kjella]

As such it's not the brightness, it's the color temperature. The spectrum of light ranges in color temperature from 1850K at sunrise/sunset to 6500K on an overcast day to 15000K under a clear blue sky. The eyes adjust to this, if you look at someone using a cell phone at night it'll probably seem to have an eerie blue glow as it has daylight color temperature. So with "nightvision" the dress looks blue/black, with "dayvision" it looks white/gold.

The people trying to read the RGB values to determine the "truth" forget that the color space assumes you have a D65 white point. Basically your LCD screen is trying to show you correct colors for overcast daylight. If you stare at the red sunrise/sunset or the blue sky for a while and then look at the LCD screen, your color perception will be off. Apparently this picture is in just the right sweet spot to confuse a lot of people.

Re:Perception

[Anonymous Coward]

why white and gold, blue and black? Actually, in RGB color codes, white is (255,255,255),which means (red,green,blue),so as gold(255,215,0),blue(0,0,255)and black (0,0,0), convert white to blue just needs to cancel same degree red and green, convert gold to black cancel red and green as well , so in dark light, all color blinds could see black and blue because red and green is 0 for them.

Re:Perception

[zoffdino]

I'm a graphic designer and photography enthusiast. I'm typing this on a NEC PA242W color-calibrated monitor, in a near-dark room. That dress is white and gold. the white part has a blue tint but I wouldn't call it blue. The colors look the same on my iPhone 5S. When I bring the iPhone outside, the blue tinge is more apparent (a short of light sky-blue) and the gold/brown turn darker, somewhat into the black category.

The different isn't in the screens, it's in your eyes, caused by environmental light. A sunny day at noon can be 100x brighter than even a well-lit room with floor-to-ceiling windows. If it's sunny in your location right now, try this: find a view point where you can frame both the sky and a patch of dirt land (no grass or foliage). Put the camera in manual mode, pick a shutter speed, manual daylight white-balance (6500K) and a low ISO, start with a large aperture (like f/4) and gradually step it down (like f/22). The sky will appear more blue and the ground will appear darker.

That's exactly what our eyes do. In darker places, the pupils open up to accept more lights, the highlights (blueish-white) gets push up to white but mid-tones and shadows are preserved. In bright places, the opposite happen: lower mid-tones and shadows are pushed to near-black, highlights are pulled down to reveal the blue accent.

Re:Perception

[Kartu]

Could you please elaborate, as I'm trying hard to imagine how on earth can that look "blue and black" as actually HIDING the context I clearly see gold color.

PS
I'm talking about this image:
http://img.washingtonpost.com/...

Re:Perception

[Lunix Nutcase]

That dress is white and gold. the white part has a blue tint but I wouldn't call it blue.

No, it's not. The original picture is overexposed and washes out the colors as shown in the picture I linked which contains a second picture showing the actual colors.

Re:Perception

[Lunix Nutcase]

It looks "blue and black" because that's what it is.

Re:Perception

[denzacar]

That's not how it would work. It's cloth. It has ripples and shades.
There would be patterned ripples across the dress where "real" colors would show in shades or highlights.

IF the photo itself wasn't messed up by the camera/software, effectively replacing the color palette in the entire photo.

William Gibson foresaw this in his "Bigend cycle" books.
Hubertus Bigend wears International Klein Blue suits just to fuck with everyone else, as it can't be represented correctly on monitors or in print - note two different whites in the color corrected photo in order to get both the skin tones and the dress right.

Gibson just didn't thought of adding shitty CCDs to the list of technology with issues with reproduction of the color.
Or, illiterate "designers".
She calls it "Royal Blue" in the video.

Sure... If one could get people to wear a computer screen, calibrated to show the web palette of colors.
There will be little difference. THERE. On the screen.
Especially if one's screen is not even close to calibrated.

On the other hand.... Trying to mix those "equivalent" values listed in RGB and CMYK.
In Web-RGB they WILL look exactly the same. And so will the blacks.
http://i.imgur.com/QdL00rr.jpg

Ask the same industry standard company to do it using their other, more professional tool, with full RGB and CMYK gamut...
http://i.imgur.com/46B6H55.jpg

And just try using the RGB and CMYK values for Ultramarine (essentially IKB).
http://i.imgur.com/Idc2pr7.jpg

The color she envisioned on her screen is NOT the color of cloth chosen for the dress, based on the color on the screen.
She wanted "royal blue" but picked ultramarine - because Web-RGB royal blue is closer to aquamarine IRL.
The person designing the dress DOES NOT KNOW WHAT COLOR IT IS.

It's not about "rods and cones" and "everyone seeing colors a little differently".
It's about people using wrong names for colors, often calling many different colors by the same name and the same color by different names.
Then it is about faulty capture technology and badly written color conversion and calibration algorithms.
Then it is about faulty display technology, which can't show the same image under different viewing angles.
THEN, and only then, MAYBE, color perception and ambient lighting might fool the untrained eye.

But it is most likely that in most cases it is again different people calling a shade of red pink and orange.
While trying to GUESS the "correct" color from a crappy photo on a crappy screen.

Re:Perception

[werepants]

It isn't about the camera's white balance. It is about the light on the dress, and the lack of sufficient context to determine exactly which light the dress is in. Is the dress in the shade, with a blown out background from a different light source? Or is the dress in the same blown out golden light as the background? The brain can choose one way or the other - if it prefers to think the dress is in the shade, you see white and gold. If your brain thinks it is washed out by the yellow-ish light, you see black and blue.

If you REALLY understand photography, you are well acquainted with the fact that outdoor light (shade especially) is dramatically more blue than incandescent light. If you've got both in the same scene, you get problems like this, and there's no good choice for the camera to make. This is why there are things like gels for flashes, because it isn't a problem with the way the photons are processed by the camera, it's the fact that physics delivers very different photon wavelengths from on object depending on the incident light source.

Re:Perception

[zoffdino]

I forgot to quantify it "the dress as shown in the picture". I know the actual dress is blue and black, but the crappy camera work was what generated so much controversy.

Re:Perception

[BlackPignouf]

Thank you. GP is just a guy with a camera who thinks he understands it all just because he was lucky to process the dress as black and blue.
I've also seen enough pixels in my youth to think that I got it right : unfortunately, there's just not enough context, and I cannot see any other color than white and gold.

Re:Perception

[ScentCone]

He's not talking about the ACTUAL dress, he's talking about the photographic portrayal of a dress is the crappily exposed and presented JPG that everyone is looking at. The dress, as recorded in the JPG, is a barely-blue-tinted light grey, and the black elements have a demonstrably uneven RGB that makes them look gold (because that data represents a color low on blue ... which is to say, it's a golden hue).

