Flashback: Mike Roberts on Some of the Unsung Heroes of VR’s History

by Matthew Terndrup • July 23rd, 2015

These opinions do not necessarily reflect those held by the editorial staff at UploadVR. 

Virtual reality is about hit the mainstream, and it’s about to hit it hard. It is showing up everywhere – from being an integral part in large scale conferences to showing up at small, close knit meetups. Practically everywhere you look, VR is booting up to alter every industry it touches; but where did it all come from? Why is virtual reality proliferating into the market like never before?

In a series of interviews and re-publications, we venture into the past to see where this exciting virtual reality ride began. Today, we interview Mike Roberts who witnessed the rise of VR firsthand in the 1990s. Roberts obtained a Ph.D. from London’s City University, where he worked in the area of parallel computing, visual programming, and graph theory. He has worked on a number of advanced virtual and augmented reality projects over the years and holds several patents related to the industry.

As the medium took off, Roberts came face to face with the wild experimentation that was happening at the time. Parties, events, and even Burning Man excursions hovered around the environment – pushing the innovation as far as it could go.

Through a several email communications, we asked Mike Roberts what the culture what it was like then and how it is similar/different from what is occurring now. So sit back, relax, and enjoy this exciting flashback in time.

So Mike, what was your first virtual reality experience like?

I’m originally from the UK, so my first experience of seeing real VR was sometime about 1990 with the Division stuff. The inception of VR in the UK was slightly behind the US, although my experience with computer graphics goes back further than this.

With some clever programming using both the main 6502 CPU and the 6522 IO processor on the BBC micro, it was possible to do very basic wireframe type graphics on microcomputers the early 80’s, as seen with games like Elite.

Comparison between 1993 Division dVisor vs 2013 Oculus Rift [source]

Comparison between 1993 Division dVisor vs 2013 Oculus Rift [source]

The UK was a hotbed of 3D in the late 80’s and early 90’s, which accounts for the large number of UK games programmers in circulation. 3D in general made it into the UK university curriculum a lot earlier than in the US; at least that has been my impression.

During my Ph.D. work in the late 80’s the Inmos guys like Steve Ghee had turned me onto reading William Gibson, and I saw code for Steve’s early work at Inmos on rendering pipelines using Transputer networks (the Transputer was an early British parallel computing chip).

Transputer Inmos T800 [source]

Transputer Inmos T800 [source]

At the research level, the state of the art at the UK research-level was systems with decent (16-64) numbers of Transputers, running Occam or C, and linked to Sun workstations (running the control software) and Inmos color graphics systems (for output). Some larger systems with thousands of processors were available in certain places, and a lot of core distributed systems work was done figuring out how to parallelize applications onto those systems.

You could do some real-time rendering and primitive ray tracing on the Transputer machines; people were doing a lot of more serious non-real-time 3D development like volumetric reconstruction, the type of stuff which was confined to the very high-end SGI machines in the US at the time. I don’t think the level of what we had (in the UK) at that time is well known, and a lot of this history is pre-web, so it’s not really available online in a very accessible form.

Steve Ghee gave me a bunch of sample T800 Transputers out the back door from Inmos to work with around that time, then went on to form Division and write a lot of their software. You can check out a great BBC video I found on YouTube, below, which shows the state of the Division demo as of about ’90.

I remember very distinctly hanging out with Steve and the crew drinking Genever (dutch gin) in Enchede, Holland at a Transputer User Group meeting in the late 80’s.

At that time I was working more in parallel and distributed computation (still an interest) with a side of graphics and in particular what became graph-based, node-based or flow programming – many of the ideas in that wider body of work associated with that formed the conceptual basis for a lot of the visual tools content tools you see in common use today, like the Maya Hypergraph and the flow-based programming in Max/MSP. By about 92-93, Intel and SGI had won, and the transputer systems were going into decline.

How did you get involved in the VR scene in the 90s?

I moved to California in early ’91 and spent some time working at a small startup called Mediashare, where I recreated a lot of my Ph.D. work on visual programming using graphs along with Art Whitten. We connected it to a multimedia engine based on-top of Smalltalk/C on the PC and Mediashare had quite an active group working on it for a while.

