Project Description

To advertise their new sport utility vehicle, Volvo hired a branding agency who commissioned a Flash game where the car was driving down a rugged mountain road, picking up outdoors gear such as tents, rope, backpacks, etc. However, this game was not intended to be played in a web browser like most Flash games. The plan was to show this game to movie theater audiences in twelve cities in the UK. The game was controlled by the entire audience collectively by leaning left and right in their seats to steer the car. It was specially arranged for there to be a simultaneous show time in all of the cities, so that they could compete head-to-head and be ranked against other cities on a scoreboard.

My Role

The idea of this game being controlled by the leaning of a movie theater audience was mostly likely based on the work I did on this technology as both an undergraduate research assistant and as a graduate student. Audience Interaction, as we called it, was originally a research project by Daniel Maynes-Aminzade in the Stage3 Research Group that I continued to develop. We experimented with a number of collaborative and competitive interfaces for large audiences to interact with projected video games, of which leaning was a particularly successful one. I had built these computer vision algorithms into a library that was used by at least six semester-long projects at the Entertainment Technology Center as well as many two week long projects in the Building Virtual Worlds course for which I was part of the support staff for several years.

Although the Flash game for this Volvo project was already created by the Flash animation/interactive studio Chunk, I was contracted to add the audience control. Since I was brought on late in the project, a challenge for me was to adapt my framework to design and technical decisions which had already been made. Because the Flash game was a mostly closed environment, the audience control component had to be developed as a separate application which could seamlessly communicate with the Flash game and also composite live video of the audience over the game's graphics. Although previously the leaning had often required manual calibration, for this installation it was important to develop an auto-calibration routine that did not require manual interaction.

Since the event was scheduled to take place simultaneously in 12 cities, I was not able to set up system in each location. Instead, I had to prepare 12 portable sets of equipment and train 12 hired volunteers who would travel to each city and operate it. We had a very narrow window within which to train each operator, have them travel to their respective locations, set up and test each location individually, do a full dress rehearsal of the networking of all twelve locations at once, and execute the event. Particularly tricky was that because we were working in movie theaters, we had only about 15 minutes at a time between movies to set up and test our system. A large part of my work on this project was managing our operators, testing each set up, updating the software, and troubleshooting technical issues, all remotely.