Bill Keays, Ron McNeil
The metaField Maze is a virtual maze game which uses a 12’x12’ interactive floor. People interact by walking onto the floor and directing the path of the marble through the labyrinth with the movements of their bodies over the floor surface. The maze, which is projected onto the floor, is a three-dimensional model which tilts according to where the player moves. The marble moves according to the direction of the tilt. The goal is to move the marble through the full course of the maze.
This installation uses a multi-purpose interactive floor system created by the authors known as metaField. The metaField was developed in an effort to create an immensive environment that had the following qualities: it should be suitable to interactive art or game applications; it should have the ability of accomodating both solo and collaborative activities; it should involve the kinesthetic input from the users; it should have a low threshold of engagement. From these criteria,a floor-based configuration was decided upon using video projection and a vision system.
The metaField consists of a video camera and projector mounted 30 ft above a 12’ x 12’ retro-reflective floor surface. The camera and projector are focused on the floor, and are attached to a Silicon Graphics O2 computer. The retro-reflective material creates a desirable high contrast between the people and the floor when viewed from above. This contrast facilitates the task of locating the people standing on the floor.
The software driving the system has two distinct components. The first is vision software known as glimpser,which accepts the incoming video stream and draws basic information from it. The second takes this information and uses it to control the display. The glimpser has the ability to find regions of a specified size and specified color from incoming video frames. The glimpser and display software run separately, but simultaneously.
The display software begins by telling the glimpser what sort of image data it requires. For the purpose of the metaField Maze, it is required to find regions of black that correspond to 4 inch squares at floor level. This granularity is chosen to allow for the recognition of the limbs of the subject while not being so fine as to slow the performance of the glimpser. Once the glimpser is calibrated to the specifications of the display software, a steady flow of location data moves between the two. The display software incorporates a three-dimensional model of a maze created in Open Inventor; this maze is projected onto the floor. The 3-D maze model is tilted on X and Y axes depending on the location data provided by the glimpser.