Due to the dress's particular color scheme and the exposure of the photo, it forms an optical illusion causing viewers to disagree on what color the dress actually seems to be. The xkcd strip sandwiches a cropped segment of the photographed dress between two drawings which use the colors from the image against different backgrounds, leading the eye to interpret the white balance differently, demonstrating how the dress can appear different colors depending on context and the viewer's previous experiences.
Below is an illustration demonstrating that the "colors" of the dresses are the same by connecting them with two lines with the above mentioned colors (all the way!):
This image has sparked surprisingly heated debate in many internet communities. A select few individuals may have prior experience with optical illusions of this ilk, but because this particular image went viral -- it got heavy exposure over such a short amount of time -- it reached millions of people who aren't so familiar with these sorts of mind tricks. To the uninitiated, the color of the dress seems immediately obvious; when others cannot see it their way, it can be a surreal (even uncomfortable) experience.
Randall is presumably pointing out how ridiculous it is for people who don't understand the underlying science to become so adamant in defending their beliefs. A spoof of the "Monty Hall problem" previously appeared in 1282: Monty Hall, where Beret Guy decides to take the goat.
To me, they both look blue/gold Mikemk (talk) 06:29, 27 February 2015 (UTC)
- To me as well. The one on the right, with a lighter background, appears more bluish and the collar is a darker brown. (The collar on the left, to my eyes, matches the face on the right.) But both definitely appear bluish with a dull yellow. - Equinox 188.8.131.52 16:50, 27 February 2015 (UTC)
What is the illusion supposed to be? The colors of the dress look a bit darker with the light background, but not very much. Is that the illusion? --184.108.40.206 07:07, 27 February 2015 (UTC)
- Agree. To me, it looks like it's definitely light blue (maybe "cornflower"?) with pale olive stripes. "Gold" would really be a stretch. It looks like that in all lighting conditions and in both backgrounds of the strip. Did I pass some kind of color-blindness test? Or fail? 220.127.116.11 07:43, 27 February 2015 (UTC)
- This has nothing to do with color-blindness, but probably with certain arbitrary constants related to white-balance adjustment that differ brain-to-brain. Many people I know insist that even though the picture looks blue, it's a dress illuminated by a blue light, and based on this assumption their brain may essentially redden the whole picture to adjust for this light. The actual picture was taken in white light, not blue light. 18.104.22.168 07:46, 27 February 2015 (UTC)
- It may also be related to white-balance of the MONITOR. I see original dress like black and blue and the one on left here as gold and light blue. -- Hkmaly (talk) 10:00, 27 February 2015 (UTC)
Apparently for some people the left-hand-side's general blueishness is adjusted against by the visual system enough to make the dress look white and gold instead of blue and brown. I am not one of those people. 22.214.171.124 07:43, 27 February 2015 (UTC)
Description says left for both 126.96.36.199 08:37, 27 February 2015 (UTC)
- Now changed. (Saw it myself before I saw your comment, and just lept straight in there. Hopefully I changed the right left so that it's right and not left the wrong left whilst producing the wrong right. Alright?) 188.8.131.52 09:30, 27 February 2015 (UTC) (Also, "hello near-IP neighbour!"... The same digits, even. Creepy.)
- Y'all are both from London, and probably live on the same street. Congrats! You made a friend! :) 184.108.40.206 16:25, 27 February 2015 (UTC)
Are they really the same colour? 'Cause to me on the blue side it looks blue and black- while on the white side it looks white and gold. Is this normal? -- FlyingPiggy (talk) (please sign your comments with ~~~~)
The figure on the right definitely has a beard. 220.127.116.11 09:38, 27 February 2015 (UTC)
I checked with ColorZilla and the RGB values are identical. From my perspective, in the one on the left the dress appears pale blue with darker brown/gold stripes, and the one on the right appears a darker blue with lighter brown/gold stripes. 18.104.22.168 10:10, 27 February 2015 (UTC)
This is just a polychromatic version of that checker shadow illusion, right? http://en.wikipedia.org/wiki/Checker_shadow_illusion 22.214.171.124 10:12, 27 February 2015 (UTC)
- That's what I thought too. But it looks the same (doesn't it?) and is the same (that, thankfully is non-subjective and verifiable with as little as MSPaint), so I'm at loss as to why this deserves a comic. 126.96.36.199 10:47, 27 February 2015 (UTC)
- This is a common optical illusion (at least I've seen this many times) - most peoples eyes perform a white balance adjustment automatically which affects the perceived colours. If your eyes don't do this then you will do well in the paint colour matching business. http://www.moillusions.com/hue-optical-illusion/ I apologise for the jarring colours in the link. 188.8.131.52 (talk) (please sign your comments with ~~~~)
- Here's a particularly good demonstration of the underlying "color perception" illusion (i.e. the Checker-Shadow illusion referred to above): https://www.youtube.com/watch?v=z9Sen1HTu5o Arcanechili (talk) 15:45, 27 February 2015 (UTC)
- Confused guy again. It seems I can only 'accurately' see shades of red (or is it blue? anyway, the first illusion in the Hue article, but not the other two) fascinating as that is, I'm probably not getting the paint job.
