This thesis presents a framework consisting of intuitive metaphors and techniques
for realizing smart displays with the help of scalable projections in everyday office
settings. Optical I/O bricks containing cameras and projectors are combined and
blended in a flexible way to achieve increasingly immersive, interactive environments
which overcome the shape, placement, collaboration and interaction restrictions
imposed by current computer and home entertainment systems.
The enhanced functionality of our interactive visual workspaces relies on a novel
technique of a per-pixel light control, which exploits the micromirror modulation
pattern used by Digital Light Processing (DLP) projectors and allows for imperceptible
structured light approaches. Displays can be activated on-demand on any surface
extracted from the camera images and dynamic changes are continuously
accounted for by a recalibration procedure based on an imperceptible per-pixel light
control handling both the cameras and the projectors in a unified way. Flexible,
freeform displays which allow for a space-efficient visualization are provided. They
continuously sense the environment and automatically adapt to collision events.
Using encumbrance-free tracking techniques we allow for a unified natural interaction with the bare hands, pens and laser pointers. An intuitive interaction metaphor with standard applications is presented, replacing traditional screens,
keyboards and mice in a flexible and effective way. Concurrent users and both
single-handed and bimanual input are supported for a direct manipulation and a
multi-touch gesture-based operation triggering. For an increased resolution, a contents personalization and an interactive visualization, the users can dynamically augment the large-scale projections with mobile high-resolution foveal enhancements using a pocket light metaphor. To further optimize the presentation at a given resolution, the design of the displays can be modified interactively, and like a jigsaw puzzle, the layout can be customized using an adaptive compositing approach. The set of techniques is complemented with a support for smart light-sensing tags, which can extract imperceptible information transmitted pixelwise in the projected workspace. Furthermore, using passive tags or faces captured by the cameras, we also present automatic display reinitialization approaches. To validate our novel display techniques and metaphors, several aspects of our interactive visual workspaces are evaluated in an extensive user study.