OBJECTIVE | SCENARIOS | TECHNOLOGY | process | SOURCES | TEAM

 

Technology
Installation
living-room augments a given room of 3.10 x 3.75 x 2.45 meters. This room is decorated with markers that enable visual position tracking. The living-room book contains the collection of different scenarios. Looking at the marker of a specific scenario causes a scene change.



Hardware
living-room runs on an Apple G5 with 1.5 Gbyte RAM. Two Fire-I unibrain FireWire cameras record the environment, one is needed for position tracking and the other delivers the background image for the AR scenery.



Software and Implementation
ARtoolkit is used for visual position tracking in the room and the recognition of the scenario changes. The scenarios are implemented in Max/MSP and Jitter. ARtoolkit was expanded with functions to adapt the position tracking to the specific situation of living-room and to communicate with Max/MSP over sockets and the OSC (Open Sound Control) protocol.
In Max/MSP/Jitter a special patch was developed, the living-room engine, that contains the basic functionalities needed to run the different scenarios, like the initialisation procedures, the rendering of the background image, the socket connection to ARtoolkit, and the scene-change function. Every scenario is a further patch that is loaded and unloaded by the living-room engine patch. In the individual scenario patches the geometry, sound, and dynamic behaviour of each scene are defined.

Little Portable Living Room
A special application was created to allow the development of scenarios independently of the physical living-room. The interface enables the user to change camera position and look-at point. The data is sent to Jitter using the OpenSoundControl protocol in the same way as the ARToolkit component.



Living room wallpaper
After experimenting for a while with 9 markers on each wall (see topmost image) it became obvious that we needed more markers and bigger ones; more markers to solve the problem when no marker is visible because the user is too close to a wall. The larger markers improve position determination, when the user is far away from a recognized marker.
A script was written to generate “wallpaper”, since it would be very tedious to mount every marker individually on the walls of the physical living-room. The script, written in Perl, generates postscript files, which can be plotted, and creates the marker configuration file, with the position information needed by ARToolkit. Different parameters can be set in the script, i.e. the size and number of markers and their distances from the edges. Currently we have mounted 5 markers in height and 6-7 along the length of the walls.