PC’s could not do 3D at that time – you needed an SGI or a specialist render box from Evans and Sutherland, or something of that sort. The Mediashare work became the basis of what became Kinetix (Autodesk) Hyperwire and was licensed to a couple of places, including Progress Software. I managed Progress’s software’s Graphical User Interface Technology group for a while and did work on the licensed code as well as some early real-time 3D node-based visualization code.

Around that time, I saw some of Ian McDowel‘s early Fakespace stuff at an event in San Francisco. That was the first stuff I saw from the US which I really liked, apart from Jaron Lanier‘s original stuff, which had a really human centric approach. I think that was at VERGE.

The BOOM (Binocular Omni-Orientation Monitor) from Fakespace is a head-coupled stereoscopic display device [source]

The BOOM (Binocular Omni-Orientation Monitor) from Fakespace was a stereoscopic display device [source]

On the Fakespace stuff, the original monochrome “Boom” and later color versions – the fact that the orientation tracking was basically done using hardwired sensors with very low latency meant you got a great sense of immersion with their gear. Also the CRT screens in their stereo display were very high resolution for the time, 1280×1024 per eye, or something like that.

Fakespace could get away with nicer displays as the weight of the display was supported by the boom, as compared to the HMDs, which were limited in terms of their CRTs due to the display weight. That stuff was running off SGI machines at the time. Display weight is no longer a major issue with LCD displays.

D’Cuckoo album cover [source]

D’Cuckoo album cover [source]

SGI had a great VR evangelist who was concerned with connecting together a lot of the early wave of VR work, Linda Jacobson, who much later came and worked at PARC for a time (she also wrote an early VR book – “Garage Virtual Reality“). Linda was in a VR band called D’Cuckoo which used to do performances mainly in the bay area. I’ve always stored that concept of a VR band for later reference. I think it’s a good one.

I saw D’Cuckoo at CHI in I think ’92 or ’93, and started attending SIGGRAPH pretty much every year after that; from there I met a bunch of the the quieter more thoughtful SGI connected folks like Paul Haeberli, and Helen Cho, who together used to run Fiat Lux, a sort of alternative conference for graphics/hacker/art folks and were a great influence to be around.

Paul Haeberli still has some great programmer manifesto stuff up which is worth a read and right on the mark for the time. Then of course there was Morph’s outpost magazine, the Digital Be-In, and whole scenes associated with those folks. Deanan DaSilva was another SGI connected person at that time, but more concerned with CG in general than VR. I remember seeing the SGI racer and speedracer aluminum chassis at Deanan’s warehouse, which was the codename for the machines that became Octane and Octane2, really important machines in the history of computer graphics. At that time, SGI would comp you machines, if they thought you were cool enough.

Via Ian McFarland (who was working at Wired/HotWired just when they started and knew my interests), I got hooked up to the VRML folks and was on the VRML list pretty much from the inception. I saw the first Labyrinth demo, via Mark Pesce, Tony Parisi, and Owen Rowley from Autodesk.

The VRML team (circa the 1990s)

The VRML team (circa the 1990s)

Peter Kennard, who now lives in NY had a major role in getting that demo together in C++ based on the RenderMorphics Reality Lab rendering engine. I’ve known Peter for what now seems like a huge length of time, but his role in the production of the original Labyrinth demo has I think been a bit forgotten. He wrote the first version of Caligari True Space on the Amiga, as well.

Tony Parisi, co-creator of VRLM [photo courtesy of Mike Roberts]

Tony Parisi, co-creator of VRLM [photo courtesy of Mike Roberts]

Flashback: Tony Parisi on Co-Creating the Virtual Reality Markup Language

RenderMorphics was another company formed from remnants of the UK graphics industry, by Doug Rabson (of FreeBSD fame), Servan Keondjian, and Kate Seekings.

Pretty much contiguous with the early VRML demos, RenderMorphics were acquired by Microsoft to form the basis of DirectX. Kate came to work in the US for MS, and was around in the scene in the 90’s, while the rest of the team stayed in London; she did some cool stuff like getting Douglas Adams to come to MS SIGGRAPH parties.

Continue reading on the next page: parties, Burning Man, meditation, psychedelics and more

Tagged with: , , ,

What's your reaction?