From: blau@.ist.ucf.edu (Brian Blau)
Subject: IG's and VR's (was Re: Regarding (C)IG's ???) 
Date: Mon, 15 Jul 91 20:17:01 GMT
Organization: VSL, IST, UCF, Orlando


Regarding (C)IG's, jraymond@BBN.COM (Jayson Raymond) writes :

>
> Although dynamic terrain isn't currently available, at least in Simnet,
> to the best of my knowledge it isn't anymore inherently difficult in an
> IG than a workstation, it just hasn't been done (funded) yet. The same
> problems still exist for workstations and IG's, problems of broadcasting
> this data to the network, and updating other systems databases, or
> working from a common database.

IST has a project called "Dynamic Terrain" and has been working on problems 
like for the past two years.  We have developed workstation based software 
which lets a bulldozer "dig" up the ground.  It runs on the parallel 
processors of a Silicon Graphics machine.  The bulldozer is interactive 
(about 10-15 hz).  Additionally, a water model was incorporated.
During this summer, two of the students in the lab added the hydrological
model described in "Rapid Stable Fluid Dynamics for Computer Graphics" in 
SIGGRAPH'90 to the dynamic terrain programs.  We can build dams and then 
break them down.

Additionally, this project is moving to a distributed platform.  The next
versions will let the terrain database be distributed across machines, locally
at first, then long haul.  The main task here is to evaluate network
performance and amount of traffic based on how the database is distrubuted.
Also, transmitting the database is some efficient data packets is a top
priority.

> In any case there are many advantages to utilizing a CIG over a
> workstation, namely because they are built specificly for the purpose of
> generating and working within polygonaly represented worlds. There are
> some disadvantages as well. Such as, some systems don't offer a very
> friendly development environment for the end user, at least compared to
> what one may be use to on a workstation. But as VR system creators
> design their systems and evaluate their needs, I highly recommend at
> least considering an IG.

There are many new software programs that give simulator database developer
are large amount of freedom.  A package called Multigen from Software
Systems is an execllent package for database development.  It can read
all of the favorite DMA data formats. It has all of the nifty stuff like
levels of detail, texturing and hierarchial models included.  The format
of the database is called "Flight" which is public domain.  Other packages
such as Target from GE give the database developer another level of detail.
Target has the ability to manipulate textures that has never been demonstrated 
before. Target is also able to reconstruct 3D objects from 2D images.  This 
package is intended for use by database developers for IG's, but at some point, 
these kinds of abilities will be needed by VR world builders.  

The two packages which I just mentioned are the so-called "state-of-the-art"
in IG database tools.  As reported by some people who work with these tools, 
" ... these database tools (Multigen, Target, etc.) were not created by 
those who need to use them on a daily basis, thus lack depth and 
maturity ... ".  This is not a thrashing, but an observation of some users.
Other database tools for IG are BBN's S1000, E&S MetaFile Editor, Syndbad and
TopGen (a MultiGen derivative) are about the only products on the market
which are available for this purpose, and are probally suited for first round
world building.  

There are other packages which may be suited better for world building.  These
are the tools used by computer graphics animators.  These tools are Alias,
Wavefront, ElectroGig and NeoVisuals.  In the long run, new tools which 
combine 3D reconstruction capabilities with modeling power will 
be needed to construct virtual worlds.

Currently, building worlds is an art.  Most of the worlds today have been hand 
crafted, and the development tools are designed for a specific VR system.  
Softare packages like Target and Alias will let world builders have freedoms
they have never experienced before.

> It sounds as though UCF IST is working on some great projects, and I'm
> glad to see a none commercial interest coordinating standardization
> efforts. Also, I hope you don't mind that I have taken the liberty of
> forwarding the Distributed Simulation Workshop announcement to this list.

Thanks for the posting.  

Brian Blau
Visual Systems Scientist
 
###########################################################################
# Brian Blau                             ## Phone :                       #
# Visual Systems Laboratory              ##     (407) 658-5513 or (5073)  #
# Institute for Simulation and Training  ## Email :                       #
# 12424 Research Parkway Suite 300       ##     blau@ist.ucf.edu          #
# Orlando, Florida 32826                 ##     blau@osceola.cs.ucf.edu   #
###########################################################################