APOD: The Same Color Illusion (2023 Dec 18)

[Chuck Almdale]

Has anyone done a pixel count on this "illusion?"
I made a 3/8" by 1 1/4" rectangular hole in some paper and covered up everything by the "connecting strip" between the letters A & B. Frankly I see a brightness gradient going down my screen as a result. This suggests to me that there are fewer gray pixels per cm2 of whatever the gray color is as you descend. So, no, to me they're not really the same color. Close, though. This may have to do with the quality of my screen, or the angle of tilt, but this is what I have to work with. To my eye, at my screen, not the same.

[Chris Peterson]

Has anyone done a pixel count on this "illusion?"
I made a 3/8" by 1 1/4" rectangular hole in some paper and covered up everything by the "connecting strip" between the letters A & B. Frankly I see a brightness gradient going down my screen as a result. This suggests to me that there are fewer gray pixels per cm2 of whatever the gray color is as you descend. So, no, to me they're not really the same color. Close, though. This may have to do with the quality of my screen, or the angle of tilt, but this is what I have to work with. To my eye, at my screen, not the same.

There is no brightness (color) gradient. The number of pixels and the pixel density are not relevant, only the value of the pixels, which is identical for all the pixels in both squares.

Your eyes may be fooling you, or you may have some artifact created by your screen. But the numbers don't lie. Which is why when we analyze images, we look at the numbers.
_

greyillusion_wikipedia_960-copy-2.jpg (13.37 KiB) Viewed 35233 times

[zendae1]

I framed the image with both my hands so that nothing could be seen from either side. The two depicted squares looked the same. When I took my hands away, the two squares remained the same; my eyes were trained. A minute or so hence, I looked again, and the two squares again looked different. (Spock eyebrow)...fascinating...

[rwlott]

I make a case that the square values are different because of context. To extract the data out of its context, yes, there is a certain raw value that is identical. But to extract the data and proclaim "puzzle solved" is to abandon the context; in effect, this is abandoning data. Basically I'm arguing that a value of a lit area is different from a value of a shaded area, regardless of discovery of certain identical information, because 'lit' and 'shaded' are data. I don't see the influence of perception as discardable.

This is about science, and about how our senses can fool us in analyzing an image. Instead of a chessboard, make this a nebula, with two physically identical regions separated spatially, and appearing different to our eyes but measuring instrumentally the same.

We would never claim the regions are different simply because of a perceptual failure. We would deliberately discard our perception and accept the instrumental measurements.

I first encountered this illusion almost 30 years ago, and even used it to illustrate Pastorian's point in my statistics classes: Context is essential for properly interpreting and communicating the results of statistical analysis. I'm not disagreeing with Chris about the need to eliminate perceptual failure in analyzing astronomical images, but context is an important element in many sciences.

[Pastorian]

I make a case that the square values are different because of context. To extract the data out of its context, yes, there is a certain raw value that is identical. But to extract the data and proclaim "puzzle solved" is to abandon the context; in effect, this is abandoning data. Basically I'm arguing that a value of a lit area is different from a value of a shaded area, regardless of discovery of certain identical information, because 'lit' and 'shaded' are data. I don't see the influence of perception as discardable.

We're delving into philosophy here, but if two things measured by certain data are identical, what difference does the context make, other than to affect perception (aka, the effect of the whole on human senses)? You're saying that the context is also part of the data? A bullet in a gun is definitely in a different context than a bullet in an ammo box, but it's the same bullet!

If I was a house painter and a client called up and said "I just love today's APOD! Could you paint my house like that? I'd like the main color to be square B and the accent trim color to be square A." If I went by the color measurements we're discussing, they'd say "What's this?? We want our money back!" What they really wanted was the perceived contrast. I then would be tasked with reconstructing that contrast and would need to identify two different color values to do that. The perception is a phenomenon so it must have data points. As soon as you isolate the squares in open space like some of the above illstrations, you are taking the information out of its original context. It is no longer a picture of a cylinder casting a shadow over a chess board. It is something else. I actually think the whole thing is cool, how in isolation, the values turn out to be identical. My point is that when isolating the data like that, keep in mind the source, and ask what information you are losing. It is as equally fascinating that we perceive something else.

