Research project P 13867 Real-Time Visualization of Urban Environments Michael GERVAUTZ 11.10.1999
This project aims at the creation of an integrated solution for modeling and real-time visualization of large and
medium-scale urban environments as a basis for applications like traffic, driving and architectural simulations,
information visualization and location based entertainment. We want to investigate new efficient modeling and
rendering methods to create a system that makes it possible to
* model large three-dimensional environments of up to a few tens of square kilometers in reasonable time.
* allow interactive visualization of this environment. A user should be able to navigate through the environment at
about 20 frames per second.
Our modeling approach is based on basic geometric, semantic and topological data typically found in CAD and GIS
databases. Visual programming is used to specify how a database is translated into a three-dimensional computer
graphics model. The main focus of modeling is to allow for the creation of a model that is suitable for interactive
rendering on low to high end graphics hardware.
To be able to maintain a sufficient frame rate during the rendering of larger environments, we want to investigate
new rendering techniques, which involve occlusion culling, ray casting and image caching. These new algorithms
shall be combined with existing acceleration methods to obtain an optimized rendering pipeline which makes use
of both graphics hardware and CPU power and is in its nature output sensitive.
Our main contributions are:
* a new modeling framework built on database translation and amplification, that allows the user to specify the
model creation in a data-flow oriented visual programming user interface.
* a new, faster algorithm for occlusion culling combined with a new algorithm for image-based rendering
incorporating image caching and ray casting that are combined to a rendering pipeline for urban walkthroughs.