Re:Perception

[jez9999]

Those two colours are "slightly off-white blue" and "gold".

different from Cornsweet

[e**(i pi)-1]

There is big difference between the Tom Cornsweet illusion which is also addressed in the XKCD http://xkcd.com/1492 Also this explanation https://www.youtube.com/watch?... While both are well done, they miss an important point. The Cornsweet paradox works for everybody. Universally. The dress paradox not. For most people (75 percent in one poll), the paradox does not work. (I myself find it hard to believe that some see initially a blue dress). But it seems that different brains work differently. This is why the phenomenon must be interesting for psychologists.

Re:different from Cornsweet

[narcc]

I'm convinced that this is just a joke that I've missed. I've tried good displays, crappy displays, various lighting, brightness settings, backgrounds, room lighting, viewing angles and probably something I've forgot. I can not get that dress to look white and gold.

It reminds me of "The emperors new cloths".

Re:different from Cornsweet

[itzly]

I have exactly the same, except I can not get it to look blue/black.

Re:different from Cornsweet

[Anonymous Coward]

I see it as Slightly Blue and Gold, and a kind of shitty picture. After looking at the 'real' pics, I now realize that I was wrong. It is in fact a VERY shitty picture.

I was curious and so I popped it into mspaint and looked at the RGB values for the various pixels.
Original pic:
The 'white/blue' regions have definitely have higher B values than R or G. The 'gold/black' regions similarly correspond to a brown (higher R+G) when under the microscope.
The second pic:
Pic as provided by the BBC and others shows wildly different RGB values. The formerly 'gold/black' regions now lack any particular channel dominance; all the pixels I looked at were an even R-G-B balance, and all about 30/255.

Re:different from Cornsweet

[narcc]

I considered that as well, and pulled down the picture from multiple sources.

Re:different from Cornsweet

[narcc]

Ah, so you're in on the conspiracy!

Re:different from Cornsweet

[BlackPignouf]

I'm in that camp too. But :
I tried to put an incandescent light bulb close to my screen, took 3 steps back and squinted a little.
The dress still doesn't look clearly blue/black to me, but the blueish tone is now much more saturated and dark, and the gold is dark enough that it could be black with some orange highlights.
The effect is gone when I walk back to the screen, but at least I can understand that it can look blue/black to some people (=morons). :)

neither

[ihtoit]

it's sky blue and muddy brown.

See, I went with the definitive and used a CALIBRATED COLOUR PICKER.

Can someone please answer

[krkhan]

Why is this particular image such a viral hit? Have people tried creating images which would appear of different colors but failed to split opinion on such a wide scale?

[Rant]
First everyone started arguing over their gut-reaction, "it's obviously color X!"

Then everyone started trying to sound smart by doing some variation of: "Colors are perceived by your brain! Can you imagine that? Your brain. Like, literally!"

The /. crowd can mostly understand without much fuss that colors are subjective. A more geek-oriented analysis -- which I'm pining to read somewhere -- would deal with what took the Internet so long to catch up to this phenomenon.
[/Rant]

Re:Can someone please answer

[JaredOfEuropa]

The real question is: was the girl who posted this image able to capitalize on its viral popularity, like the owner of Grumpy Cat?

Re:Can someone please answer

[serviscope_minor]

What's wrong with that?

It's made people realise and think about that our brain is what interprets what colours mean and in deeply non-trivial ways. It's certainly only for a short time in many cases but it's got people thinking about things they'd not normally think about.

I can't see anything wrong with that. I'm a computer vision person, so until today colour interpretation has been an esoteric topic that no one really cares about. Today I can actually talk to non-experts on a topic find interesting, and one that they temporarily find interesting too.

I like this.

Re:Can someone please answer

[smooth wombat]

I said elsewhere that this is a scam for the following reasons.

Her Tumbler account (Swiked) shows the initial photo asking what color is the dress. A day or so later she posts a second picture of someone wearing the blue/black version, stating:

this is the dress as i saw it on the day of the wedding. blue and black. It's just that one photo bUT it's so weird???!??

Here's the thing, she does not say the second photo is the same dress ON THE DAY of the wedding, only that the photo shows what she saw.

What most likely happened was she took the first picture on a different day and because it's white/gold, it took on the cast of the lighting to give a blue hue and her friends had a disagreement over the color which started the whole thing, especially if someone is red/green colorblind.

Second, if you look at the top edge of the dress in the first picture you can clearly see what would be considered "virgin" light, i.e. light which is not reflected but directly falling on to the dress. If the claim is that our eyes are fooling us because of the bright background or because we can't be sure what the lighting situation is, this light should produce a different color under it, just as light shining across a rippled surface can produce different colors.

Except it doesn't. There is no transition from the slice of light to the shaded portion. It's one continuous tone. In fact, as the picture shows, there are multiple shadows of different angles and lighting conditions which should produce different colors, but they don't. For instance, under the cape/shawl toward the upper right, the area under the shadow is darker but not a different color.

Further, if our eyes are being deceived by the bright background, covering up all but a small portion of the dress should reveal the true colors because then there would be nothing to confuse us. Except that doesn't work either, the dress stays as white/gold with a blue hue.

And finally, if you look at the edging in the middle of the picture then down to the lower left corner (our left), you can see shadows under the gold edging. If the dress in the picture was blue/black you would not see such distinct shadows as are shown in the picture.

So, the dress in the original picture IS NOT the same dress she shows a day or so later but is the same style.

One final thought. The opposite of blue on the color wheel is gold and the opposite of black is white. However, if you look at what is blue and black in the "good" picture, they are not the correct parts. The bodice is blue which means in the reversed portion the bodice of the original picture should be gold and the edging should be white. Except that is not what is shown. The bodice is white and the edging is gold.

And for the record, I am not red/green colorblind. I pass all the Ishihara color tests without issue.

Re:Can someone please answer

[serviscope_minor]

I said elsewhere that this is a scam for the following reasons.

Except a good chunk of slashdot, absolutewrite and a few other also completely unrelated forums and IRC channels are in on the "scam" and have a bunch of people who have joined the conspiracy to pretend it's blue and black. Or white and gold, in which case I'm in on the scam and hereby declare I got my note from a shady black vehicle with blacked out windows this morning at precisely 5:50am at the dedicated drop point.

It's not a scam, because it frankly doesn't matter what the original colour of the dress is.

The interesting thing is it's sufficiently close to some average threshold of human perception that nearly half the population perceive it as completely different from the other slightly-more-than half.

At that point it wouldn't actually matter if it was a 'shopped image of a dress covered in purple unicorns.

