almquist@cis.udel.edu (Squish) (12/27/90)
Recently I posted an article about the possibilities of a deck being created
very cheaply with today's technology. I have been informed of some possible
alternatives. First, for a recap: VR can become reality NOW. All we need is
a distributed virtual reality fileserver (which would contain the database
about world(s) and provide message routing and SOME CPU power for decks
connected to it) and a deck (this is a cheap, fast piece of hardware that sits
between the user and the VR worldserver and interprets the information to and
from the deck - depending on your budget, the deck can be a stupid ascii
terminal all the way up to, perhaps beyond, a system that handles position,
tactile (in the form of pressure and temperature), visual, auditory, motion
queing, and limited keyboard I/O). SO, we need a VR worldserver and a deck.
The optimal system that I envision would consist of the following:
o SUN Sparc SLC mini-motherboard w/ 8MB memory
o SUN's sound chip could remain or be replaced with technology used by NeXt
o anywhere from 1 to 6 serial ports for: space position identification,
tactile, visual, auditory, motion queing (directed air jets optional),
limited keyboard I/O
o one scsi port for optional local data caching, local storage, etc.
o one network interface port - ethernet? fiber optics? you chose
o small keyboard, lets say a standard keyboard minus function keys and numeric
pad has approx. 62 keys so, we create a SMALL keypad with 20+ OVERSIZED keys.
This key pad could be strapped to your left thigh (seeing that the glove is
right-handed) or on your arm like the small one in the ABYSS. Until graphic
engines and CPUs become fast enough, why slow down one's performance? Give
'em a glove and small keypad. This could/would speed up one's response
times. I/O would be over a serial line (that's all that would be needed).
o dataglove of some type. I/O would intially be over a serial line (future,
fiber optics?).
o eyephones or display of some type. I/O again would initially be over a
serial line (future, fiber optics?).
o small headset with headphones and 3d positional tracker. Two more serial
lines.
o (optional) tactial feedback using the concept of inflatable air pockets/
cells. Gas or liquid used to inflat pockets could be heated or cooled to
add temperature sensing - could be very valuable for air/liquid flow analysis
- you'd see the pretty colors and you'd be able to feel an associated
temperature - I'm picturing that neat picture of the shuttle on re-entry.
Again, one serial line needed - more added or a good scheme created in the
future to handle all the information in a timely fashion - again, future,
fiber optics?
o (optional) for the rich people, limited motion queing in the form of directed
air jets and/or a motion platform/chair/etc.
What are we talking here for price?
SUN mini-motherboard, 8MB, sound chip minus the monitor, one ethernet port,
one scsi port, two serial lines, at educational discounts approx. $1500? It
cost approx. $2500 with monitor at educational discount rates.
Small keypad using serial line to connect to system - $30.
Dataglove. Tough question - I feel that the PowerGlove could easily be
improved to meet our needs - hey, $100 plus $200 improvements is still cheaper
than VPL. So, approx. $300.
Eyephones, tough one. Looking at GameBoy or the Lynx system I feel that we
could create a cheap, affordable eyephone-like system for around $1000.
Headphone and 3d positional tracker, also tough. Perhaps using the optical
gyro technology we could get this down to $1000? I can't properly price this.
SO, were talking a VR deck for under $4000. IF we had GOOD VR software
available, that would perhaps replace PCs and low-end workstations. By
software I also include good software development tools. One of the
necessities for VR is to make it able to be easily and quickly modified to meet
a given users need. Mom and pop or Mister Physics isn't going to want to
spend a long time (years?) developing a world to play in. Options, hardware
graphic engines, tactile controllers and compressor, air jets and/or motion
platform, etc.
Another possible route to take was brought to my attention by Dan Wilcox:
From: "D. Wilcox" <wilcox@cfht.cfht.hawaii.edu>
Subject: Re: A deck DOES exist - Just change the packaging ;-)
SUN does make a VME-based SPARCengine 1E. It has 1 S-bus slot, 8M
of memory, ethernet, SCSI, serial port, an optional additional card with up
to 64M of memory and an optional graphics S-bus card (either color or mono).
There are a plethora of cards available for the VME bus. It's fast, modular,
and available. It is not exactly low cost (~$4000/card, depending on the
complexity). But I/O is cheap (<$1000 for 64 channels), etc. Available for
VME are many different CPUs, RAM, video controllers, frame grabbers, cases,
power supplies, etc. You could easily assemble a VR system without having to
build any hardware.
Check out Control Engineering VMEbus Buyers Guide ($44.95) from:
The Microcomputer Interface Group
1350 E. Touhy Ave.
P.O. Box 5080
Des Plaines, IL 60017-5080
(800) 992-4447 (have your VISA card ready!)
You don't like the SUN approach, how about this from Erich Boleyn:
From: erich@eecs.ee.pdx.edu
The NeXTdimension board is also great (it has the NTSC ports, plus the i860
RISC chip on it). I have also heard a fairly substantiated rumor that NeXT
is planning to use Renderman as the imaging scheme for the 3.0 operating
system (with an optimized version for the i860 coprocessor, of course ;-).
Two of these boards plus some eyephones, datagloves (what kind of connectors
do these use? The NeXT has two RS422 ports, or even SCSI, if supported), and
software using NeXT imaging commands for the NTSC section of the screen
(in 24-bit color, if the screens support it) would be hot. Just make your
own window manager that runs under the server that would have you logged
into a VR-based system, instead of the normal workspace manager, all neat
and modular on top of Mach. I don't know, NeXT architechture, if not as
well-used right now, seems to have a pretty good VR-type setup already
available (NeXT cube + 2 of the NeXTdimension color boards, or NeXT cube + 1
or 2 of the boards being worked on in Seattle, Wash. right now). A cube
with big hard drive would run you ~7000-8000 right now (plus a memory
expansion, maybe), then the two color cards (with an i860 on each one) for
~3000 apiece. It does seem a little expansive though.
Upon further discussion, a few of us have decided that a VR deck can be built
very cheaply using today's technology. ALTHOUGH, one thing that keeps coming
up: SOFTWARE.
From: "D. Wilcox" <wilcox@cfht.cfht.hawaii.edu>
It would sure be nice if someone would put together a VR system from this
stuff (the SUN VME stuff), because then other peolple could build it also
with a minimum of effort, at least hardware wise. Since these systems can run
UNIX, OS-9, VxWorks, MS-DOS, X-windows, etc, there should be quite a lot of
software and expertise to bring up a VR system quickly.
I'm a software person and I can see the development of necessary software
coming to life soon. BUT, I don't know what work is going on for the hardware
aspect of VR. Any lurkers out there wish to enlighten us?
From: erich@eecs.ee.pdx.edu
>Yes, but I'm still thinking that a VR deck could be made using today parts
>for around $3-5k.
So do I, but it would be somewhat against the normal tactics of a company
to make such a thing (though don't ask me why, it would be a smart move that
would get a lot of supporters). I dunno, maybe you could seek out a company
and ask them to make or adapt an existing hardware platform that would go
well with the software, then together would make a great product... (say,
maybe we could convince NeXT of this, yeah, that's the ticket ;-)
Anyone have ideas? Cheaper solutions? A working solution? Working on this
problem? Etc. Etc. Etc.
- Michael Almquist
"Engineers bring ideas to reality,
Virtual Reality Idealists bring reality to ideas!", Squish, Dec. 90'