Here's my attempt at explaining this phenomenon:<p>Our Retinas don't send raw data to our brain like a camera sensor would. Instead, the neurons in the retina already do some pre-processing, like boundary detection, or movement detection. The brain then receives signals where movement was detected.<p>This movement detection is basically just detecting changes in light level, so it works better when the contrast is high (dark gray vs. white) and works less good when contrast is low (dark grey vs black).<p>So our brain gets stronger "movement" signals when there is high contrast, and it looks like the part that has high contrast is moving faster. Since the image is designed in a way that the boundary of head/body always have low/high contrast or vice versa, it seems they are moving with different speeds.
The illusion was very strong on my mobile screen and non-existent on my desktop screen. The screen calibration was different in the latter I believe so the pigeons were appearing as a slightly different colored grey blocks visually distinct from the vertical bars.
I don’t feel this is much of an illusion. I think you just can’t see the movement of gray on black. Or maybe this is what illusions are in a very simple form?
I was in bed reading this with an eye closed and was very confused until I discovered the illusion works for me only when I watch it with both eyes open.<p>I found this surprising—I know three dimensional optical illusions depend on focus and perspective, but I was unaware two dimensional optical illusions were as well.
It would have been interesting to control the width of the vertical stripes too , to see at what point it "starts" being a illusion. I assume for very thin stripes it might not work so well .
This seems to be, for lack of a better analogy, an effect of rounding colour intensities in the "motion compensation" processing of the brain. The grey and black become perceived as identical and thus no motion is detected in those areas where the two colours are overlapping.
I wonder what is the maximal extent of the craining and the shrinking illusion that can be achieved.
Is it to do with the velocity? The grid thiknes?<p>If the illusion is an error, what maximizes it?