[Chris Peterson]

I make a case that the square values are different because of context. To extract the data out of its context, yes, there is a certain raw value that is identical. But to extract the data and proclaim "puzzle solved" is to abandon the context; in effect, this is abandoning data. Basically I'm arguing that a value of a lit area is different from a value of a shaded area, regardless of discovery of certain identical information, because 'lit' and 'shaded' are data. I don't see the influence of perception as discardable.

We're delving into philosophy here, but if two things measured by certain data are identical, what difference does the context make, other than to affect perception (aka, the effect of the whole on human senses)? You're saying that the context is also part of the data? A bullet in a gun is definitely in a different context than a bullet in an ammo box, but it's the same bullet!

If I was a house painter and a client called up and said "I just love today's APOD! Could you paint my house like that? I'd like the main color to be square B and the accent trim color to be square A." If I went by the color measurements we're discussing, they'd say "What's this?? We want our money back!" What they really wanted was the perceived contrast. I then would be tasked with reconstructing that contrast and would need to identify two different color values to do that. The perception is a phenomenon so it must have data points. As soon as you isolate the squares in open space like some of the above illstrations, you are taking the information out of its original context. It is no longer a picture of a cylinder casting a shadow over a chess board. It is something else. I actually think the whole thing is cool, how in isolation, the values turn out to be identical. My point is that when isolating the data like that, keep in mind the source, and ask what information you are losing. It is as equally fascinating that we perceive something else.

The context here, however, is a simple illustration of how our senses can fool us when looking at an astronomical image. Not picking colors to paint a house!

[johnnydeep]

Has anyone done a pixel count on this "illusion?"
I made a 3/8" by 1 1/4" rectangular hole in some paper and covered up everything by the "connecting strip" between the letters A & B. Frankly I see a brightness gradient going down my screen as a result. This suggests to me that there are fewer gray pixels per cm2 of whatever the gray color is as you descend. So, no, to me they're not really the same color. Close, though. This may have to do with the quality of my screen, or the angle of tilt, but this is what I have to work with. To my eye, at my screen, not the same.

There is no brightness (color) gradient. The number of pixels and the pixel density are not relevant, only the value of the pixels, which is identical for all the pixels in both squares.

Your eyes may be fooling you, or you may have some artifact created by your screen. But the numbers don't lie. Which is why when we analyze images, we look at the numbers.
_
greyillusion_wikipedia_960-copy-2.jpg

Hmm. Now that the two squares are isolated, the color of the A is definitely lighter than the color of B (and blurrier)! Though they both consist of many different pixel colors, A has more 0x"5n5n5n" values and B has more 0x"4n4n4n" pixel values. I wonder if that affects our perception of the illusion when the rest of the squares are present.

[Chris Peterson]

Has anyone done a pixel count on this "illusion?"
I made a 3/8" by 1 1/4" rectangular hole in some paper and covered up everything by the "connecting strip" between the letters A & B. Frankly I see a brightness gradient going down my screen as a result. This suggests to me that there are fewer gray pixels per cm2 of whatever the gray color is as you descend. So, no, to me they're not really the same color. Close, though. This may have to do with the quality of my screen, or the angle of tilt, but this is what I have to work with. To my eye, at my screen, not the same.

There is no brightness (color) gradient. The number of pixels and the pixel density are not relevant, only the value of the pixels, which is identical for all the pixels in both squares.

Your eyes may be fooling you, or you may have some artifact created by your screen. But the numbers don't lie. Which is why when we analyze images, we look at the numbers.
_
greyillusion_wikipedia_960-copy-2.jpg

Hmm. Now that the two squares are isolated, the color of the A is definitely lighter than the color of B (and blurrier)! Though they both consist of many different pixel colors, A has more 0x"5n5n5n" values and B has more 0x"4n4n4n" pixel values. I wonder if that affects our perception of the illusion when the rest of the squares are present.

greyillusion_wikipedia_960-copy-2b.jpg (12.27 KiB) Viewed 35207 times

[Chris Peterson]

Has anyone done a pixel count on this "illusion?"
I made a 3/8" by 1 1/4" rectangular hole in some paper and covered up everything by the "connecting strip" between the letters A & B. Frankly I see a brightness gradient going down my screen as a result. This suggests to me that there are fewer gray pixels per cm2 of whatever the gray color is as you descend. So, no, to me they're not really the same color. Close, though. This may have to do with the quality of my screen, or the angle of tilt, but this is what I have to work with. To my eye, at my screen, not the same.

