3D City - Adaptive Capture and Visualization of Cityscapes
by Thomas Jung and Ines Ernst
The project 3D City is concerned with the development of an innovative
technique for visualizing cityspaces.
The capture of existing cityscapes is normally an extremely time-consuming
matter. It involves the three-dimensional surveying of objects
and their provision with surface textures for the purposes of
visualization. Generation of the models requires a great deal
of subsequent reworking by hand. The large three-dimensional models
can only be rendered in reasonable quality by using high-end graphics
computers.
On the other hand, there are techniques that allow panoramic shots
to be visualized interactively in photo-realistic quality, eg
QuicktimeVR. Here, however, the viewer cannot move freely, but
only turn and alternate between different camera positions, as
no depth information on the panoramic shots is available. Adding
depth information to such shots should allow the viewer to move
freely in cityscapes that have been captured photographically.
A new computer-graphics 2 1/2D rendering technique allows the
generation of images for viewer positions from which no photograph
was taken. On the basis of image and depth information, the rendering
algorithm decides which object parts are visible from the current
viewer position and renders them according to their appearance
in the corresponding panoramas.
Using several panoramic shots, the depth values for individual
pixels or image sections can be automatically reconstructed. Specialized
tools allow depth information to be added semi-automatically for
further image sections.
What emerges is a visualization environment with the following
features:
- photo-realistic rendering quality: since the basic information
is obtained from photographs, the image quality is comparable
with that of multimedia environments such as QuicktimeVR
- free navigation in the parts of the scene that are rendered in
the pictures taken; no fixed camera positions
- moderate computation time for rendering; it increases only slightly
with the complexity of the scene
- use of modern 3D graphics cards, the techniques employed being
based on 3D algorithms
- compatibility with 3D graphics: the technique can be used eg for
rendering high-detail static components in VRML (Virtual Reality
Modeling Language) scenes.
Possible application scenarios include:
- simple generation and updating of 3D city maps or 3D information
systems
- helping architects by speeding up the job of drawing up tenders
- capturing the state of building and redevelopment areas
- supporting architectural presentations by enabling planned objects
to be rendered in a natural, lifelike setting.
Please contact:
Thomas Jung - GMD
Tel: +49 30 6392 1779
E-mail: tj@first.gmd.de
