Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!zaphod.mps.ohio-state.edu!uwm.edu!ogicse!milton!hlab
From: lance@motcsd.csd.mot.com (lance.norskog)
Newsgroups: sci.virtual-worlds
Subject: Re: How about VR on $1 a day?
Message-ID: <1991May21.011155.15840@milton.u.washington.edu>
Date: 20 May 91 23:24:32 GMT
References: <1991May20.011348.7025@milton.u.washington.edu>
Sender: hlab@milton.u.washington.edu (Human Int. Technology Lab)
Organization: Motorola CSD, Cupertino CA
Lines: 42
Approved: cyberoid@milton.u.washington.edu




3rd-party 8514 cards are now $500 here in Silicon Valley.  They
should be fast enough to do animated displays, and include painter's
algorithm polygon fills.  E.g. you copy from bitmap A to bitmap B,
the chip uses an outline in rectangle C to decided when to actually
write in B.  The problem with 8514 is the chip sets that I've studied
don't support low-resolution clocks; you can do full-color with one
buffer, 16 colors with dual-buffer screens, and no quad-buffering.
Stereo VR needs quad buffers, two for display and two for drawing
the next frame.

The most interesting card I've seen is from Antex in Gardena, CA
(213-xxx-xxxx).  For $900 you get VGA, NTSC, stereo 12khz 14-bit 
sound, 1mg RAM, and a TI DSP chip.  With an assembler loop on the
TI doing sound management interspersed with drawing from a
2D display list, and a 386 feeding it the 2D display list 
and sound start/stop/position commands, you could do a really
nice low-end VR system.  

As for tracking, this Polhemus stuff is for the birds.  It should
be easy to put a few ultrasonic or infrared transmitters in the
corners of the room, and place 2 receivers each on your head and hands.
You've already got a wire bundle feeding the glasses, more
wires from the receivers is fine.  PC lab cards that could
read these are $200-$400.  If all the transmitters transmit
at different frequencies, you can just pull the frequencies out
with a multiple notch filter circuit and feed those into the lab
card A-Ds.  You already have a 386/387 (or in my case a 486-33, smirk)
for doing transforms.  You have to decide which signals are 
are obscured or reflected, build XYZ for all receivers, then extract
roll/pitch/yaw from the receiver-pair deltas.

You need to build a model for dealing with reflections of your 
ultrasonic/infrared transmissions.  Ray-tracing has been used in
this modeling domain.  

You're welcome to build and sell from these ideas.  I want one
and I flunked electronics.

Lance Norskog