From: edlantz@aol.com (Edlantz)
Subject: INDUSTRY: New VR Display Classes, etc.
Date: 20 Aug 1996 20:04:14 -0400


This is my summary of the Siggraph panel discussion "The Future of VR:
Head Mounted Displays Versus Spatially Immersive Displays."  Anyone with
an opinion on this topic is encouraged to post on this thread (the panel
that wouldn't die...)  -  Ed Lantz

***********************************************************
Organizer: Ed Lantz of Spitz, Inc.
Panelists:  David Bennett of Alternate Realities Corp., Bertrand de La
Chapelle of VIRTOOLS, Steve Bryson of MRJ/NASA Ames, David Zeltzer of MIT
Research Lab of Electronics, and Mark Bolas of Fakespace

The viability of spatially immersive displays (SIDs - projected VR
environments such as the CAVEtm and domes) for certain VR applications
was re-affirmed by all panelists.  It was agreed that human-to-human
communication is essential for many real-world applications such as
CAD workstations.  SID's can deliver a VR-like experience to a greater
number of participants at once, making it more cost effective for
group entertainment as well.  Also, with SIDs, the graphics engine is
not burdoned with re-rendering constant head rotation updates - but -
a larger instantaneous field-of-view does have to be rendered,
requiring more graphics pipes (good for SGI, bad for cost).

HMDs (and head-coupled displays like the Fakespace's Push(tm) and
Boom(tm)) make more sense for single-user systems where direct,
hands-on interaction with the VE is required.  SIDs make more sense
for multi-user systems where indirect interactivity is OK, or multiple
users are passively viewing the interaction of a single HMD user.  One
exception is a multi-user space, especially where avatars are used to
represent HMD users who are not co-located.  But it was mentioned that
the subtlety of human communication (facial gestures, body language)
would be difficult to convey using an avatar with current
technologies.

Mark Bolas dissented from calling a non-stereoscopic display "VR."  I
personally disagree, as stereopsis only provides highly useful 3D cues
in applications where near-stationary objects are being manipulated at
close range (admittedly, alot of VR applications do this... but not
all).  Other 3D cues including motion parallax/shear, object rotation,
shading, depth cues such as fog and object size, etc. are very
effective and can create the illusion of 3D immersion when employed in
a SID.  Also a SID provides, by definition, accurate angular
(look-angle) representation of the VE, a very strong spatial
navigation cue.  Stereopsis is hardly useful at all when flying over
terrain at a distance, or for fast moving objects.  Add the eye strain
factor of conflicting cues (vergence vs accomodation), the hassle of
eyewear, etc. and you will find that many venues will not want to
bother with a stereoscopic SID.

In my opinion, one of the most important outcomes of the panel was the
adoption of a third general VR display class: the Virtual Model
Display or VMD.  This includes the ImmersaDesk(tm), the Virtual
Workbench, and the Responsive Workbench which are best suited to
representing 3D object models rather than fully immersive workspaces.
Although varying degrees of visual immersion can be obtained in
flat-screen desk-type displays, these displays really shine when a
model such as a human body is displayed in its entirety.  Here there
is no need to create a "sense of presence" within a room.  Rather, as
Bertrand de La Chapelle put it, the model "emerges" from the display
rather than the user being "immersed."  Thanks to Steve Bryson,
working with Mark Bolas, for coining this new display class.

I believe that these three display classes, HMD/HCD (head-coupled
display), SID, and VMD, are useful tools for matching display type to
our customers' requirements.  Often, display issues are confusing to
VR customers and a handy-dandy classification system clarifies the
tradeoff issues.

A list of SID/VMD display manufacturers and integrators can be
obtained by emailing me at Edlantz@aol.com.  I will gladly add anyone
to this list, or make corrections as required.

Thanks to all the panelists, and to you who participated in the panel at
Siggraph or via USENET.

Ed Lantz  
Spitz, Inc. 
P.O. Box 198  
Chadds Ford, PA 19317  
tel: (610) 459-5200 x27   
fax: (610) 459-3830     
email: edlantz@aol.com     
www.libertynet.org:80/~spitz  
                                                            
* Spitz:  fifty years of visual immersion  *