5 ms·

In my other life I'm an artist, with a lot of training and experience in color theory and painting. I'm honestly a bit surprised that this whole thing is confu

by thraxil 12y ago

In my other life I'm an artist, with a lot of training and experience in color theory and painting. I'm honestly a bit surprised that this whole thing is confusing and dividing so many people. This seems like a straightforward demonstration of Color Constancy (http://en.wikipedia.org/wiki/Color_constancy http://en.wikipedia.org/wiki/Color_constancy) and the distinction between local color and perceived color (http://practicalpainter.blogspot.nl/2010/05/local-versus-perceived-color.html http://practicalpainter.blogspot.nl/2010/05/local-versus-per...). I had thought that this kind of thing was more widely understood outside art education, but perhaps I'm wrong. The short version is that the dress could be either and the photo doesn't really give enough context to say anything definitive. The longer version is that all color perception is relative. Light of a particular makeup of colors hits an object, it absorbs and reflects different amounts at different frequencies, and that's what hits your retina. Our brains then attempt to "reverse engineer" the "local color" of the object by comparing it to the surrounding light (presumably because local color is more useful for identifying and discriminating objects), and that's what we "perceive". In normal conditions we do this really well. But when you strip out the context, our brains keep trying but are easily fooled. James Gurney tends to do an excellent job dissecting this phenomenon: * http://gurneyjourney.blogspot.nl/2010/01/color-constancy.html * http://gurneyjourney.blogspot.nl/2012/11/yellow.html * http://www.amazon.com/Color-Light-Guide-Realist-Painter/dp/0740797719/

