I actually had to cut the squares out in Photoshop and paste them together in a new document before I believed it.

That seems to be the standard reaction. Erik Olson was the one who brought this to the attention of the SMOFs mailing list, saying "Best illusion ever. When you don't believe it, fire up your favorite image editor and compare pixels." The first message I actually read about it was Beth Friedman's reply to him, saying she'd cropped it down to just those squares and only then believed it.

A perfect illustration of the difference between Liberals and Conservatives. Move the object casting the shadow, or the position of the light, and their appearance will change.

All sorts of analogies to be made here.

Bloody hell, that's impressive. I also had to isolate the pixels before I believed it, though I used a handful of post-it notes rather than mucking about with software. As soon as I removed the post-its from the screen, the illusion was every bit as strong as it was initially.

The contrast bit is the same reason you see black on computer monitors which are physically incapable of displaying black.

I did the whole photoshop thing too, and noticed that if you take a tiny little square filled with the color of the "lighter" block and move it around on the checkerboard in the area between the "darker" and "lighter" blocks that the illision actually has a point at which the color "switches over".

I sent the URL to co-workers and college friends. One of the latter called long distance to cuss me out for playing with his head. He suggests drawing a line of the Mystery Color across the checkerboard.

A co-worker did a low-tech comparison. He printed out the image and snipped out Square A and Square B. Same color, you betcha.

Careful, Stefan - I tried the low-tech print-it-out solution, and from my printer, the two squares were not exactly the same color, though closer than they'd seem.

Playing around with PhotoEditor, I find that the secret is that the squares diagonally adjoining A and B change shade much more gradually than the fuzzy shadow edge suggests. Even quite close to the corners they're a markedly different shade from the AB color. This is more strikingly seen in the part of the "light" square just to the upper-left of B.

I was intrigued by this latest in a series of generally successful optical illusions. Something, however, didn't seem quite right. I tried swatches in Photoshop, but was still unconvinced. Finally, I tried the 'print and snip' method, which left me with a little piece of the mystery color.

As an experiment, I tried turning my back on it for a minute or two, then suddenly spun around to confront it. Not only did I discover that it was in the process of quietly changing its hue, I also caught it trying to hack into my personal bank account!

This breakthrough led me to try various methods of ascertaining its intentions. It finally broke and admitted that it was placed here surreptitiously by agents of a foreign power (an evil one whose name you would recognize instantly) to undermine our very American way of life!

I disposed of it properly, and wiped all records of it from my computer, and urge you to do the same. It seems to me that your web page uses what looks suspiciously like the same shade in one or two key places. I think this also explains the mysterious Grays (and their cognate fellow travelers, the Greys) of Roswell, NM.

Don't let emotion sway you in this matter. The truth, objectively viewed, is black and white.

I'm disappointed. I was hoping to come into this comments section and see people posting things like, "Wait a minute. I just examined those pixels in Photoshop and they are so different! This isn't an optical illusion; it's a fraud."

Alas, now I have no choice, but to disbelieve my eyes at all times. This is going to make driving a lot more difficult.

Back around 1978, I saw a lecture by Edwin Land (of Polaroid) . He showed a similar (and equally vivid) illusion, except with color rather than gray scale. His contention was that your perception of color depended on the sequence of edges that your visual system processed in going from the "background" (or some reference point) to the patch in question, not just (or even mainly) on the actual red-green-blue brightnesses. At the time, his theories of vision were still pretty radical, but I think a lot of them have been adopted into the main stream of thought about how vision works.

Also, if you've ever been to a "Mystery Spot" they do the same thing with visual orientation -- build a house where all the visual cues for vertical and horizontal are actually tilted. Even though you *know* that's what's going on, it's nearly impossible not to see those cues as closer to vertical/horizontal than they are, with the result that it feels like gravity is tilted the opposite way, pendulums appear to hang at an angle, etc. (The DC Metro system has a number of escalators that are pretty good unintentional Mystery Spots -- look across at the opposite escalator and your visual system tries to make the escalator horizontal, with all the people leaning over at a steep angle.)

The Swiss Alps are full of accidental Mystery Spots, because there is no visible flat horizon and the ground slopes in all directions. At one place I found a cow trough sticking out of a slope in which water seemed to be flowing uphill by a few degrees. Again, the illusion of a misplaced vertical was so powerful that it was impossible to shake intellectually.

The Swiss Alps are full of accidental Mystery Spots, because there is no visible flat horizon and the ground slopes in all directions.

Ditto the Colorado Rockies. There were several times where Mike Pins and I were convinced that we were either driving on a flat, or a downhill slope -- but the GPS and Altimeter clearly showed us ascending a pretty decent grade.

Airplane making a steep turn.

Actually all the illusions are correct except the Craik-Obrien one with the gray gradient:

http://www-bcs.mit.edu/gaz/demos/craik.html

It is *not* a true illusion, because although they have taken squares out of the same region, the region itself has a light-to-dark *gradient* variance so the left side of each portion of the square is indeed a bit lighter and the right is darker. Think of it as every pixel to the left being lighter than every pixel to the right by an infinitesimally increasing degree.

When juxtaposed to "itself" in the exact same orientation, the right side is touching the left side, so of course they are different. However if you were to flip *one* of the clone squares 180 degrees in either direction so that sides of the same gradient touch, you will see either a concave or a convex strip.

Makes sense?

:-)

Vera

Here's one that just showed up on Geek Press. Circles you don't look at whirl; circles you do are still. Freaky. (But then, so are most optical illusions.)

---L.