Re:Can someone please answer

[tehcyder]

You sound like one of the Reddit geniuses who identified the wrong Boston bomber.

Re:Stupid post

[Lunix Nutcase]

Nope Slashdot is now Digg 3.0.

It's neither

[krkhan]

The image is actually a PNG with some lines criss-crossed over a zero alpha-channel, everyone just has different crappy wallpapers.

TED Talk by Bray Lotto

[Mike Morgan]

http://www.ted.com/talks/beau_...

I can see both now

[Anonymous Coward]

If you see the white/gold image, scroll down the page below the image, squint, and slowly scroll up from the bottom. You will see the blue/black.

The white is not blue

[gweihir]

That is an effect from the picture sensors and optical brighteners (which give white a blue/violet tinge from converting UV to visible). However I am completely mystified as to where anybody sees black. The small horizontal stripes?

Re:The white is not blue

[serviscope_minor]

I think they perceive the picture to be under intense orange light, like incandescend bulbs. This will wash the blue out to grey and make the black (which always appears as dark grey) a dar orange (i.e. gold) instead. Once your perception of one colour is fixed, the others will follow, or rather, they go as a group basically based on what you perceive the light colour as.

That said no matter how hard I try, it still looks white and gold.

Re:The white is not blue

[gweihir]

Sounds reasonable. But the gold is just to bright to be black in any light. At most its is a 50% or so gray.

The interesting thing is however the reminder that different people may see colors differently.

white (with *maybe* teeny blue tinge) and gold

[cascadingstylesheet]

Not seeing where black comes into it. Don't see any black at all.

I looked at the PBS story version.

I was looking at this

[MrKaos]

and then I realized I just don't care.

Re:I was looking at this

[serviscope_minor]

and then I realized I just don't care.

Huh, how brains work isn't interesting! Who knew? BTW: the title of the site is "news for nerds ... " and this certainly qualifies. Just because it's popular doesn't make convincing yourself that you're indifferent any cooler.

And yes it is interesting. If you've ever tried to write code to interpret colours correctly andautomatically, you will be in awe of how good we are at it. Anything that sheds light (pun intended) on such a fascinating process is interesting.

Re:I was looking at this

[MrKaos]

Huh, how brains work isn't interesting! Just because it's popular doesn't make convincing yourself that you're indifferent any cooler.

I suppose you are right and I don't mean it that way. The stuff about the brain interpreting colours is interesting but I find that when hollywood gets involved in geek stuff they make it lame and uninteresting.

Re:I was looking at this

[tehcyder]

I can only apologise on behalf of the New World Order government which recently passed the law requiring you to view and comment on this thread on pain of up to ten years imprisonment or an amusingly heavy fine.

sigh

[s.t.a.l.k.e.r._loner]

This "____-gate" shit needs to stop. Now.

Re:sigh

[hughbar]

Stopshitgategate ?

Re:sigh

[freeze128]

...and what the hell is a "color scientist"? Is there really a need for them in our society? Do they do anything besides comment on trending photos?

Re:sigh

[Livius]

Aside from Shawinigate (the pun was just too good), it doesn't exist outside the US.

Re:sigh

[s.t.a.l.k.e.r._loner]

Touche, my young son. Touche.

Re:sigh

[tehcyder]

This "____-gate" shit needs to stop. Now.

Since Watergate was back in 1972, I think your protest is a bit tardy.

Comparison over multiple screens

[MassEnergySpaceTime]

Earlier tonight, I compared the same picture on the same site using 3 different computer monitors side by side and 3 different tablet screens.

To me, the white/blue part of the dress is sort of a pale light blue on all 6 screens. But they're different shades of pale blue. On one screen, the blue stand outs a little stronger. On another screen, the blue seems more faded towards white.

For the black/gold/tan part of the dress, on some screens, the tan color seems more faded, making the darker part stronger, and I COULD call it black. I know it's not PURE black, and it's not as black as that cow patch thing in to the left of the dress. But I could call it a shade of black. On other screens, the tan part stands out more, and I would definitely not call that part black. I don't know if I would call it "gold", but I would call it tan/light brownish.

So I think the screen settings is one variable that contributes to what colors the user thinks they see in the dress picture.

For the situations where different people are looking at the same screen or printed photograph, my guess is that the variability comes from the color/brightness/etc sensitivity of their eyes. For example, in my own eyes, one of them sees the wall in a brighter shade of white (and possibly slightly red tinted) than the other eye. Perhaps those who aren't as sensitive to blue might see the blue/white part of the dress as a shade of white, and call it white.

I guess this picture is just one of those freak pictures where the colors are at some borderline that could be interpreted as one shade of color or another.

Slashdot can do video, but not an image?

[wonkey_monkey]

Couldn't Slashdot post the image in the summary? I mean, it's not even like it'd be an illustration; this article is literally all about a specific image.

You can do it with annoying autoplaying videos, so why not a simple JPEG?

Re:Slashdot can do video, but not an image?

[slashdice]

and what does frequent contributor Bennett Hasselton think about it?

Histogram

[TrashGod]

For reference, here's a histogram of the image's RGB color components.

obligatory schoolhouse rhyme

[jabberw0k]

In http://xkcd.com/1492/
Columbus sailed the ocean gold

There's a third camp

[Zcar]

Blue and orangish brown. That's what I see.

Re:There's a third camp

[David_Hart]

Blue and orangish brown. That's what I see.

That's what I see too.

Granted, my first off-the-cuff reaction was white and gold, partially because I was expecting white/gold or blue/black due to the choices presented. Expectation has a lot to do with what we perceive. However, when I looked closer for a minute or two my color perception shifted to blue/burnt orange and that's all I see now.

Color Illusion

[physicsphairy]

The XKCD plot just makes me see gold and white at different levels of brightness. But I did find this color illusion featuring yellow and blue. The dogs are actually the same color, which you see if you look at them individually through a small aperture
http://i.imgur.com/sh5NwCK.jpg

Make it pretty obvious that at some point your brain switches from wanting to see blue to wanting to see yellow based on the color context. It would appear some of us are slightly different in where transitions like that occur.

Same thing with Daleks

[wonkey_monkey]

Is the Dalek on the left red or yellow?

http://horman.net/avisynth/dal...

Re:Same thing with Daleks

[brianerst]

Um... orange?

What's really intersting here is....

[rizole]

...not the colour perception issue IMO but that this has caused such a big storms on the tubes in the first place. I got shown the picture yesterday and could see it could be blue or white and it's difficult to tell which given the poor lighting. I mean we've all more or less got cameras in our pockets or bags all the time these days, how have so many people missed seeing poorly lit, bad quality pics with white balance issues?

Re:What's really intersting here is....

