Worm Walker uses frequency information from an image to control a separate stochastic walk blit of its pixel data. Programmed in C++ using SDL and FFTW libraries. 
This is a flow chart of our process for a stochastic walker. There are three probabilities that determine the movement of the walker; vertical, right, and up. First, it decides whether to go vertical or horizontal, and then depending on that, it chooses a direction along that axis on which to go. The walker also has a step size that determines the distance it moves. 
left right walk  This was our first attempt visualize the behavior of a stochastic walk. The following ideas progressed from what we observed here. 
pixel spitter  This was an attempt to make two seperate mechanisms, one to read pixel data from a source image, and the other to display it on screen. What you see is the stochastic walk on the source image being displayed on screen sequentially, left to right, top to bottom. 

two independent walkers  Here, you see a stochastic walk on the source image being blitted to the display screen along the path of a second stochastic walking mechanism. When the seed and probabilities for both walkers are the same, then what you see is a revealing of the original source image. When you change the probabilities of either one of the walkers independently, they become outofsync. This makes the resulting image become incrementally more abstract. 

fft scan  The next step in our progress was to attempt to modify the behaviour of the stochastic walker dependent on the frequencies of the source image. Probabilities are weighted according to high frequency content to make the walker attract to edges in the image. What you see here is a red square bobbing around on screen along the edges due to its stochastic nature.  
The FFT analysis is performed directly above, below, to the left and right of the x,y coordinate of the walker. 

worm walker (no alpha)  A combination of fft scan and random pixel spitter. It runs a single stochastic walker on the source image and an number of independent stochastic walkers on the display screen, where by each destination walker on the display is moving according to the Fast Fourier Transform analysis as described above. 


worm walker (with alpha blitting)  This is the same as above, only now the blits to screen are made with a transparency setting. This shows the history of the walks on screen. 
