Michael Frank Deering: VR: SPAR VR
I’ve been interested in VR since I first heard about Ivan Sutherland’s pioneering experiments in 1968. However, for many years the technology just wasn’t up to the task. So when I was at Schlumberger Palo Alto Research, as soon as the first handheld LCD TV’s became available, I immediately pushed for building a color head mounted display (HMD). Along with others, we built our own HMD, and also experimented with what became known as “fishtank VR” using a normal workstation color display screen in stereo. The work at Schlumberger was conducted by Andy Witkin, Henry Sowizral, Howard Davidson, myself, and Jay Freeman.
The first handheld LCD TV’s available in the United States were from Seiko, and were quite expensive at approximately $900 each. (Casio didn’t enter the market until sometime later with a cheaper but lower resolution product.) These early devices were generally considered cute but over priced gadgets. Macy’s (which was still selling consumer TVs) decided to reduce their inventory a bit and put the Seiko TV’s on sale for approximately $600 each. I convinced Schlumberger management that this would be a good thing to do some experiments on, and bought three of the devices at the sale. The idea is that we would probably destroy one taking it apart, but Howard Davidson not only took one apart without breaking it, but also found the internal NTSC video input to the LCD display system (TV’s in those days didn’t usually have external video inputs).
After removing the cases and unnecessary electronics, we then re-mounted two of the displays on a helmet for wearing on your head. To re-focus the displays at something closer to optical infinity, I bought some Radio Shack flat plastic frensel magnifiers and reversed them.
We tried really hard to get a Sun workstation to generate two video signals to be converted into NTSC and put into the headmount, but in those days it was not possible have multiple color displays. So as an interim “kludge” we did what it turned out most other people did as well: we rendered the left and right eye onto two windows on the same Sun screen, and then pointed two inexpensive NTSC video cameras at one window each, and then fed the camera video output signal into the headmount. This sounds pretty awkward, but given the low resolution of the Seiko displays, the cameras weren’t the resolution limit.
For head tracking, we used (again as others did) the Polhemus 6-axis magnetic tracker. The headaches of dealing with that tracker gave me a lifelong aversion to any trackers with significant lag, relative or absolute error, ill-defined interfaces, and metal interference problems. (This is not Polhemus specific; I have issues with most all trackers ever manufactured.)
When all this was put together with a simple rendering environment (what else? a cube in space) one could stand with the headmount on, and by turning and moving one’s head “look” around a virtual environment. We used a second Polhemus sensor as a 3D wand, and eventually acquired one of the first DataGloves from VPL research for our system. (Aside: many years later at Sun, I bought the assets of VPL, and donated the original prototype VPL DataGlove and BodySuit to the Computer History Museum.)
The Fishtank Stereo Display
Besides the headmount display, we also experimented with rendering stereo images onto a normal workstation CRT display, but with the two eye’s viewpoint controlled by the 6-axis tracker. We called this “fishtank display”, because that’s what Howard Davidson remembered 3D displays in 1940′s science fiction novels being called (usually tactical displays on space ships bridge during battle). (Later when I re-introduced a high quality version of these displays at Sun, other people in the VR field asked me what they should be called, as I was generally credited for inventing the modern workable version of them. I could have grabbed for some immortality by suggesting “deeringTanks” or some such, but I just repeated the name Howard Davidson had suggested (I had read the same novels).)
In order to make stereo display work with the Sun rendering board, we had to get at the stereo buffer swap vertical retrace sync signal. This signal was not available as an output from the Sun rendering board, but we obtained the circuit diagrams of the board from sun and found the right point to tap in. It was this experience that caused me to insist on a stereo sync output signal on ever 3D graphics board I later designed at Sun, starting with the GT.
Initially the Fishtank stereo display ran the same demos as the HMD: a cube sitting in space. Later we tried prototyping up some more CAD like applications.
While Schlumberger had for most of its history been an oil-field services company, in the early eighties they acquired a number of electronics companies in a diversification effort, as oil has always been a cyclic business, and while electronics was as well, up to that point in history it has tended to be counter-cyclic with oil. (Unfortunately for Schlumberger, as of the eighties electronics switched to the same cycle as oil, and it turns out that most electronics companies that were available for sale (like Fairchild and Applicon) tended to be not the most healthy.)
So much of the job of Schlumberger Palo Alto Research was look up to five years out for new product opportunities for the electronics businesses of Schlumberger. OK, so where would Virtual Reality fit in? While Applicon’s mechanical CAD/CAM business might first come to mind (and was one we followed up on, especially with the Fishtank stereo display), another early opportunity was in electronics test, measurement, and repair. It turned out that some of the market for test equipment was for doing 2 day turn around repairs when British warships came into port for very brief servicing. The idea of mapping all test signals from several different specialized instruments into one augmented reality head mounted display, allowing much quicker debugging and repair of complex circuit boards.
What Happened to all This?
For many years Schlumberger Palo Alto Research (SPAR) was a wonderful environment to work, and I am proud to have been part of most of its history. While SPAR never really achieved the high goal some had set “to be the Xerox PARC of the eighties”, if you look at what the alumni of SPAR have gone on to do, it is nevertheless quite impressive. (I won’t mention all the people here, but will point out that just one of groups to come out became CISCO.) Unfortunately Schlumberger’s attempt to apply their management magic to the electronics business did not work out very well, and by 1987 they had decided to retrench mostly to their main oil services businesses, and to shut down SPAR. The VR work at SPAR happened towards the end, and thus no publications (or any PR on it at all) were made. But the experience of having built these early systems passed on to the organizations that people dispersed to, including my work at Sun, and Henry Sowizral’s work at Boeing (before I hired him to work on Java 3D at Sun).