[Al Al Cool J]

Really? because I'm surprised at those who downplay it.

When you see it as blue-black it looks really blue-black. When you see it as white-gold it looks really white-gold. It's astonishing when you have two groups of people looking at the same thing at the same time under the same lighting conditions and seeing such radically different things, to the point that is unimaginable for each that the other person could be seeing what they say they're seeing.

What I find intereresting is that in all of recorded history, no painting, photograph, or object has ever exhibited this property in such an extreme and shocking way before. If it only work on LCDs then that might explain it, but I don't know if that's been established.

Its light blue and brown

[TomGreenhaw]

I checked with photoshop...

White-n-Gold, Then Black-n-Blue

[Zamphatta]

I originally saw it as white & gold when I saw it on a forum Thursday night, and figured it was being posted by trolls 'cause they were asking people what color it was even when it was obviously white & gold. Friday morning I saw it on the news on TV, and they explained that some people see it as black & blue. I thought that was a bit crazy, but after a few minutes, I was suddenly able to see it as black & blue......and I can switch back & forth between the two color sets if I try hard. Awesome.

Omg, my GF talked to me about it

[Eloking]

For the first time ever, /. reported one of the uninteresting daily gossip from of girlfriend. SHAME ON YOU /. !!!!

Re:Omg, my GF talked to me about it

[burtosis]

Lmafo yes the same happened to me +1000

Re:Omg, my GF talked to me about it

[tehcyder]

For the first time ever, /. reported one of the uninteresting daily gossip from of girlfriend. SHAME ON YOU /. !!!!

Yeah, GF, sure. You could at least try to sound plausible.

Re:Omg, my GF talked to me about it

[Eloking]

For the first time ever, /. reported one of the uninteresting daily gossip from of girlfriend. SHAME ON YOU /. !!!!

Yeah, GF, sure. You could at least try to sound plausible.

Yeah I do have a girlfriend and she's into stuff like Vampire Diaries, Pretty Little Liars, Gossip Girl, Keeping Up with the Kardashians and the likes.

But I convinced to start LoZ and she finished OoT and MM, just started WW, which is nice.

And yeah she showed me that dress picture in the morning asked me what colour it was.

XKCD to the rescue

[McLae]

OK, so why is XKCD the only place that explains this so it makes sense? 3 million bloggers, countless news stories, and this one cartoon ups them all. Question, is XKCD genius, or the others idiots? (or both?)

While I was reading this...

[JohnnyDoesLinux]

I realized that I the time wasted could have been used for a good fart. ... Ummm nevermind.

Dressgate??!!!

[Anonymous Coward]

Dressgate? Seriously? Just kill me already.

Re:"reflections"

[jrumney]

The black looks gold because the camera's auto white balance removed the blue. Simple as that.

rubbish

[rkoot]

1. save picture. 2. open picture in image editor 3. invert colours. 4. voila.

It's neither

[renergy]

Hardly I'm the only one who see (dirty) gold and (light) Blue(ish)...

Absolute stupidity

[Khyber]

Shitty potatophone camera fucks up picture, the internet loses its fucking mind over colors.

Re:Absolute stupidity

[itzly]

okay, okay, but what colors do you see ?

Re:Absolute stupidity

[Khyber]

I don't see colors, I see the electro-magnetic spectrum in the range of 380nm-740nm.

Colors are for artists.

Re:Absolute stupidity

[Al Al Cool J]

Have you been in a room with two different groups of intelligent perceptive people, all looking at the same picture on the same screen and seeing two totally different sets of colours, and screaming at each other because of it?

Maybe you have, maybe you haven't. But those who haven't experienced it, are probably not the best persons to judge it. Just saying.

Re:Absolute stupidity

[Foreign Entity]

The point isn't that the picture is shitty. The point is that two people are observing the SAME picture, and seeing two drastically different color combinations. The psychology behind this is facinating.

Re:Absolute stupidity

[serviscope_minor]

Yes but it's popular and sadly too many slashdotters not only feel that have too much nerd cool to ever be interested in something normal people are interested in but feel the need to tell us about it too.

Re:Absolute stupidity

[tehcyder]

I don't see colors, I see the electro-magnetic spectrum in the range of 380nm-740nm.

Colors are for artists.

Everyone say hi to slashdot's new GUI programmer.

Re:Absolute stupidity

[tehcyder]

Yes but it's popular and sadly too many slashdotters not only feel that have too much nerd cool to ever be interested in something normal people are interested in but feel the need to tell us about it too.

It's similar to the hipster "I liked X before it was popular" thing. Intellectual snobbery.

Are you serious?

[Rambo Tribble]

You actually expect us to answer that question without a Pantone chart?

It's a trick question

[HeadlessNotAHorseman]

I blogged about it here (with a bonus explanation for how religion works), but in essence the question as asked presents you with two incorrect answers (each possible answer has one colour correct, one incorrect), so you pick one of them and then argue with everyone else that you are right.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Auto-adjust levels Corel Photopaint

[BenJeremy]

Bam, Blue and Black.

Q.E.D.

Bad lighting, bad camera, lazy picture taker.

Re:Auto-adjust levels Corel Photopaint

[itzly]

But that's not what this is about. It's about two groups of people seeing different things when looking at the same picture.

Re:Auto-adjust levels Corel Photopaint

[thejynxed]

What is interesting to me, is the fact that the brain itself might be operating automatically in a way that color correction software does when working with photographs for some people, while for others it simply does not do so in a consistent manner, or at all.

That being said, my first self-response when seeing the photo was "The entire photo looks wrong".

I saw a kind of bluish-white and the kind of brown with yellow hints that you see when you shine a flashlight that gives off "yellow" light through dyed black cloth (which is also similar to the effect you get when you apply a little chlorine to cloth that has been dyed black).

Another thing that I found personally interesting to me, was that from my perception, the overall photo itself seemed to have had a blue tint to it, as if the room it was taken in was being lit by a LED light similar to the ones I have in my bathroom (when first turned on, the light they give off is very blue-shifted until they warm up and become a very bright white).

White & Gold

[jomcty]

In our house, three of us see the dress as white/gold, while two of us see blue/black. My parents saw the dress as blue/black. Colorblind Assistant sees basically soft brown/soft blue. Interesting indeed.

Actual Dress Colour

[XB-70]

The actual colour of the dress is "No, That Dress Does NOT Make You Look Fat."

Re:Stupid post

[dcollins117]

If it makes you feel any better about this issue appearing on a tech site, it's part of what makes computer vision hard. Colors change under different lighting conditions so how an algorithm treats color information when identifying an object or analyzing a scene is an interesting problem.

Huh?