bjornsing 12y ago
I completely agree that's what's going on. But what's fascinating here (to me at least) is how individual our interpretations are. Can you see what assumptions the gold/white crowd is making in their interpretation? Even with the slider in the OP I can't see it. :) And even worse: I can't understand what assumption I need to make (about the color of objects in the background or shadows or something) to be able to see a gold/white dress. Phrased differently: I can't imagine how I would construct a scene with a gold/white dress and then take a picture that looks like the one in the OP (regardless of the position of the slider). Scary! :)
- lucio 12y ago
  do you really see "black" here ? https://cdn0.vox-cdn.com/uploads/chorus_asset/file/3452082/no_black_dress.0.png https://cdn0.vox-cdn.com/uploads/chorus_asset/file/3452082/n... Objectively, there's only "yellow-brown" (gold) and "light blue" which most people "mentally" correct to "white"
  - smoe 12y ago
    I'm one of the light blue/gold people and couldn't image seeing black anywhere. But when I look at the picture upside down I see it as blue/gold and when I mask the lower half, as blue/black with a gold touch. http://s4.photobucket.com/user/ometauru/media/YNOPKWp_zpstcvt2hqd.png.html http://s4.photobucket.com/user/ometauru/media/YNOPKWp_zpstcv... Looking at the different things around and behind the dress, everything dark has a similar golden color. Maybe some people correct that to black in the same way most do with the light blue to white
    - bjornsing 12y ago
      I see blue/black, but blue/gold makes sense to me too. It's more white/gold that seems improbable to me.
    - Squarel 12y ago
      I am the same, I cant see either white/gold or blue/black, and I am surprised that the options for most people seem to be white/gold or blue/black. light blue/gold barely gets a mention in most articles.
  - quarterto 12y ago
    local colour vs perceived colour LOCAL COLOUR VS PERCEIVED COLOUR LOCAL COLOUR VS PERCEIVED COLOUR nobody is arguing that the perceived colour of the dress is black.
  - bjornsing 12y ago
    It's easier to understand the gold part than the white part to me. But I think I assume that the goldish hue comes more from the light, since the further down on the dress I look the more black it seems. Phrased slightly differently: The yellow-brown (gold) color is most visible on the upper torso where the fabric (to me) is likely to reflect a light source specularly. On the lower part of the dress (where I assume) light is more ambiantly reflected I see black. Looking at your color squares to the right it's quite clear to me what my visual cortex is doing: It's assuming that specular reflection expains the difference in color in the left column. So each color there is a superposition of two: the ambient color of the dress and the specular reflection of the light source. Separating those two gives me black dress and yellowish lighting. UPDATE After writing the above I think I understand why it seems so weird to me to see white: I parse the light source as yellowish, so a white dress would never look blue in that light. To me seeing a white/gold dress requires contradictory hypothesis: the light source needs to be blue to explain bluish tint on white stripes; but it also needs to be yellowish to explain the "gradient" in gold/black stripes from top to bottom of the dress.
pfortuny 12y ago
Just a silly anecdote but a friend of mine had a room painted "this exact same pearl grey" used in another room. After having it grey, he realized it did not look grey but blue (I know why it was, because the only window of that room got the light reflected from a large bronze-green roof). He had it repainted a different tint. That was amazing.
MatthewWilkes 12y ago
Once I got to the word "Constancy" I started reading this in Reverend Lovejoy's voice.
- soylentcola 12y ago
  Sweet constancy...
nilkn 12y ago
My perception of the dress's color has flipped from white/gold to blue/black and now back to white/gold again over the course of the last day. I can say that, in my case, there's been a fairly strong correlation between my own ambient lighting conditions. I've only seen white/gold in well-lit settings and I've only seen blue/black when I was laying in bed with all the lights off. Whether that's coincidence or not I cannot say, but I felt it was an interesting data point.
logfromblammo 12y ago
I think it's reasonable that we would have a default light-color model for illumination by natural sunlight, and that our brains would use that model to subtract the color of natural direct and diffuse sunlight from an objective color measurement upon the retina to determine a perceived inherent color for an object, which is then held as a constant. This would allow the same rock or tree to be recognized as a landmark at both noon and civil twilight. Since the invention of artificial illumination (starting with fire), humans have had access to multiple light sources, which may be of different colors and coming from different directions. The default illumination model is insufficient to handle this without ambiguity. This phenomenon is also used for dramatic effect in The Abyss, wherein the only light source available in the bottom of an oceanic trench is a monochromatic glowstick, and to disarm the nuclear bomb, Bud has to cut the blue wire with a white stripe and NOT the black wire with a yellow stripe. The actual wires used in filming were both black with a white stripe.
jgehrcke 12y ago
I really agree with the essentials of what you write: - "surprised that this whole thing is confusing and dividing so many people" - "the dress could be either and the photo doesn't really give enough context to say anything definitive" I have used the same two arguments to explain why this hype appeared in the first place, and why the "scientific" analysis many people try to follow is conceptually wrong. Blog post and corresponding discussion can be found here: https://news.ycombinator.com/item?id=9120940 https://news.ycombinator.com/item?id=9120940
ohhko 12y ago
Hey, I'm currently researching color theory. If you don't mind: Except Realist painter by Gurney, what are some other great resources about color theory that you would recommend? My main paper will be about color in marketing. Cheers, and good to see another scholar of colors!

by thraxil 12y ago

bjornsing 12y ago

I completely agree that's what's going on. But what's fascinating here (to me at least) is how individual our interpretations are. Can you see what assumptions the gold/white crowd is making in their interpretation? Even with the slider in the OP I can't see it. :) And even worse: I can't understand what assumption I need to make (about the color of objects in the background or shadows or something) to be able to see a gold/white dress. Phrased differently: I can't imagine how I would construct a scene with a gold/white dress and then take a picture that looks like the one in the OP (regardless of the position of the slider). Scary! :)

