Update (2/17/10)
I have updated the project. The footprints are now smaller, and after several optimizations, the flash does not slow down. Thanks for the tips everyone!

Note: The flash is above: it begins as complete whitespace.
Controls
left arrow/esc: reset
up arrow: spawn footprints quickly
down arrow: spawn footprints leisurely
numbers 1-9: level of independence, or “propensity to carve one’s own path”
clicking and dragging the mouse across the screen: suggest new paths

Zip file containing flash and noResizeHtml page:

link
(zip this to your desktop and open the html file to view).

Initial Murmurings – Conceptual Development
I have spent a lot of time walking through snow recently, and so my first idea was to procedurally generate lots of different kinds of footprints. The idea was to create all kinds – reptilian, mammalian, hooves, paws, claws, etc – some similar to existing animals and many created through random chance. As I began to work on this project, I began to think about how the footprints should travel across the page.
I noticed when walking through snow that it’s much easier to take a path already traveled, even if it doesn’t lead directly to my destination. I decided to work on modeling this behavior, where footprints tend to follow pre-existing paths. I dropped my original idea because I felt this new development was interesting enough and different shapes and sizes of prints would clutter the screen.

Underlying Tech
To create this behavior, I initialized a 40×30 vector field that spans the screen that would gradually be altered as footprints traveled across it. For example, a footprint traveling east would influence the vector field in the locale of the footprints in an easterly direction. Secondly, I added the vectors local to the footprint as a weighted average to the footprint’s original trajectory – thus, it both influences the vector field as it travels across it, and its path is influenced by the field. As time passes, vectors which coincide become stronger, leading more footprints to more firmly stick to the path. Eventually, clear routes are
visible.

Playing with the Program
I allowed the user to manipulate several parameters. Namely, they can influence the vector field using the mouse, allowing for users to create their own path, either on the fly or before running the simulation. They can also change how “independent” the footprints are using the number keys – highly independent footprints will not follow preexisting paths as much while highly suggestible footprints will almost immediately join previous tracks. Both of these can lead to a variety of terrains appearing with their own subtleties. The controls are listed near the top.

Looking Ahead
The next step, I believe, is to render it in three dimensions, with mountains and valleys dynamically forming from traveled paths. With 3D topography, I think the project would become aesthetically inspiring as terrain formed in front of you.
Secondly, there are a number of other behaviors that could be modified or added. For example, currently footprints will tend towards the most commonly traveled path, seemingly forgetting (or never having had) a destination to begin with. This leads to river-like behavior. However, with a set destination, perhaps road-like routes would form.
Another consideration would be to add an interface to allow people other than me to use the program. However, since the project is currently in its infancy, I think simply describing the controls is sufficient.

Jon Miller