LEE NN Se 
DE 
Ces = 
== Seo SOR e 
Lo uuu ooo eee 
es = ATUM, 
bez SS 
     
    
    
   
  
  
  
  
  
  
  
Figure 3. The predictor is seen in use over a Digital Terrain 
Map representing the North Atlantic. The predictor shows 
future position based on current velocity inputs. The tails of 
the predictor show the path through the data space. 
Variations on the flight metaphor are other more 
constrained vehicles. These typically limit the freedom 
in some way that is suitable to some application. The 
key aspect to these metaphors is that the user feels that 
he or she is controlling a vchicle which can be used to 
explore the terrain (or other data space). The viewpoint 
is fixed to a vehicle which can then be driven around 
the environment. Bicycles in virtual environments are 
an example of this metaphor. Another is a treadmill 
used to simulate architectural walkthroughs (Brooks, 
1986). 
3. DATA MANIPULATION IN 3D 
The problems of data manipulation and viewpoint 
manipulation both involve six degree-of-freedom task 
performance. It takes three numbers to position an 
object in space and three more to specify its 
orientation. There are many solutions to the problems 
of 3D object placement, including virtual turntables 
(Evans et al, 1981), or virtual trackballs (Chen et al, 
1988) whereby the mouse is used to control a set of 
higher level widgets which control rotations and 
translations. We have had considerable success using 
the Bat, described above for free object placement. 
However, in many GIS and CAD operations we find 
that it is better to used constrained placement. Often 
the user wishes to only manipulate one degree of 
freedom at a time, a rotation about a particular axis, or 
a single axis translation. Therefore, whereas we use the 
Bat extensively for viewpoint manipulation, we 
generally use a conventional mouse for object 
placement in the applications we are currently working 
on. 
4. THE FLEDERMAUS SYSTEM 
We (Mark Paton and myself) are currently building a 
prototype 3D Geographical Visualization System at the 
University of New Brunswick that incorporates many 
of the advanced interaction techniques talked about in 
this paper (it is called Fledermaus). Our primary use of 
the flying technique is in a data visualization and 
editing system for oceanographic research. This 
enables users to fly over a digital terrain map 
represented in thinned wire frame form, The thinning 
is necessary because it is impossible to render the 
entire surface at the highest resolution and at a 
reasonable upgrade rate. As soon as the user stops 
flying, the surface fills in two steps: a medium 
resolution step and the highest resolution step. The 
idea is to keep adding information to the user and to fill 
in the otherwise unacceptable delay required to 
compute a fully rendered high resolution image (on our 
current IRIS Crimson VGX™ it takes 16 seconds to 
render a 2000x2000 digital elevation model). In order 
to make a high quality video of a flyby it is only 
necessary to hold down a different button on the Bat 
and from that time forward the sequence of viewpoints 
is recorded while flying. These viewpoints are later 
used to generate high resolution frames which are 
captured one at a time on our Panasonic LQ4000 laser 
disc recorder. The videos we have made in this way 
(using previous versions of the system) have appeared 
in the Canada pavilion at the recent Seville World's 
fair, on Canadian and German National Television as 
well as at numerous scientific and technical 
conferences. 
We are currently adding Georeferencing, to the 
visualization system and it now has support for 
multiple objects. This enables us to, for example, 
compare two different DEMs of the same surface, by 
co-registering them and moving them up and down 
with respect to each other. We have also recently added 
support for closed polygon meshes with the goal of 
visualizing schools of fish in the water column. We 
are currently investigating various techniques for stereo 
visualization and by the time this paper is presented 
should have stereo movie capabilites. 
ACKNOWLEDGMENTS 
I gratefully acknowledge the support and assistance of 
Dr. Larry Mayer who has been instrumental in 
encouraging the practical application of laboratory 
ideas. The research described here has been primarily 
funded through NSERC grants. 
REFERENCES 
Brooks, F.P. "Walkthrough - A Dynamic Graphics 
System for Simulating Virtual Buildings," Proc 
1986 Workshop on Interactive 3D Graphics, F. 
Cro 
pp-$ 
Chapma 
Prec 
Virt 
Con 
Syn 
Can 
Chen, M 
Inte: 
Con 
No. 
Connor, 
D.C 
Dim 
Intei 
of C 
Evans, K 
Val 
of fr 
Vol. 
Gibson, 
Perc 
Gleicher 
Cam 
Sigg 
Mackinl. 
"Raf 
3D. 
Sigg 
Phillips, 
"Au 
mani 
Grap 
Marc