Graphics Reference
In-Depth Information
or sensing devices. In general, what is the best way to populate virtual environments?
A recent additional level of complexity arises in the context of distributed and
multiuser environments. See [CMBZ00] for some comments on networked virtual
environments.
An alternative to VR is augmented reality (AR). This is a hybrid approach where
one displays three-dimensional virtual objects in a user's natural environment. Typi-
cally, a user would see the world via a head-mounted camera and a display with virtual
images superimposed onto the real scene. Some applications of AR are showing pro-
posed buildings in their actual setting, interacting with machinery, and medical data
for patient. For example, AR has been used to superimpose internal ultrasound data
onto a patient.
There are of course many connections between VR and geometric modeling. One
area where VR has stimulated research in geometric modeling has to do with the fact
that one often deals with large models that have to be displayed quickly in interac-
tive environments. Decimation algorithms come in handy here, that is, one wants to
remove data that is not really needed for accurate display. See the end of Section 14.3
for a few more details and some references.
It seems that currently, in addition to the standard architectural walkthroughs,
virtual prototyping, and medical, therapy, and entertainment applications, IVR is
having its biggest impact on scientific visualization where the huge amounts of data
one has to cope with makes it hard to grasp its meaning. The current successful appli-
cations are rather specialized however and one is left with the question of when IVR
will be used in the more general purpose sort of way that its strongest advocates dream
about.
16.3
Geometrically Intelligent Modeling Systems
Modeling systems have come a long way. A variety of paradigms have been developed.
They were discussed in Chapter 5, but it is clear that overall what we have is a lot of
ad hoc approaches, some of which work reasonably well within their domain. A uni-
fying structure is missing in CAD/CAM/CAGD. Of course, the ad hoc nature of current
state of geometric modeling is not surprising, given the short time that computers
have existed. Furthermore, it is important to have something that “will get the work
done” now. The niceties of having a unifying “theory” are things that can be indulged
in at some future point in time. Besides, there will always be special problems that
are solved in special ways. Nevertheless, it is the author's dream to see a truly general
and complete modeling system that will understand all the geometry that mathemat-
ics understands at the time.
There are really two basic parts to CAD/CAM/CAGD systems: the underlying geom-
etry and the interface between it and the user. Since the geometric coverage is
adequate for current users and industry, if one overlooks the constant efforts to make
algorithms and data structures more efficient, it is not surprising that most of the new
work is related in one way or another with the user interface. The topic [ShaM95]
explains how, by looking at the design process, one can create feature libraries and
automate some common model manipulations. The authors refer to features as being
based on an extensional view of design, where one records what practitioners com-
monly do. They argue that one also needs to capture the intentional aspect of design,
Search WWH ::




Custom Search