[daveime]

Am I the only one who sees blue and gold?

Re:Huh?

[iggymanz]

no, the "white" I see as defintely of a bluish hue, but of the kind I expect on a white shirt in shadow. So I say "gold white" looking at the dress, but looking at it in corner of my eye while looking away from it I see black / purple! My family is half/half divided on what the colors are, which is funny

That's just one part of the elephant

[SuperKendall]

It's primarily an effect of the camera fooling around with the white balance and contrast/light to try to get a correct image and failed.

This is not the whole story - the first time I saw it I saw a strongly blue/black dress, which no apparent white balance issues..

The white balance being off is part of the effect to be sure. But so is the surroundings (as XKCD refers to). But that's not the main story either, because no matter what I do both sides of the XKCD image look at least light blue. In some conditions the dress in the original photo appears pure white.

I think it's combination of all those factors but then at the end, a large push from the brain one direction or the other.

The individual theories are like the blind man around the elephant, each convinced it knows what is there from a part...

Add to it that the picture looks different depending on which display you have

That's the thing - looking at the EXACT SAME DISPLAY from minute to minute can show you different colors. I've seen it flip multiple times, on the same monitor in the same lighting conditions. That's the aspect that is amazing to most people, is they can see it change under almost any condition.

Re:That's just one part of the elephant

[arglebargle_xiv]

... that dress...

What dress?

Looks different on same display at same angle

[SuperKendall]

Ask the person next to you, and they will tell you it is blue and black. Turn your screen towards them and the effect will be reversed.

I looked at the image on my phone. It appeared strongly blue/black.

I showed the image to my wife, straight on. Same angle. To her the image appeared STRONGLY white/gold, when I asked her if there was any blue at all she said no...

Same image on the same display at the same angle, within a minute or so.

Re: Looks different on same display at same angle

[shahlak]

Same here. I see white & gold and my husband blue & black irrespective of the angle. I also read an interesting theory that most men see blue & black and women white & gold. I forwarded the pic to one of my girls group on whatsapp, but was surprised that majority said blue & black. Another, theory says mostly older people will see blue & black. My parents saw blue & black, so it could be true for them. But my husband, me and my younger sibling are under 30. Everyone else except me saw blue and black.

Wrong interpretation

[SuperKendall]

It is blue and black, but if you up the lighting, and/or display it against a white background the black lace part looks golden.

In the XCKD comic the dress on the BLUE background has "black" that appears golden. In the original image the dress is on a white background, but apparently to 75% of people it appears white/gold...

Re:Wrong interpretation

[serviscope_minor]

In the XCKD comic the dress on the BLUE background has "black" that appears golden. In the original image the dress is on a white background, but apparently to 75% of people it appears white/gold...

On the blue background, you believe the light is blue. So, the blueish tinge on the white is from the light and the golden gold is more or less unchanged. On the one with the orange faced person, you believe the light is strong yellow, so the illuminated colour must be blue, not white and that implies the black will have an orange hue.

The thing is not the background but what your brain intreprets the light colour as. Tweaking the background is not the only way to achieve it but it is a way to do it reasonably reliably. What's amazing is that some people's brain succeeds in correctly interpreting the light and colour despite having very little and washed out background.

But anyway, the point is the ligt not the background.

Re:Wrong interpretation

[SuperKendall]

The thing is not the background but what your brain intreprets the light colour as. Tweaking the background is not the only way to achieve it but it is a way to do it reasonably reliably.

As I said though, it's not really the same effect. The dress looks Blue/Black to me in the XKCD comic on both sides, on the dark background it just looks like a lighter shade of blue.

At times the original dress image to me (and others) has looked pure white, not just a light blue. I agree it's about interpretation of the light color, the surprising thing is how strong the interpretation is - either way. That's something the XKCD comic does not convey at all.

Re:Wrong interpretation

[serviscope_minor]

the surprising thing is how strong the interpretation is - either way.

It's *vast*.

If you're a bit of a programmer and have some time, try the following.

Find a red object (or any other colour of your choice) and photograph it in a variaty of conditions, e.g. indoors, outdoors under different lughts, with or without a flash and so on and so forth. Write a script to extract the pixels from the object (make sure the object completely covers some square or circle in the middle of the image to make this possible, you don't need all of the pixels after all), and create an image out of the resulting pixels. You might want to sort them by for example absolute intensity on one axis and relative proportion of blue on the other (and otherwise random).

You will be amazed at the sheer variety of colours that make up the same "red". Not only that but you always strongly perceive the object as the same colour despite these.

The multiple parts that make up the effect

[SuperKendall]

Everyone seems to have a singular theory as to why the dress looks different, but each one by itself normally does not have such a dramatic effect as this.

In the interest of science I am going to lay out all of the various aspects of the dress I think together contribute to the ability of the dress to appear strong blue/black or strongly white/gold:

Main aspects:
White balance of course, lending a slight yellow cast to the blacks.
Pure white background.
Hint of golden object to far left bottom.
Gradient strength of lighting from top to bottom, that is to say the bottom blacks are purer black than the very top.

Questionable but probable:
Brain itself deciding to lock into a specific color instead of a washed out color.
Possibly frequency of stripes?

I think the brain is the wildcard here, I think all of the other conditions contribute to leading the image to a place where it's exactly on the cusp between two possibilities for the brain to register.... either the brain sees a washed out blue, thinks of white in shade and thinks of the dress as white so then the black gets shifted along with it to gold.

Or the brain latches on to the black first, darkens that, and then the blue gets pulled along with it and seen as a strong blue.

Re:The multiple parts that make up the effect

[serviscope_minor]

Basically, given there colours, there are two equally plausible interpretations: normal blue light or very strong yellow light. Depending on which you perceive the colours look gold/white or black/blue. It seems in this regard the brain won't go half way and is either all in one interpretation or the other.

I think it happens this way because the photo is genuinely ambiguous between two quite different interpretations and is exactly in a region where people differ, rather than everyone going one way (pedantically, colour is always ambiguous---try interpreting any colour under low pressure sodium lamps).

Re:The multiple parts that make up the effect

[SuperKendall]

I agree, the amazing thing to me is the same person can see it differently at different times (but very close together) given different cues before they see the picture.

As you say, all around it's right between two possibilities and the brain can take it one way or the other.

Re: Typical nerd

[DocHoncho]

Even worse, a clothes hanger!

Crazy People

[PPH]

Trying to evaluate colors based on a digital photograph, which may have been white balanced who knows how. And then uploaded through Instagram filtering algorithms. And finally viewed on various displays that have been tweaked (or mis-configured) to suit different users' tastes.