There is no brightness (color) gradient. The number of pixels and the pixel density are not relevant, only the value of the pixels, which is identical for all the pixels in both squares.

Your eyes may be fooling you, or you may have some artifact created by your screen. But the numbers don't lie. Which is why when we analyze images, we look at the numbers.
_
greyillusion_wikipedia_960-copy-2.jpg

Hmm. Now that the two squares are isolated, the color of the A is definitely lighter than the color of B (and blurrier)! Though they both consist of many different pixel colors, A has more 0x"5n5n5n" values and B has more 0x"4n4n4n" pixel values. I wonder if that affects our perception of the illusion when the rest of the squares are present.

[Pastorian]

We're delving into philosophy here, but if two things measured by certain data are identical, what difference does the context make, other than to affect perception (aka, the effect of the whole on human senses)? You're saying that the context is also part of the data? A bullet in a gun is definitely in a different context than a bullet in an ammo box, but it's the same bullet!

If I was a house painter and a client called up and said "I just love today's APOD! Could you paint my house like that? I'd like the main color to be square B and the accent trim color to be square A." If I went by the color measurements we're discussing, they'd say "What's this?? We want our money back!" What they really wanted was the perceived contrast. I then would be tasked with reconstructing that contrast and would need to identify two different color values to do that. The perception is a phenomenon so it must have data points. As soon as you isolate the squares in open space like some of the above illstrations, you are taking the information out of its original context. It is no longer a picture of a cylinder casting a shadow over a chess board. It is something else. I actually think the whole thing is cool, how in isolation, the values turn out to be identical. My point is that when isolating the data like that, keep in mind the source, and ask what information you are losing. It is as equally fascinating that we perceive something else.

The context here, however, is a simple illustration of how our senses can fool us when looking at an astronomical image. Not picking colors to paint a house!

I was illustrating that perception has data that can't necessarily be measured but can be approximated. It's not lost on me that certain aspects of perception needs to be set aside to do your work, and that, it seems, math becomes a sixth sense through which to observe. But has there ever been a true-to-life APOD? The data gathered is parsed back into digestable visual representations of that information.

[bls0326]

Similar to using Color Picker, I used a "Replace Color" routine. Replace any Area A color (+/- alittle) with bright blue. Here is the results on my computer monitor. Not just the darker looking squares, but some bleed over into white squares in the "shadow" of the cylinder.
https://drive.google.com/file/d/1ESblyM ... sp=sharing

[Confused]

Also, when I copy the image and paste it into Irfanview then copy a portion of each square (a portion of the background) and paste each into a new image then they look the same color.

[Avalon]

Hey, we need Ann, the color expert, to weigh in on this one.

[Ann]

Hey, we need Ann, the color expert, to weigh in on this one.

Nothing much to say that hasn't been said already. But yes, human color perception is relative.

Consider a black cat:

Click to view full size image

This particular cat is being partly illuminated by red light, so that one of his ears looks red.

Now imagine that you, after protecting the kittie's eyes, exposes it to absolutely brilliant white light. Using your marvelous camera, you take a picture of the black cat. Under the blinding white light, the black cat will look white.

Okay, I can do better. Forget about the real kittie and imagine a stuffed toy, as black as a typical black kittie. Send it into space and make it orbit the Sun. When the light from the Sun reflects off the orbiting black toy, it will gleam white.

Just like the Moon looks white when the Sun shines on it, an orbiting black toy would look white in the gleam of sunlight. Credit: Drew J. Evans

---

Now consider a white sheet of paper:

Click to view full size image

Remove basically all light from the room. Use your marvelous camera and take a visible-light picture of the white sheet of paper. It will look black.

