9
$\begingroup$

This question was originally asked on Physics, then moved to Cognitive Sciences.

Consider the following image: enter image description here
You might want to display the image in a new page, in case it gets resized for mobile displays.

On the top half, there is a pixel-sized checkerboard pattern with alternating black and white pixels; on the bottom half, there's a black to white gradient. Now, I don't know if you see the same, but for me, when viewing from distance or defocusing my eyes and the top half blending into one color, I can't find any color in the gradient below which would match it.

With simple arithmetics, I would guess at first guess the resulting color be either RGB $(0.5, 0.5, 0.5)$ or, with gamma, $(\sqrt{0.5}, \sqrt{0.5}, \sqrt{0.5})$. I cannot help myself, but the resulting color appears a lot warmer than the metallic gray which would appear if you for example zoomed this page or used a blur filter. I tried to add some yellow to the gradient, and the result looks more similar to the perceived color.

Now, based on the comments, it appears some people perceive yellowish tint and some don't. And I do on my LCD computer screen but not on my mobile display. Thus I guess it's based on some property of the display.

Why don't we perceive the resulting color like real gray? Where does the yellow color come from?

I have a theory: Based on the color arrangement in a typical LCD pixel, one white pixel would contain the red and green color together and blue on the right side. Blue color appears darker to human eye than colors with the same physical intensity, so a white pixel is more green than red than blue. Green and red have roughy the same perceptional insensity, and mixing red and green color addivitely gives yellow, thus the yellow tint? Well, shouldn't then all white pixels on the display appear a bit yellow?

Is this possible? Are there any other explanations?

A side question: Do you know any computer image scaling algorithm or blur filter that tries to mimic this, simulating a blurry vision of a human eye correctly?

$\endgroup$
  • $\begingroup$ Excellent question! I've been wondering about this myself. It's not that the checkerboard pattern has a yellow tint... it's the grays that are incorrect. When I compare what I see on my laptop to my phone, where the checkerboard pattern seems to have roughly the same color as the grays, the gradient on my laptop appears distinctly more blue. Perhaps a better question would be to ask why do shades of gray look blue on most LCD computer screens. $\endgroup$ – Quinchilion Jun 8 '16 at 14:11
  • $\begingroup$ @Quinchilion Interesting perspective. I'm so used to how gray looks on my screen that I might not notice if there's some blue to it, though it looks really metallic to me. It might be relevant that blue is the complementary color of yellow. $\endgroup$ – IllidanS4 Jun 8 '16 at 14:28
  • 1
    $\begingroup$ Monitors are frequently very poorly calibrated. That's the most likely cause of issues with images not looking grey. Further, assuming the display is meant to be sRGB, then (and this is from memory so please take with a grain of salt) the RGB values that should correspond to a B&W chequer board are about 186. What you may find helpful is to repeat your experiment separately with each of R,G & B using a Primary/Black pattern VS Black->Primary blend. The crossover points are ideally meant to be at the same position but frequently aren't. $\endgroup$ – Simon F Jun 9 '16 at 10:34
  • $\begingroup$ @SimonF 186 corresponds roughly to the middle of the bottom gradient, and that place also seems to match the intensity of the top rectangle. $\endgroup$ – IllidanS4 Jun 10 '16 at 23:30
10
$\begingroup$

Because your monitor is not properly calibrated.

On my screen at home the top and bottom parts have the same hue. At my office though, the top part tends to looks a bit yellow compared to the bottom part that looks more red.

The difference: my screen at home was from a series that was decently calibrated out of the factory, and on top of that I did calibrate it properly with a color calibration tool. That was quite some years ago and its colors have probably degraded since, but it still makes a difference.

A poorly calibrated monitor will not display exactly the color intensity requested for R, G and B, resulting in differences between the color that should be displayed and the one actually displayed. It can even vary depending the area of the screen (many consumer level LCD screens tend to have more light leaking near the edges for example).

The image you included is a good example to highlight a calibration issue, because it allows to compare directly the average color at 50% intensity (combining two halves at 0% and 100% intensity) with the color displayed when requesting 50% intensity. Moreover, the gradient allows to see if the hue is constant from 0% to 100% intensity.

$\endgroup$
  • 1
    $\begingroup$ Yup same exact result, i even went as far as calibrate the one not calibrated showing hue and poof gone. $\endgroup$ – joojaa Jun 9 '16 at 16:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.