The xkcd cartoon illustrates one kind of optical illusion. But that's not what is going on here (or on Instagram/whatever). Because you can download two copies of the photo, one that appears white and gold, the other that appears blue and black. And you can actually verify, using various graphics tools, that the colors are actually different. It's not really a visual illusion produced by human perception. Its the result of massive post processing of a digital image.

Re:Crazy People

[itzly]

It's not really a visual illusion produced by human perception. Its the result of massive post processing of a digital image.

If it's not a human perception issue, how do you explain two groups of people in the same room, looking at the same picture, and strongly disagreeing about the colors ?

Re:Crazy People

[PPH]

I get what's going on now.

It may have started as a demonstration of poor color encoding, white balance and perception. But it quickly morphed into a demonstration of suggestibility.

There are (at least) three dress pictures out there. Not counting the re-balanced original. The first was of a blue and black dress, probably taken with a cheap camera phone. And that started it all off. The dress, way overexposed (mis-balanced) looked white and gold (tan, yellow, etc.) But the photographer knew it was supposed to be blue and black. So they posted this and started the discussion. Finally, some professional equipment was taken out, the dress photographed and proven beyond a shadow of a doubt that it's blue and black.

But there are at lest two other photos out there. A photoshopped copy of the original (thanks, /b/tards), with the colors changed to white and yellow. And a second dress with a very similar striped pattern, white and tan. But if you look closely, its not the same one and it has been cropped so as to eliminate background cues. Now, if you show these two pictures to people, some will still say "blue and black". But its not a matter of perception anymore. Thee two photos were never blue and black. Its all about suggestion. People 'know' that its blue and black because they've been told so. So that's the correct answer.

It's sort of like the Jimmy Kimmel Man on the Street prank interviews, where someone asks a passer-by a question about something that never happened. And they formulate an answer, just to sound well informed. And they might even believe that the incident in question actually happened when they walk away.

Perception of Colour before a Light or Dark Bkgrnd

[johnrpenner]

GOETHE explained this through the effect of turbidity on the perception of colour.

Light shining through a darker medium yields yellow; whereas an illuminated turbid medium before a dark background yields blue.

Check out: Light Darkness and Colours @time: 23:30 on youtube:
https://www.youtube.com/watch?...

This is remarkable

[LostMonk]

This is certainly a first . . . I don't believe there ever was a subject that lost my interest or got on my nerves that quickly.

You've got the color channel information

[g01d4]

Why the controversy regarding some bizarre conjunction of lighting and camera response? Match the color channel information with color test images and you're done.

Re:You've got the color channel information

[iggymanz]

Wrong answer, the "color" in question here is entirely a matter of human perception. Color channel information is irrelevant to issue.

Re:Stupid post

[serviscope_minor]

God, this, 1000 times this. This is a deeply fascinating topic (yes I am a computer vision person), and it's sad to see so many people crowing about how uninteresting it is because it's popular outside of nerd circles.

Re:Perception has nothing to do with it...

[Fallen Kell]

You have a shit calibrated monitor/display. The reason why most people see white+gold is because the majority of monitors have crappy color calibration, lumen balance, contrast, and white/black levels, especially "out of the box". My monitors are calibrated at the factory and come with custom color map for each monitor from the factory, so that they have less than 0.1dE2000 from sRGB.

This is why your iPhone 5 or 6 shows the image and it looked black+blue (they have "decent" color calibration of under 2.5dE2000, but that still is not even close to the 0.1dE2000 of a really good monitor), and most probably is still pushing way to many lumens for environment, which washes out the image (making it look white+gold).

Too much information is lost in the photo

[mark-t]

Sampling the pixels directly from the image reveals that the dress color in the photo has a hue between in or around the range of 230 or so, which happens to be blue. However, because it appears that the dress was not directly lit, that hue may be arising because of diffuse interreflections with its surroundings, something that anything which has a lighter shade can be very susceptible to if the only light hitting it is diffused, and which is the kind of lighting that this dress does appear to be exposed to in the photo.

So there is simply too much information about the surrounding lighting conditions that has been lost from the photo to ascertain with any certainty what the actual color of the dress is... at least from this one photograph alone.

Debating the matter is pointless, because it is impossible to actually arrive at a logically valid conclusion merely from what one can see in this photograph except through sheer guesswork.

W&G, was Re:White balance and contrast in came

[Anonymous Coward]

Here is a pretty good explanation of why this might happen.

Why is it that my mod points always expire right *before* I want to use them? I used eight of fifteen purely on posts where I only sort of wanted to mod. I had to give up commenting to do so. Here, I log in so as to mod -- and my mod points are gone. And I have no real interest in commenting!

Grrrr.

Anyway, regardless of the general quality (or lack thereof) of gizmodo, this was a decent explanation. It points out that in the picture, the colors are pale blue and dark gold. However, the original dress is a darker blue and black. The colors in the picture are incorrect. People who see it as blue and black are seeing past the problems with the picture while those who see white and gold are being fooled by the bad colors in the image.

Actual dress is the blue and black one on the left in this picture: http://media.gotraffic.net/ima...

Re:Green or Brown tie?

[Mes]

Had a similar experience with brown and green. Bought a beautiful chocolate brown couch for my office, but when it was brought indoors under the florescent lights, it was an ugly puke green. I was sure they delivered the wrong couch until I took a cushion outside and it was dark brown again.

It is not just the ambient light.

[Ecuador]

Me and my wife, as well as other couples see it in the same light but interpret it differently. I was sure it can't possibly be interpreted as blue/black, at most I would say it could be thought of as blue/bronze, but I asked my wife in the same room what color it is and she immediately said "blue/black of course". Neither of us could "see" it the other way. Until I opened an image someone made of how the white gold would look in a proper photo. So, after looking at this photo for a bit, I went back to the original and I could finally see it as black/blue!

Re: Typical nerd

[penguinoid]

Worse, it isn't even being occupied by a pretty girl

Was too distracted discussing the physics/psychology of the color to notice.

Wow

[Simulant]

Ya'll must be lonely.

It's painfully obvious from the comments that few people actually looked at the picture with along someone else. The camera doesn't matter. Nor does the quality of the picture. Two people looking at the same (admittedly crappy but that's beside the point) picture on the same device can see different colors. What you see may also change when you look again. I usually see white & gold but on two occasions it's been blue and black.

&is "teal" blue with greenish tinge or vice-ve

[Ungrounded Lightning]

... blue and brown. Just now, I opened the Washington Post link on my 24" screen in a sunlit room, and it was clearly white and gold.

