Abstract

Current supercomputer simulations of complex physical phenomena are generating massive data sets that continue to grow in size. One approach to harnessing the rich information content in such data sets is the use of large-scale display environments, including, large tiled display walls and virtual reality caves. Tiled displays can show fine detail while keeping the entire model in view, however, techniques for interacting with such displays are somewhat undeveloped.

The most common interaction techniques are standard mouse and keyboard input for large-scale 2d applications and 3-dof input devices for immersive applications. Our aim with this project is to develop a vision-based interaction system which uses cameras aimed at the display surface and a laser pointer for interaction with the display.

Currently there are 2 interaction tasks that we are targeting. The first comes from the need for flexible interaction with 2d applications. Sitting at a small terminal while trying to control the applications on the display wall is often non-intuitive and inconvenient. Our project makes use of the laser pointer as a mouse position on the tiled display. Adding buttons to a laser pointer (or using a standard remote control with built-in laser pointer) provides all that is need for full, untethered mouse based interaction with the display. The second target is one that comes from the need to construct displays at minimal expense. The main area of focus on large tiled-displays is often a small subset of the entire screen. While it is desirable to view data in very high resolution, it is wasteful for an entire display be at that resolution while only paying particular attention to a small region. For this reason we are developing a system which uses a single high resolution projector to project an inset that renders a portion of the displayed image in resolution much higher than the rest of the display. By making use of a pan-tilt unit, we are able to move the location of this inset. We use the laser pointer tracking system interact with the system and specify the location of the inset.

Acknowledgments

Contact