From: richard@.ist.ucf.edu (Richard Dunn-Roberts)
Subject: Re: VR and CIG's
Date: Fri, 28 Jun 91 20:22:49 GMT
Organization: VSL, IST, UCF, Orlando


In article <1991Jun26.164251.15308@milton.u.washington.edu>, galt@es.com (Greg
Alt - Perp) writes:

> A while back, someone from Florida (Univ. of Florida?) said that they were
> using an ESV for VR.

The work you are referring to is taking place at the University of Central
Florida's Institute for Simulation and Training, in the Visual Systems
Laboratory.  We are working with two channels of Evans and Sutherland
ESIG-500 image generator as our visualization platform, and plan to have
the first world on this system up and running by the end of August.  We are
interested in networking issues associated with distributed virtual
environments, so this work will use a heterogenous suite of workstations for
various components of the virtual world.  Computational platforms included in
the first world using the ESIG (we've done previous worlds on SGIs) MAY
include SGIs, Sparcs, and a Harris NightHawk.  This is not intended to
be a low-cost VW ... we want to throw everything we can at the problem, and
then determine a reasonable subset of capability/functionality for
low-cost VW's.

In article <1991Jun25.154814.27460@milton.u.washington.edu>, jraymond@bbn.com
Jayson Raymond) writes:

> Also, most CIG's are frame driven, as opposed to the
> run-to-complete processing of typical general purpose systems.

and

> Also of benefit are the now emerging standards.  Database
> standards allow database's to be shared and hosted on other hardware
> platforms. And networking standards are being established so that anyone
> can "plug-n-play" into existing networks of simulators.

Jayson is correct that IGs have a lot of benefits for use in VR.  However,
highly interactive simulations such as VR still present some inherently
difficult problems that IGs cannot handle in current incarnations. Some of
the problems of using IG's include:

     - limited polygon budget.  Most high-end systems are capable of
     providing on the order of 5K-6K textured polygons at 60Hz frame
     rate.  Cost for these systems is on the order of $1M per channel.
     For lower end systems, polygon budgets of 1K-2K at 30Hz or less
     is common, and it is real easy to exceed this budget with a world
     of any complexity.

     - Dynamic models on IGs are very limited.  For example, BBN's SIMNET
     has moving tanks and aircraft that can interact. However, the terrain
     database is static.  This is actually a much larger part of the
     database than the moving models.  This helps reduce correlation
     problems, but also reduces the number of interesting things you can
     do with the database.  Ever tried to hide a tank without first having
     a bulldozer dig a big hole in the ground?  This is related to the
     polygon budget problem.

However, IGs will probably be used a lot in commercial VRs, especially
those used for training simulations.  There is quite a bit of research
and industrial development being done to try to solve these problems.
For example, here at IST we are coordinating the efforts to standardize
distributed interactive simulations (DIS) network protocols.  This effort
is being sponsored by the DoD. DIS is discription of the data packets to
be exchanged by simulators of differing fidelity (ie. E&S, GE, BBN).
Currently the DIS standard is in its second draft and is under review
by many industrial simulator vendors.

Finally, there has been development in software for database construction.
Newly developed IG's, such as GE Comp VI, have database capabilities built
into their rendering engines.  This takes database development into a
different arena.  For example, say there is this nice little town in
California that would be a great place to visit for the weekend.  Instead
of flying there, go to the library, get pictures of the place (preferably
from overhead) and feed them into the database production software.  The
result is a textured world ready to run a VR system.  This capability
already exists and is in use in for some military applications.  Currently,
this capability requires human help, but new image processing and computer
vision algorithms will help make this process automatic.

---------------------------------------------------------------------
Brian Blau, Visual Systems Scientist              blau@ist.ucf.edu
Richard Dunn-Roberts, Visual Systems Scientist    richard@ist.ucf.edu
---------------------------------------------------------------------
Visual Systems Laboratory
Institute for Simulation and Training
University of Central Florida
12424 Research Parkway, Suite 300
Orlando, FL 32826-3271

(407) 658 5073/5074
---------------------------------------------------------------------