Monday, March 06, 2006

Appearance vs. Reality


The checkershadow illusion is one of the more remarkable "illusions" I've encountered, that goes to the heart of how we (as visual information processors) interpret and categorize the world; it also goes to the heart of the question, "Are you really sure of what you are looking at?"

Believe it or not, the squares marked A and B are exactly the same shade of gray! If you do not immediately believe your senses (as I suspect you won't!), just copy/save the image to a jpeg file and use any image processing program to sample the actual luminosity of each square...truly amazing!).

The "explanation" is that our visual systems require more than just luminosity to assess the shade of grey to be assigned; it also needs such features as local contrast and boundary effects. A complete explanation is provided here.

The checkshadow illusion was devised by Edward H. Adelson, Professor of Vision Science in the Dept. of Brain and Cognitive Sciences, Massachusetts Institute of Technology. His site includes many other startling illusions that explore the nature of perception and interpretation of reality (see his Illusions and Demos), as well as technical papers explaining his theories and findings. Adelson's site is a must-see for all photographers who "believe" they know all there is to know about appearance, reality, and the true nature of tonal gradations.

Additional references (and illusions) appear on the Perceptual Sciences Group homepage.

1 comment:

David S. Mazel said...

It's good to see, no pun intended, a discussion of what one actually sees and doesn't see. Moreover, another question is: When one takes a picture, how true to reality is the final picture? I think a good question to ask is: How can we measure the duplication of reality in a photo? What is the fidelity of a picture and how do you know it?

In the image here, we see two different shades, yet they are identical from an pixel value view. Does that metric make them equal? Perhaps equality should be taken from what the human being interprets and not from a pixel measurement? That is, if you think the colors are different, then they are different. The fact that they may have the same image value (i.e., pixel value) is irrelevant. Is that the way to distinguish images or colors? I don't know the answer to this, but the question is worth thinking about.

I am reminded of the movie The Matrix: How do you define real? (Morpheus). This post is another step along that thought.

By the way, here's another thought: While the explanation for how the human visual system is of interest, I also think we should consider that humans, when they see something, compare what they see to what they have seen sometime in the past. That is how we are able to interpret scenes so quickly. We have a stored interpretation in mind and compare what we see to what we expect to see. I suspect, and don't know, that this may have something to do with how we interpret the image here as well.