Impossible colors | Hacker News Reader

Impossible colors

en.wikipedia.org

90 pointsm0th8730 comments

Back when I was a wee art student, I had a phenomenal printmaking professor who was an expert in inks. One of her projects was a monotype, and mine included a cloud background, but I couldn't get the colors right: a dusky yellow cloud background. You've probably seen the hue following an afternoon thunderstorm late spring.

I'd mix, and mix and end up with a dark yellow, that when printed over blue turned green, or reddish or some other stupid hue. And my printmaking professor came over, and she said to use gray, and some yellow, and some brown (please, don't ask for the exact ink mix). When we were done, the ink sat on the slab as a puce, but printed slightly yellow.

However, when printed over a thalo blue, it was exactly like I needed: yellow-blue. Not green, yellow-blue. It was the damnedest thing, and not lost on the other students: I had to remix it at least six times within a week for other people.

The color doesn't reproduce at all on screen, so I have to assume it's partially an optical illusion.

The color doesn't reproduce at all on screen, so I have to assume it's partially an optical illusion.

I'm sure you know this, but RGB is not able to represent all possible colors, and monitors themselves are not able to represent all of RGB. This gives us a somewhat limited subset of available colors for use on screen, which is why (I assume) photographs don't look as colorful as the subject did real life. The film / CCD can't record all color information, and the print / monitor can't display it all.

Is there any known, or conceivable technology, that would improve the color field for displays?

Adding color channels can help. For example, Sharp sells LCD televisions with an extra yellow channel. Problem is that these need software computes/approximates/guesses at the perceived yellowness of pixels based on their RGB (for computer monitors) or HSV? (for television) values. I have no idea how well we know color perception to pull that of reliably, whether it actually can work across the population (including the colorblind), or whether they implemented the best-known solution.

There are "HDR" displays that can display a wider range of colours, my university had one.

HDR (high dynamic range) displays are those that can reproduce wider differences between dark and light. What you're talking about would be called a wide gamut display, such as an LCD with RGB LED backlighting or a special wide-gamut fluorescent backlight.

RGB is not able to represent all possible colors

Indeed, but it’s not completely an issue of gamut: the hue is always off even if it’s balanced against a card.

When I look at the blue/yellow thing on Wikipedia, I mostly get a headache. Depending on the angle I view it at, one color or the other dominates and it can change colors while I'm looking at it.

I don't seem to get a blue/yellow unless you count where I can get it to be yellow in the middle with some blue at the edges that tends to merge with the black border.

And I find that, hilariously, my visual system actively fights the merging of the boxes. When I successfully look at a real stereogram, there is a point at which my eyes snap into alignment and the stereogram pops out. In this case, it's like the opposite: I try to steer my eyes to the point where the blue and yellow boxes overlap, but if I get close they skitter apart. It's like trying to bring two identical magnetic poles together.

I'm sure this stuff is highly idiosyncratic. Everyone's visual system is different. It develops via visual feedback at early ages and so is different for everyone, sometimes to one's detriment:

http://www.lazyeye.org/

It would probably help if the "+" were a more visible color than white, as it gets lost a bit in the yellow when crossing.

My visual system fought back too. I'm normally able to free-fuse, but in this case my right eye just started hurting.

I don't get a single blue-yellow colour that I'm happy with, instead it either swaps from one to the other and back (patchily), or it holds semi steady as if I was looking at yellow through a blue mesh (or blue through a yellow mesh) but doesn't feel like a single colour, more like a checkerboard grid of two colours.

I get the same checker effect. That, or I see a gradient of blue to yellow. Maybe it's just my LCD.

I've seen the blue-yellow color only once before -- during a cluster headache. Seeing it again did not bring back the fondest of memories.

Also note imaginary colors: http://en.wikipedia.org/wiki/Imaginary_colors

You can see some imaginary colors here: http://www.skytopia.com/project/illusion/ipage-et.html

So... does this blue circle go away eventually? Been a few minutes and I still can't look at a white surface without seeing a clearly defined blue circle...

Yes, it goes away. But it may take a little while.

That Skytopia thing is crazy, you definitely need to stare for the full minute to get the effect though.

Also, any time I do these afterimage experiments I notice that my eyes are not very well aligned.

Yeah, I was rather astonished the first time I tried it. I expected it to be unconvincing, but nope, that is definitely a brand new shade of cyan.

Wauw, this is a whole new HN - does it come as a Chrome extension ?

It just works!

Ow.

This reminded me of a fact that astounded me when I first read it: Magenta is a colour created by our brain - there is no wavelength for it.

Color is not a physical phenomenon, but a perceptual one; there are some wavelenghts that, in isolation, lead to perception of specific colors, but actually, there isn't a wavelength for any color.

For example, one can have metameries (http://en.m.wikipedia.org/wiki/Metamerism_(color)), where different physiclaal signals lead to perception of the same color.

Right, although we have been trained otherwise by rainbow diagrams from birth, colors do not map 1:1 to wavelengths, and most of the light we observe is a mixture of wavelengths; we can only sample an intensity from each of three fuzzy bands of the spectra, and the rest is in our heads.

As someone who enjoys both pesto pizza, and red wine, I can safely say that reddish-green does in fact exist.

I once read that colour-blind synaesthetes (or color-blind synesthetes, for you furriners) were able to perceive colours through synaesthesia that they were physiologically incapable of perceiving with their eyes. Thus they abnormally associated concepts with normal colours that were abnormal to them.

They used to teach us elementary school that "blue+yellow=green". With the advance of science we fix some of our naive illusions. I don't know how many new we introduce...

Neon brown.

I can't cross my eyes :(

Reply on news.ycombinator.com