The Dress

[Aaroh]

Its white with gold stripes. It depends on the brightness

 

[ProjectWHaT]

Here's the most important question:

Is it black with blue stripes or blue with black stripes???

 

[Aaroh]

When its bright, that is with the light on in the living room, its white with gold stripes. When in a dark room, I think its blue with black stripes

 

[FT-1]

Here's the most important question:

Is it black with blue stripes or blue with black stripes???

Looks like it is primarily blue with a black lace design on it.

 

[GranTurismo916]

Here's the most important question:

Is it black with blue stripes or blue with black stripes???

Blue with black stripes, officially. That's what the actual dress is.

 

[WhiterunGuard]

Personally I have no clue why the f I clicked this page, but I see dark brown and light blue, I'm pretty sure this due to overexposure as has been stated previously...or is it all in our heads...maybe there is no photo at all!

 

[Wolfe]

Never saw the white/gold except in the xkcd comic, pretty convincing:

After all this, I think I'm even more concerned by so many people interpreting a bright and overexposed photo as if it was the scenario on the left.

 

[tarnheld]

When its bright, that is with the light on in the living room, its white with gold stripes. When in a dark room, I think its blue with black stripes

This is getting strange: When i looked at the photo in a darkened room, it was white/gold for me. Before, when watching it in a room with sunshine coming in, it was blue/black. Must be a peripheral vision thing...

 

[Imari]

After all this, I think I'm even more concerned by so many people interpreting a bright and overexposed photo as if it was the scenario on the left.

I take it the one on the left is supposed to look white/gold? I get where the gold comes from because it's legitimately a gold colour in parts, but I've never seen the blue as white.

It's possible that I've spent so much time working with colour matching pigment samples that I've trained myself to see true colour instead of what my brain thinks I should be seeing, but the whole thing just seems weird to me.

 

[TexRex]

 

[watermelon punch]

I really hope the next internet hype will be about the McGurk effect!

Particularly interesting part of that is that people used to watching dubbed movies may be less susceptible.

 

[tarnheld]

@TexRex: If The Dress is that new cybercounterterrorism thing, i like it! 👍

@watermelon punch: You mean @Johnnypenso's PD disclosure video?

But seriously, i love that perceptual glitches, you wouldn't guess how many of those are used in computer games to increase 'realism'.

 

[Liquid]

This whole thing is really funny in a "it's actually not funny at all kind of way". Almost as though in 2015 some people have only just discovered what an optical illusion is (or can be) and that colour in the human eye is about perception and not absolutes.

 

[InSight]

Personally I have no clue why the f I clicked this page, but I see dark brown and light blue, I'm pretty sure this due to overexposure as has been stated previously...or is it all in our heads...maybe there is no photo at all!

A glitch in the Matrix...

 

[Johnnypenso]

This whole thing is really funny in a "it's actually not funny at all kind of way". Almost as though in 2015 some people have only just discovered what an optical illusion is (or can be) and that colour in the human eye is about perception and not absolutes.

Also making worldwide headlines this week:

 

[TexRex]

I don't give a flying 🤬 that this is blowing up Twitter and Facebook--I don't have a significant presence on social media anyway--but it has no business coming up on the big networks' evening news broadcast. It isn't news. I don't think anyone took me seriously when I said I hope ISIS strikes again, it's just something that would surely displace this bull pucky. But hey, apparently a political opponent to Putin has been assassinated...


...so there's that.

 

[watermelon punch]

@TexRex
I think it's just evidence that people like controversy itself more than caring about the substance of any disagreement.
It doesn't even matter.

And so clever people can be powerful, and really manipulate whole groups of people if they wanted, to do whatever just by feeding their need to feel right about something & to get excited about railing against someone else or other groups they think are wrong.

... oh wait... that's already happening.

 

[TexRex]

Well as I said, my issue isn't that there's any kind of debate so much as that someone decided it's newsworthy.

 

[watermelon punch]

That's my point... ANY controversy is "newsworthy" under this idea.

 

[Pupik]

Well as I said, my issue isn't that there's any kind of debate so much as that someone decided it's newsworthy.

Philosophically, it challenges how we learn what color "is" and that color is not quite an absolute in the context of public opinion. Color can be dictated by chemical composition and reflectivity, but not exactly by visual detection.

RGB is a measurement, an absolute, but people do not work with absolutes in a concrete world. Cameras can deceive, and photos can lie; they've done so for decades, well before computers, color monitors, and image sensors. Whether accidentally or through intentional manipulation, this is something that just hits the mark as to be a bit of an optical illusion.

Example...what color is this sign?