184.108.40.206 11:52, 4 March 2015 (UTC)
The comic is a reference to the debate around the coloration of this dress. The band in the middle of the image shows some of the material of the dress. To some people, including me, the dress is obviously, unquestionably black and blue. But to others, including my wife, it's obviously, unquestionably, black and gold. 220.127.116.11 (talk) (please sign your comments with ~~~~)
- And to others it's apparently a number of other combinations - I've seen claims of white/gold and blue/orange. However, surprisingly few people seem to have seen this link to the manufacturer's page for what appears to be the same dress; available in 4 colour combinations which according to the manufacturers' descriptions are ivory/black, scarlet/black, pink/black and royal-blue/black, with pictures available of all versions. As such I'm happy to accept the pictures doing the rounds are probably the blue/black variant (although most of the over-exposed versions I've seen appear light-blue/goldish-brown to me. 18.104.22.168 (talk) (please sign your comments with ~~~~)
Our eyes are too efficient, which makes this illusion work. In dim light we dilate our eyes, so an enclosed room with one lamp seems bright, though it is a cave compared to the outdoors. If the bulb in our lamp is of a warm tone, our eyes adjust so we believe we see colours as though in daylight. I think that's what's happening in the dress illusion -- we are trying to allow for perceived lighting conditions in the photo -- so the actual illusion is in our guess as to what those light conditions actually are. And finally an artist quote: "I can paint you the skin of Venus with mud, provided you let me surround it as I will." - Eugene Delacroix 22.214.171.124 13:28, 27 February 2015 (UTC)
Some Notes on camera color correction: it's worse than, and is not just an optical illusion. It's a camera screwup.
Cheap cameras will try to adjust colors based on formulas that guess what the correct color scheme is.
If you take a photo while in the shade on a sunny day, you likely get most of your light from the bright *BLUE* sky. This can make you look awful. The camera is set up to guess the correct exposure. In this case the camera follows the rules, and guesses, wrongly, that the the overall majority color in the center of the photo is white, and transforms the rest of the colors to match
It's a camera screwup. It also depends on how bad your viewing device is behaving, because, based on how dark the screen is, you then get the optical illusion effect that Randall posted.
This is a secondary effect, and not the real reason why behind what is going on in the first place. The correct rendering of the camera screwup is going to be, on most devices with normal color rendition, white with gold. Because some monitors are lighter or darker depending on viewing angle, this also impacts color perception.
We can then get the actual optical illusion after all that. But as we have seen with good photos of the actual dress in normal light, the camera got it wrong.
TL;DR: It's a cell phone camera screwup. 126.96.36.199 14:22, 27 February 2015 (UTC)
- Thank you so much! That is the only good explanation I've read. The colour correction introduced by the cellphone is massive. But I still don't know how anyone can see the gold as being black unless their computer monitor is completely maladjusted. 188.8.131.52 04:30, 28 February 2015 (UTC)
- The first time I saw the alleged image of the same dress by the manufacturer (which is clearly to me black and blue), I couldn't believe they where the same dresses, because, by looking at the viral image, I can't imagine a lightning scenario and brain correction that would cause black to look gold. The camera adjustment screw up makes sense to make them the same dress. Therefore, I hereby challenge someone to buy the black and blue dress and take a photo of it, doing whatever necessary to make the black look gold. 184.108.40.206 22:40, 10 December 2015 (UTC)
Is it possible that there's a connection between this comic and 690: Semicontrolled Demolition? Some discussions I've seen about this topic involve the choice between white/gold and blue/black, so Randall coloured his dress gold/blue. --220.127.116.11 16:20, 27 February 2015 (UTC)
- It's more likely due to the fact that the colors of the actual picture - that is, if you use a color picker - are roughly the same as those in the comic. 18.104.22.168 17:06, 27 February 2015 (UTC)
I'm reminded of 356: Nerd Sniping, only the perpetrator has managed to snipe the entire interwebs 22.214.171.124 (talk) (please sign your comments with ~~~~)
No one has yet mentioned that ignoring color, the two images are also mirror images of each other. On the left, the figure is looking slightly toward her own right shoulder; while on the right she is looking slightly to her own left. Most likely not at all relevant to this discussion, but usually folks on this forum are very quick to point out even insignificant details (like I'm doing right now) :)
126.96.36.199 19:01, 27 February 2015 (UTC)
- Similar Illusions and explanation
I feel like this line seems out o place, or at least badly worded (using half the URL as the text...):
"Similar types of illusions can be seen at Optical illusion#Color_and_brightness_constancies and at echalk. (requires Flash®player)."