Though the sensations are vastly different, brown is really dark yellow. The underlying color of that part of this dress seems to be very near the perceptual boundary (probably just on the yellow side of it). This picture seems to have the dress in a non-obvious shadow, so when it is viewed by someone whose visual system doesn't adequately pick up the shadowing and compensate, it crosses the boundary and appears light brown rather than dark yellow.

Another perceptual oddity is that a very slight bluish tinge to white makes it appear "whiter than white", especially in sunlight or other strong lighting. (I suspect this works by mimicing the differential response of the various color sensors in the eye when exposed to very bright light, though blue may also "cancel out" a bit of the yellowing of aging cloth.) Laundry products up through the 1950s or so included "bluing", a mild blue dye for producing the effect. (It fell out of use when it was replaced by a fluorescent dye that reradated energy from ultraviolet as blue, making the cloth literally "brighter than white" {where "white" is defined as diffuse reflection of 100% of the incoming light}, and which, if mixed with detergent products, would stick to the cloth while the surficant was rinsed away.) I suspect some of the "blueish is brighter" effect is going on here.

When I view the picture straight-on on my LCD display, the light cloth on the upper part of the dress appears about white and the image appears somewhat washed out. Meanwhile the lower half has a bluish tinge. So I suspect the cloth is actually nearly-white with a bit of blue. (Viewed off-axis it's very blue, but the other colors are over-saturated and/or otherwise visibly off-color. So off-axis viewing makes it look more blue and this probably adds to the controversy.)

Another color-perception issue is "teal", a color between blue and green. There are paint formulations of this color that give the sensation of "distinctly blue with a greenish tinge" to some people and "distinctly green with a bluish tinge" to others, even under the same lighting and viewed from the same angle. (I'm in the "slightly-bluish-green" camp.)

The first place I encountered this was on the guitar of the filksinger Clif Flint. (On which he played _Unreality Warp_: "... I'm being followed by maroon shadows ..." B-) ) Apparently his fans occasionally had arguments about whether his guitar was blue or green, so he sometimes headed this off (or started it off on a more friendly levl) by commenting on the effect.

no, SAME image perceived two different ways

[iggymanz]

Group of people looking at the SAME picture have one of two different views of the color. Half the people in my family saw white and gold (myself included), half blue-black. In fact, if I have that image in corner of my vision it is black-purple, but goes to white - gold after only moving head twenty degrees toward image.

It's very much how human brain perceives color based on (simulated in this case) ambient lighting; this picture is on the dividing line so to speak

It's not blue&black and neither white&gold

[koinu]

Please learn to use your eyes properly.

Purkinje Effect

[NimbleSquirrel]

There are several factors that make this unintentional optical illusion really interesting.

The first, demonstrated by the xkcd, shows that the colours will appear markedly different with different coloured backgrounds. It doesn't fully explain what we are seeing here though, as people are seeing the two different states with the same background to the dress.

I believe that the main illusion comes down to the Purkinje Effect, and how our brains interpret colour. Under the Purkinje Effect, in lower light levels our peak visual sensitivity shifts to the blue end of the spectrum. At higher levels it shifts away from the blue end of the spectrum as the rod cells in our eyes reach a point of saturation and stop being effective. So, when ambient light is bright enough, we just don't perceive blues as well, and we just don't see differences in contrast as well (as the rod cells are responsible for contrast vision).

If your eyes are adapted to bright light conditions (and the threshold here varys from person to person), you will likely see white and gold. Due to the shift away from blue, dark greys in the image appear more yellow. The blue also becomes apparently lighter to the point that our brains interpret it as a white dress in the shadow in daylight. If you go into a darker environment and wait (it takes about 10-15 minutes for the rhodopsin in the rod cells to regenerate), you willl likely see the blue and dark grey/black.

The first time I saw this, I had just walked in from outside. I saw white and gold. Then after sitting in a darkened workshop for 30 minutes, I saw it as blue and dark grey.

You are not alone

[the grace of R'hllor]

I fail to see it as anything other than blue and gold (well, brown/yellow that might be called gold)

Poorly-posed questions, not color perception

[Anonymous Coward]

In fact, my first reaction was "the dress is obviously white and gold in a terrible photograph, what kind of idiot would say it is blue and black?" I saw people attribute it to various effects that really cannot account for such a huge disparity, concluded they are all crazy, thought this was an interesting subject that I should post on, and then did research, which lead me to increasingly suspect that extremely few people actually initially believe that the image looks anything like royal blue and black, but they either read Wired's analysis of problems with the color balance (unlikely), or saw the designer's stock image which has good lighting and clearly shows the dress to be blue and black (likely), and then either disingenuously claim that they thought it looked blue and black all along (because the dress the original image has a slight blue tint), or they simply say "it's blue and black, duh!" without explaining that the "duh" is "searching for a better photo." And at the same time people are shouting "it's light blue-grey, I can tell by the pixels, dumbasses!" and others are construing that to support their position of "white" (of course! it's a bluish photo of a white dress!) or "blue" (exactly--the pixels are blue, because it's a blue dress!). Few people know what question is even really being discussed, and those that completely explain their answer (Wired) are either ignored or skimmed by everyone else, whether readers or bloggers.

The blog article in TFA, which is actually under the banner of the Washington Post, walks straight into this fallacy. Presumably the other media commentators are as well, or the WaPo blogger would have noticed. They see the furious difference of opinion on the twittersphere, or whatever crap they're using for "research," and then indiscriminately repeat it all, along with some people smugly pointing out how everyone is getting the wrong answer (to a question that they haven't bothered to define). The implicit conclusion is that the image has some mystical property that makes people deranged, though most of them are too stupid to realize it (even as those people are thinking something similar), until the WaPo blogger finally badly quotes a Wired blogger who actually figured out most of the truth. They realized that the common conclusions about what is wrong with the white balance are inconsistent over the whole image, and that if they balanced it from assumption that the darkest point should actually be black, the dress surprisingly turns blue and black. The WaPo blogger ignores most of the subtlety, because like everyone else, they just want to say "the dress is actually [blue and black] because people are too dumb to account for [the brain]" instead of "the color balance of this photograph is skewed in a very unexpected way that combined with its obvious background overexposure, leads most people to guess that it is a white/gold dress in underexposed overly-blue lighting--a very common white balance problem--instead of a blue and black dress in extremely overexposed lighting with disproportionate red saturation, which we were only able to realize after repeated filtering attempts." The real answer is apparently too long to be supported by the bulk of the combined twitter-/blogo-sphere, and so the "controversy" continues.

