If you're like many regulars to ScienceBlogs you probably found the cool Purple Nurple optical illusion over at Omni Brain. I don't really understand why a static object appears as though it's pulsing, but I do enjoy the effect. Did you ever wonder how much an optical illusion can be distorted and still maintain the illusion? Mighty Optical Illusions has a bunch of items similar in effect to Purple Nurple. I grabbed the one below (it reminds me of a bunch of almonds). It has a very cool wavy effect.
(much more fun below the fold)
The fun begins when we start to manipulate it. For this little adventure I used the very useful freeware viewer/editor XNView. First, is the color what gives it the effect? Here's a negative version:
OK, how about if we equalize it?
Perhaps the orientation has something to do with it. We can flip it and rotate it:
Still it waves (I kind of like the flipped version even more). How about swapping some colors? In the first version, RGB are swapped to BRG, and in the second to GBR. Both appear to reduce the illusion, the second swap more so than the first.
So much for the simple stuff. How about wholesale alteration of the shapes? First, let's look at Slice. Here the image has been chopped up but the effect is still apparent.
Next comes Shear:
Lots of sharp jaggies so it's not the simple, smooth shapes that give it the effect. Next comes Swhirl:
This whirlpool-like effect produces heavy-duty spatial distortion but the illusion remains. Tile is kind of like Shear but there are obvious black voids in the new image. The illusion seems to be less intense so perhaps there's something about the contiguous nature of the pattern that is creating the illusion:
Our next edit is Waves:
This also produces considerable distortion of the image but yet again, the illusion continues. At this point I decided to try adding Gaussian Noise (at 50%):
The illusion is also reduced. Further increases in noise reduce the illusion even more. Finally, I cropped the image so that only a small segment was left:
Illusion gone.
Brain & Behavior
Comments
Wow I would have expected some of those treatments to abolish the illusion. Interestingly, some of them seemed to almost halt the wave when I was looking directly at the image, but when I saw it out of the corner of my eye it went crazy. For instance, this happened in the "equalized" image, as well as several others.
Posted by: Jason | January 8, 2008 3:03 PM
The GBR version didn't move for me at all, which was really weird
Posted by: Hadyn | January 8, 2008 3:22 PM
I just barfed on my keyboard. Thanks.
Posted by: mg | January 8, 2008 3:58 PM
The cropped version certainly contains no illusion - but if you just look at that upper-left corner of the original image, there's not much illusion going on there either.
Posted by: Pete | January 8, 2008 7:28 PM
In fact, I think it has to do with your eyes moving---you need to have regions of picture going in and out of your fovea (by saccading over it) for it to work. If the picture stays constant on your retina (just try staring at one place) then the illusion goes away.
Posted by: Pete | January 8, 2008 7:32 PM
I think Pete hit it. I found that for me the illusion of motion depends on eye movement. It appears when I first look at the image or when I move the focus of my view around the image, but the apparent motion stops when I stare steadily at one point.
Posted by: Mark P | January 8, 2008 10:40 PM
Whoa... I just noticed that the "almonds" seem to spin if you scroll down while looking at the image
Posted by: Monty | January 12, 2008 6:06 PM
This illusion works much better for me on my monitor than when i print it out on paper. Moving in closer than my minimal focal distance also diminishes the effect.
Posted by: Tegumai Bopsulai, FCD | January 16, 2008 3:54 PM