Splitting up FFT patterns

by
Will Wolcott and
Anne-Marie S Hansen

When analyzing a picture with FFT (see the picture to the left), a very significant pattern occurs. Around the center of the picture, low frequency image content is represented while higher frequency information such as edges and rapid changes are seen around the outside of the image. This project analyzed vector graphics using the Fourier transform in order to reconsile an organized image into an organized Fourier representation. When the vector graphics move in video format, subtle patterns in the Fourier domain emerge coresponding to patterns in the image domain.

How can we change the aesthetics of the FFT patterns? (see the picture to the right) By splitting the FFT pattern up in small squares that are then distributed randomly inside the frame where the FFT analysis takes place is adding another layer to the FFT pattern. The random distribution becomes an indexial sign of the FFT pattern.

When a FFT analysis is run on top of a movie in realtime, and the movie is split up into random squares that change position realtime too, the movie is abstracted into a living texture that suggests both a certain order that is more or less camouflaged by a layer of chaos. (The pictures aboyve and below are two still images from the same movie sequence).