The many degenerate properties of the infamous photo are interesting, but mostly this "controversy" seems to be a fairly common, if unusually clear illustration of the problems with human mass-communication. Everyone gets a little information, leaps to a conclusion, sees that people disagree, and starts generating rationalizations to explain how the bulk of humanity is morons, except for them. Well, extremely few people are actually "morons," but a whole lot of people really are terrible at both receiving facts from communications and in turn, explaining them in their own communications. In f

YOU INSENSITIVE CLODS!

[occasional_dabbler]

I'm colourblind

Although the model sure does look hot in that clingy grey and slightly greyer dress ;-)

Just making shots here...

[kefalonia]

The eye pupil is known to exhibit interesting behaviour at times,
http://en.wikipedia.org/wiki/P...
one notable being photic reflex (which also affects a quarter of a population)
http://en.wikipedia.org/wiki/P...

IMHO, human vision is still incompletely understood at whole population (global) level,
with all sorts of exceptions and special trade-off cases being documented:
http://en.wikipedia.org/wiki/T...
http://discovermagazine.com/20... ### check this one!

Finally, let's not forget, that it is well known that manly colour vocabulary is 4-bit, while females have true colour sets ;-O
http://www.smithsonianmag.com/...
http://io9.com/5919311/some-wo...
https://www.google.be/search?r...

Last but not least: make sure you see the image of the OP in fractional ways (say, top 10th of the image),
along with another person that sees it in the alternative mode. You may come up with surprises. ;-)

Pixel Colours - Light Blue&Gold

[wylderide]

I can only see it as it appears in the image: as the pixel colours. http://knowyourmeme.com/photos...

Not a good comic today

[Trogre]

That has to be one of the worst xkcd comics I've seen in a long time. The left picture shows a white and gold dress against a blue backdrop, and the right picture shows a blue and gold dress against a yellow backdrop. In neither picture does the "gold" look remotely like anything that could be called black.

I think a lot of this confusion is coming from the fact that the white balance of the picture is such that the blue fabric looks like evening light scattering off a white surface (a very light blue), so our eyes are interpreting that as the "white" point, and correcting everything else in the picture to match. So we have a way overexposed very dark object made to look like a slightly underexposed light object.

Blue Brown

[fox171171]

I can only see it as blue, period. Not trolling - I really cannot see this as white in any circumstances, even the XKCD "color balanced" bit I still see it as blue (albeit a much lighter blue on the left).

When I looked at the image, I saw a bluish and brownish dress.

In the XKCD image, I can say that they are both blue in color, with the dark room one being lighter blue in appearance to my eye, yet I know my interpretation would be more that it was in fact a white dress in a darkly lit area.

I have sunglasses that are brown tinted, and technically everything I see with them on has a brown tint to it, yet I know my brain is ignoring this (unless I am really thinking about it) and perceiving the colors for what I know they are. Your brain does an incredible amount of processing and interpretation of the things you see.

On reviewing the image again, I can see the over exposed background which does suggest the exposure of the dress itself is darkened, and therefore is white, or at least much lighter than it appears in the image.

I am reminded of this as well:
http://en.wikipedia.org/wiki/Checker_shadow_illusion

What colour is the dress?

[mgcarley]

And the answer to this $64 question is: 50 shades of who gives a fuck.

The Emperor's New Clothes

[rkinch]

I see no dress at all. Check out the Emperor's New Clothes.

Here's a good read

[PJ6]

in case you're interested in learning more about color perception than the links provide.

http://blog.asmartbear.com/color-wheels.html

If you like that, there's even more to read on color theory.

Amazing!

[Gallomimia]

Something has actually gotten the sheeple interested in science!

This makes the news / slashdot??

[MoarSauce123]

Geez...the world doesn't have anything better to debate over? I think the dress looks hideous no matter what the color. There ya have it!

The trick is to remove the context...

[rgbatduke]

What color the dress is is impossible to tell from a photograph for countless reasons. What color the pixels in the photograph are is beyond question. They are light blue. One can simply box everything but a small patch of color from the dress and out of any context at all it is not white. If one has any real doubt, one can always go into the image itself and look at the RGB of the image.

The dress itself could be white, could be blue, could be grey -- and reflecting light from some blue source (like the sky, like a blue wall in the background behind the photographer). One would have to be there to know, since there are no other foreground objects to use to normalize our beliefs. But the pixels -- the pixels are what they are, and it ain't white or any of the nearly balanced fifty shades of grey.

rgb

better question... is the dress on or off?

[swschrad]

One Genuine Internet Point, redeemable wherever Genuine Internet Points are honored, for the correct answer.

now everybody move on with their life.

Re:It's a distraction from a real story

[tehcyder]

Whenever a talking point like this hits every mainstream media outlet at the same time, now apparently Slashdot too, then I find myself wondering what poltical blunders are they trying to distract us from right now?

Yes, I too am utterly unable to think about more than one thing a day. There have been times when I followed a cute kitty link to other cute kitties literally did no work for a week, and once a particularly amusing Chuck Norris meme caused me to miss the entire Arab Spring.

I know my white sheets under blue light

[epine]

It appears to me as blue and black. Definitively. First viewed as a whole image, full screen on a tall portrait IPS and then checked on a second landscape IPS (these two screens long ago adjusted to show matched colours).

Both the black and blue appear to me somewhat blown out. Actually, the mottled black almost appears as a mutant non-colour unlike anything one sees in real life (the colour balance algorithm of a digital camera subtracting blue from the black is a perfect explanation for this).

If I scroll so that I can only see the top of the dress (down to just past the horizontal black band across her upper back) I can almost conceive of how some people see this dress as white and gold. What I actually perceive is an ambiguous image under false, untrustworthy light.

In my bedroom I have several unusual light sources which I regularly use. In addition to an incandescent lamp, there an extremely yellow bug lamp and a bright and narrow-spectrum blue LED light intended to shift circadian rhythm.

I love the yellow bug lamp because it's initially so dim I can turn it on briefly while my wife sleeps to find my socks, plus I often use it for reading late in my day so as not to expose myself to blue light. I also had red and green light sources for a while, before I discovered yellow bug-light perfection (the red and green bulbs were 40 W coloured-glass bulbs that constantly smelled bad because they instantly baked any stray dust—a failed experiment).

I have pretty good sense in my bedroom of which colours are more or less trustworthy under vastly different lighting conditions. Even under my narrow-spectrum blue light (in an otherwise dark bedroom) I can't make anything white look like this photograph. In my bedroom under a pure blue illumination (75% between 450 and 480 nm, centered at 464 nm; with 490 nm attenuated by 10 dB compared to the spectral peak) the highlights on my white sheets where the light is strongest are more saturated, and the dimmer regions are less saturated, opposite my impression of this photograph.

Perhaps people who spend a lot of time watching TV in dark rooms where people are wearing white clothing are conditioned differently.