My point: A black toy will look white when it is illuminated by brilliant light. A white sheet of paper will look black when it is not illuminated at all. If we were somehow able to place the brilliantly lit black toy next to the unilluminated white paper, the cat would look white and the paper would look black.

Black cat ➔ ███ White sheet of paper ➔ ███

And remember: In the darkness, all cats are gray.

When it's dark, all cats are gray.

---

Ann

[Confused]

This particular cat is being partly illuminated by red light, so that one of his ears looks red.

I think most people's brains understand that.

imagine a stuffed toy, as black as a typical black kittie. Send it into space and make it orbit the Sun. When the light from the Sun reflects off the orbiting black toy, it will gleam white.

Is it your intent to imply that the Moon is Black?

Remove basically all light from the room. Use your marvelous camera and take a visible-light picture of the white sheet of paper. It will look black.

As above, I think most people's brains understand that.

[Confused]

Hey, we need Ann, the color expert, to weigh in on this one.

I think no one has explained what the requirements are that causes our brains to be confused. What are the features of this image that causes the misconception?

[Chris Peterson]

Hey, we need Ann, the color expert, to weigh in on this one.

I think no one has explained what the requirements are that causes our brains to be confused. What are the features of this image that causes the misconception?

Our brain interprets the different colors created by obvious shading as the same color.

[Lasse H]

Sorry but they don't look the same color to me, with or without the bridge. Not even close, one is definitely lighter, one darker.

But that's the whole point of an optic illusion ! - In this case: that they "don't look the same color", but yet they are.
If you measure the color in the two squares, you will find that they both are RGB(120, 120, 120).

[Chris Peterson]

Sorry but they don't look the same color to me, with or without the bridge. Not even close, one is definitely lighter, one darker.

But that's the whole point of an optic illusion ! - In this case: that they "don't look the same color", but yet they are.
If you measure the color in the two squares, you will find that they both are RGB(120, 120, 120).

And the reason is because of the way our brain compensates for shadows. We interpret the chessboard as having squares with only two colors, uninfluenced by shading.

[Ann]

imagine a stuffed toy, as black as a typical black kittie. Send it into space and make it orbit the Sun. When the light from the Sun reflects off the orbiting black toy, it will gleam white.

Is it your intent to imply that the Moon is Black?

The color of the Moon is about the same color as asphalt.

Ann

[Chris Peterson]

imagine a stuffed toy, as black as a typical black kittie. Send it into space and make it orbit the Sun. When the light from the Sun reflects off the orbiting black toy, it will gleam white.

Is it your intent to imply that the Moon is Black?

The color of the Moon is about the same color as asphalt.

Ann

Freshly laid asphalt!

[Ann]

What color is grass? It depends on the illumination:

Click to view full size image

Ann

[Chris Peterson]

What color is grass? It depends on the illumination:

Click to view full size image

Ann

It should be noted that this is why those tools that try to create a screen color (RGB values) for some given wavelength of light are marginally useful at best. Because color is a physiological phenomenon much more than a physical one. Our brain interprets the image you provided, and we readily understand that both the left and right sides of the image are showing grass of the same color. A measurement will show they are different, of course. What color are oxygen or hydrogen emissions from a nebula? The wavelengths are very accurately known. The color depends on the intensity of the emission.

[Ann]

I like this picture, which won third prize in a Swedish photo competition. You can see how the snow has melted on the left side of the lawn in the sunshine, but the snow is still intact where the grass is in shadow. Of course, I really like the very blue color of the snow as well, which is due to the fact that the blue light from the sky is reflecting off the snow, but the more yellow light from the Sun hasn't reached the part of the lawn that is in shadow.

Ann

[Confused]

both the left and right sides of the image are showing grass of the same color

Actually, no. In the image there are darker colors on the left and lighter colors on the right. This can be verified using tools. What we see in the image is a true difference in colors. Very often the difference between an area in a shadow and an area with direct sunlight seen in photographs is dramatic, apparently because our eyes are better at making relevant adjustments but what is relevant here is the image, just the image. I think our eyes can properly interpret the image of grass provided previously.