I would like to create a program that constructs and deconstructs an autostereogram. I am very interested in the mechanism in which people can perceive them and think it would be interesting to explore how long you can see the object pop out once you know that it is there. The user will be able to interact with the speed and amount of “random” generation of the pixels to see how long they can see the object clearly. I would like this to help people who have trouble seeing the objects and also to be an interesting perspective for people who can see them and don’t quite understand how or why.
I have not yet decided what the object will be, I think that I will be able to pick an object once I have tested it some more, but I think I will choose something that is easily recognizable.

http://en.wikipedia.org/wiki/Autostereogram#Simulated_3D_perception