lucio 12y ago
do you really see "black" here ? https://cdn0.vox-cdn.com/uploads/chorus_asset/file/3452082/no_black_dress.0.png https://cdn0.vox-cdn.com/uploads/chorus_asset/file/3452082/n... Objectively, there's only "yellow-brown" (gold) and "light blue" which most people "mentally" correct to "white"
- smoe 12y ago
  I'm one of the light blue/gold people and couldn't image seeing black anywhere. But when I look at the picture upside down I see it as blue/gold and when I mask the lower half, as blue/black with a gold touch. http://s4.photobucket.com/user/ometauru/media/YNOPKWp_zpstcvt2hqd.png.html http://s4.photobucket.com/user/ometauru/media/YNOPKWp_zpstcv... Looking at the different things around and behind the dress, everything dark has a similar golden color. Maybe some people correct that to black in the same way most do with the light blue to white
  - bjornsing 12y ago
    I see blue/black, but blue/gold makes sense to me too. It's more white/gold that seems improbable to me.
  - Squarel 12y ago
    I am the same, I cant see either white/gold or blue/black, and I am surprised that the options for most people seem to be white/gold or blue/black. light blue/gold barely gets a mention in most articles.
- quarterto 12y ago
  local colour vs perceived colour LOCAL COLOUR VS PERCEIVED COLOUR LOCAL COLOUR VS PERCEIVED COLOUR nobody is arguing that the perceived colour of the dress is black.
- bjornsing 12y ago
  It's easier to understand the gold part than the white part to me. But I think I assume that the goldish hue comes more from the light, since the further down on the dress I look the more black it seems. Phrased slightly differently: The yellow-brown (gold) color is most visible on the upper torso where the fabric (to me) is likely to reflect a light source specularly. On the lower part of the dress (where I assume) light is more ambiantly reflected I see black. Looking at your color squares to the right it's quite clear to me what my visual cortex is doing: It's assuming that specular reflection expains the difference in color in the left column. So each color there is a superposition of two: the ambient color of the dress and the specular reflection of the light source. Separating those two gives me black dress and yellowish lighting. UPDATE After writing the above I think I understand why it seems so weird to me to see white: I parse the light source as yellowish, so a white dress would never look blue in that light. To me seeing a white/gold dress requires contradictory hypothesis: the light source needs to be blue to explain bluish tint on white stripes; but it also needs to be yellowish to explain the "gradient" in gold/black stripes from top to bottom of the dress.

pfortuny 12y ago

Just a silly anecdote but a friend of mine had a room painted "this exact same pearl grey" used in another room. After having it grey, he realized it did not look grey but blue (I know why it was, because the only window of that room got the light reflected from a large bronze-green roof). He had it repainted a different tint. That was amazing.

MatthewWilkes 12y ago

Once I got to the word "Constancy" I started reading this in Reverend Lovejoy's voice.

soylentcola 12y ago
Sweet constancy...

nilkn 12y ago

My perception of the dress's color has flipped from white/gold to blue/black and now back to white/gold again over the course of the last day. I can say that, in my case, there's been a fairly strong correlation between my own ambient lighting conditions. I've only seen white/gold in well-lit settings and I've only seen blue/black when I was laying in bed with all the lights off. Whether that's coincidence or not I cannot say, but I felt it was an interesting data point.

logfromblammo 12y ago

I think it's reasonable that we would have a default light-color model for illumination by natural sunlight, and that our brains would use that model to subtract the color of natural direct and diffuse sunlight from an objective color measurement upon the retina to determine a perceived inherent color for an object, which is then held as a constant. This would allow the same rock or tree to be recognized as a landmark at both noon and civil twilight. Since the invention of artificial illumination (starting with fire), humans have had access to multiple light sources, which may be of different colors and coming from different directions. The default illumination model is insufficient to handle this without ambiguity. This phenomenon is also used for dramatic effect in The Abyss, wherein the only light source available in the bottom of an oceanic trench is a monochromatic glowstick, and to disarm the nuclear bomb, Bud has to cut the blue wire with a white stripe and NOT the black wire with a yellow stripe. The actual wires used in filming were both black with a white stripe.

jgehrcke 12y ago

I really agree with the essentials of what you write: - "surprised that this whole thing is confusing and dividing so many people" - "the dress could be either and the photo doesn't really give enough context to say anything definitive" I have used the same two arguments to explain why this hype appeared in the first place, and why the "scientific" analysis many people try to follow is conceptually wrong. Blog post and corresponding discussion can be found here: https://news.ycombinator.com/item?id=9120940 https://news.ycombinator.com/item?id=9120940

ohhko 12y ago

Hey, I'm currently researching color theory. If you don't mind: Except Realist painter by Gurney, what are some other great resources about color theory that you would recommend? My main paper will be about color in marketing. Cheers, and good to see another scholar of colors!