If you know that American regulatory signs are white, then your answer is white. If if you didn't know that, then you might say blue. After all, that car is white, and the sign is not quite the same color as the car, yet neither is it the same as the sky.

 

[Imari]

After all, that car is white, and the sign is not quite the same color as the car, yet neither is it the same as the sky.

Except that the sign is in shadow, and the sign is the same colour (or near as dammit) as the parts of the car that are in shadow. The sign is either white or very close to it.

 

[watermelon punch]

Except that the sign is in shadow, and the sign is the same colour (or near as dammit) as the parts of the car that are in shadow. The sign is either white or very close to it.

You've further made his point.

 

[LeMansAid]

Just as with the dress, there's no ambiguity with the colour, only with the question being asked, and the evidence present.

 

[Imari]

You've further made his point.

Not really.

He's arguing that without information about US road signs, you could assume that the sign was blue. Or that was how I understood it, anyway.

That's not true, there's enough information in the picture to rule that out. If the sign was blue, it wouldn't appear the same colour as a shaded section of a white object. And the car is clearly white, you've got a fairly wide range of lighting states visible, so as long as you assume that the actual paint is the same colour all over the car you've got a good baseline for how white appears in this image.

Remember, no matter what the actual colour may be in the picture, your extrapolation of the real colour has to be consistent. If a blue object is going to appear white in an image, then it's going to impact the entire image. The fact that there's a known blue object (the sky) that doesn't appear washed out would indicate that the blue channel hasn't really been messed with to a great extent.

The dress picture is funky because there's very little reference in it, which should probably be leading people to assume that what they see is what they get. Ie. either blue/gold or blue/black depending on how you interpret the shadows and whether you know that there's a particular sort of fine black cloth that can be reflective like that in similarly lit situations.

The only way you get white is assuming that shadow is causing all of the colour change on the dress. Which is possible, but requires additional assumptions that are not supported by anything in the image. There isn't a white object to compare to and establish how the camera is treating shadow.

 

[watermelon punch]

Well also.. there's a big difference between a crappy exposure phone camera shot in shadow under artificial lighting inside a store... and a view in a photo being out in the sunlight on the road, taken with a fairly decent camera.

Most of us have more experience looking at the latter than the former.
Though I suppose the younger generation might actually have more experience looking at crappy exposure indoor phone pics of close-up objects.

 

[tarnheld]

Let's imagine The Dress is the only photo of the murderer in a murder case, and there are two suspects, one known to be dressed in white/gold during the murder and one dressed in blue/black, same dress. Now you be the judge -- or the defendant of any of the two.

I don't think photo analysis would help here, as there is no information about white balance, and you would have to make guesses about the actual lighting during the shot. RGB can be deceiving nonetheless.

 

[LeMansAid]

@Imari I agree with you.

Well also.. there's a big difference between a crappy exposure phone camera shot in shadow under artificial lighting inside a store... and a view in a photo being out in the sunlight on the road, taken with a fairly decent camera.

Lighting plays tricks on good cameras, bad cameras, and the naked eye. Fabrics in particular can be quite deceptive.

 

[Ibonibo]

I don't think photo analysis would help here, as there is no information about white balance, and you would have to make guesses about the actual lighting during the shot. RGB can be deceiving nonetheless.

One would load it for exemple into LR or PS. Move the t° slider and see that it only needs a slight change to become an actual representation of what reality is.

I have seen this pic on different screen, from IPS to TN to Oled. Big screen small screen, ambient lighting on or off.

All I see is blue and black.

I think the science channels explain it pretty well. Their are actually neuroscientist giving explinations about it.

I think thats it, the picture has a very special composition (bad white balance) on a very small margin that divides a group of people into the 2 camps.

 

[tarnheld]

Move the t° slider and see that it only needs a slight change to become an actual representation of what reality is.

Does that temperature slider have a tick for reality? It's really hard to reconstruct a light spectrum when only presented with RGB values. Think about it: light with specific spectrum hitting object that takes away some of the wavelengths and enters camera sensor with specific wavelength sensitivity. All the sensor outputs is a number, all spectrum information is lost. Camera sensors are made to be adequate for human perception -- with all it's limitations.

 

[Ibonibo]

The human eye has more limitation to wavelenghts than electronics.
With some filters removed a sensor can do IR, UV.
It's even possible to see radiation with a sensor.

The wavelength that is absorbed, reflected by an object gives that object the colour we see

here it's all about white balance

What do you see now?
Only corrected the temp by -5 and the exposure by -0.83

Yes some of the black are still washed out and have a greyish, brownish slug to it.
A bad camera, a bad picture playing tricks on how our brain understands that pic....