Also, this seems similar to the checker shadow illusion (link to page on website with explanation of said illusion).
On another note, this page seems rather disorganized and uniformative about the phenomenon behind this illusion.
Zweisteine (talk) 19:04, 27 February 2015 (UTC)
This explanation is lacking in a particular detail: in Randall's drawing, what colors do the dresses look. I'm color blind, so they don't look different to me, but I couldn't name them. 188.8.131.52 00:13, 28 February 2015 (UTC)
- I guess it depends. For ME it is blue and gold. --DaB. (talk) 01:57, 28 February 2015 (UTC)
LOOK, Randall. Schrödinger was a crossdresser. This was Schrödinger's dress. My Schrödinger fanfics say so and you can't take away my dreams! 184.108.40.206 02:15, 28 February 2015 (UTC)
Wow. This is amazing. Starting out, I see only blue and gold in both sides of the comic, the photo strip down the center, and the photo it's based on. I see only slight differences in color among all these images. Just variations in brightness. Has anyone else noticed that the colors Randall has chosen are almost precisely complementary? Adding the R, G, and B values from each color cited in the explanation gets you 248 R, 248 G, and 247 B. The values in the center strip vary alot but I have found examples of each color that match within 2 bits in an 8 bit RGB system. I wonder how and why Randall chose these values. NewScientist has an article on this too. Trivia, re Monty Hall discussion: In the 60's muscle car era the Pontiac GTO was nicknamed "The Goat" so maybe the car IS the goat! ExternalMonolog (talk) 02:44, 28 February 2015 (UTC)
So, does this have anything to do with Christopher Columbus? Mikemk (talk) 03:14, 1 March 2015 (UTC)
The current version of the transcript says that on the right Megan is "yellow against a buttercup background." I think I'm going to change that to "brown" and "white." (If we want to get super nerdy about it, then name-of-color.com says the actual colors are luxor gold, equator and old lace.) Pesthouse (talk) 02:29, 2 March 2015 (UTC)
The weird part is that I see no difference between the dresses. The same blue brown, but if I stare long enough, right side looks darker. Despite that, you cannot convince me that checker B is the same colour as A in the checker colour illusion.
220.127.116.11 06:43, 3 March 2015 (UTC)
Okok.... in the image, the dress is technically light blue and dark gold. HOWEVER, it is obvious that the real dress could not be this color, because the lighting is weird and you can tell that it is really white and gold. Also, that is a more reasonable color for a dress anyway... QED 18.104.22.168 22:29, 25 May 2015 (UTC)
I actually saw it just as it was (blue and brown). I can't understand why anybody else saw it differently. 22.214.171.124 09:28, 22 July 2015 (UTC)
The regular 'dress' looks white and gold to me, but this looks blue and brown. Aargh! -- Jadzia.dax (talk) (please sign your comments with ~~~~)
It may seem petty, but this site usually points out small errors Randall makes and I think there is one in the mouse over text. I'm not native in English, so it's likely I'm wrong. But »only to find there was unexpectedly disagreement over whether the thing they'd revealed was a goat or a car« doesn't look right to me. I'd want it to be either »»only to find there was unexpected disagreement over whether the thing they'd revealed was a goat or a car« or »only to find there unexpectedly was disagreement over whether the thing they'd revealed was a goat or a car«. --126.96.36.199 17:04, 29 November 2017 (UTC)
- I'm not a native English speaker either, and I also take a little longer to understand that phrase because of the way it's worded. Your first suggestion looks better. The second one, though, not as much. GuiRitter (talk) 17:35, 29 November 2017 (UTC)