From: cdshaw@cs.UAlberta.CA (Chris Shaw)
Subject: Re: W Industries Introduces First Working VR Entertainment Unit.  
Date: Mon, 14 Jan 91 14:08:43 MST
Organization: U of Alberta, Dept of Computing Science


In article <14298@milton.u.washington.edu> Bob Jacobson writes:
>A W Industries unit, according to the company's brochure, costs
>approximately 20,000 pounds, or about $35,000 at current exchange rates.  

>The Virtuality system includes sight stimuli, with 65K colors,

16 bit frame buffer.

>dynamic self-shadowing, and 32 bit resolution from 50mm to 200,000km 

Great. Forward into the past. This is called throwing away floating point
and doing all your modeling in unsigned fixed-point math, just like Sutherland
did in 1967. This strictly limits the scale of "worlds" that you may want to
explore.  For example, let's say you bought a W box and suddenly got interested
in molecular dynamics. What do you do?? Scale everything up by 1,000,000,000,000
(or so) just to get your molecules big enough to not all render in the same
spot. Even for human-sized things this resolution is a little coarse. I
would have expected between 1mm and 5mm resolution.

>sound offers frequencies from 10Hz to 14KHz, 25 second duration. 

Should be careful what's being specified here. Are these samples? What's
the amplitude resolution? Sampling frequency? Or is it a MIDI device?

>The network operates at 10Mbytes in 1510 byte
>packets, accommodating up to 16 users. 

Sounds like Ethernet's nominal numbers. You never get anywhere near that in
real life.

>The computer system is
>apparently proprietary (though built of standard components), displaying
>30K polygons independently, transformed, clipped, and shaded (assuming
>50% back faced removed; the update rate is 50MHz.  There are four
>visual channels.  The system is currently operable at 230V.

More holes in the spec. How big are the polygons? Are they screen-aligned?
Are they rectangular? How many actually end up on the screen? (This is a
"marketing polygons" number, perhaps) Also, how big is the raster? What
kind of shading? Flat? Gouraud? Phong? 

So anyway. 15,000*50 = 1.5million polygons per second. Methinks another typo
is involved, and that's 15K polys/second, which is respectable, but by no
means stunning. I think that "50MHz" is actually 50Hz, which is British TV
vertical refresh rate. Refresh rate is not update rate.

By comparison, the i860 can nominally do 50,000 Gouraud shaded 100-pixel
marketing polygons per second, using 8-bit pixels, and you get floating point.

Actually, the W box is reminiscent of a system that Atari's Advanced
Graphics Division showed at SIGGRAPH 89. They didn't market it for whatever
reason, but it, too, was supposed to be 12K-15K polys/second, and
$10,000-$30,000 (I don't quite remember prices). It sounds like W has done
more or less the same set of hacks with more recent technology.

>The Visette weighs 2.9Kg (what's that, about eight pounds?).  

6.5 Lbs

>It uses stereo think film transistors, with 102K pixels; 

102400 pixels = 320*320, which is roughly 1/2 * 1/2 British TV resolution.
By the way, NTSC is 640*480. In the VPL EyePhone, you get 1/2 * 1/2 NTSC, or
320*240. The real questions are... can the W goggles accept NTSC, and is it
tough enough to go on tour?

>There are two models of Virtuality.  The first is a sitdown version that
>looks like the body of a small jet fighter in videogame configuration.
>I/O devices are two joysticks and various mode selection buttons.
>
>The second version of is a standup model.  "Experiences" are selected by
>buttons on a console.
>The experiences that are being demonstrated are apparently loaded into a
>CD-ROM and then accessed by the control system.

VR-in-a-can. Just add water & stir. Still, looks like fun. More of a video
game than a development type system, but what the hell!

>Waldren seems to have designed a useful low-to medium-range
>virtual worlds system that could become a valuable part 
>of a more comprehensive ensemble.  Is this impression well-founded?

Perhaps. The I/O is what's interesting. Also, the dual-signal stereo graphics
would be useful, compared to the network-style lashing up you have to do
with workstations. The rest of the graphics system is hard to judge, but it
looks like a very special purpose hack.

>Bob Jacobson

--
Chris Shaw     University of Alberta
cdshaw@cs.UAlberta.ca           Now with new, minty Internet flavour!
CatchPhrase: Bogus as HELL !


