91-11/W.review Report on presentation by W Industries. Leicester Polytechnic Branch Meeting of the British Computer Society, 9th October 1991 Chris Hand, Department of Computing Science Leicester Polytechnic, Leicester, UK cph@leicp.ac.uk (c) Chris Hand, 1991 Introduction On October 9th 1991 Leicester Polytechnic was visited by two representatives of W Industries, one of the major forces in the UK's rapidly growing Virtual Reality industry. The talk was presented by Terry Rowley, Simulation Director of W Industries. Also present was Sales Executive Chris Yewdall who answered some questions and ran the demonstration at the end of the presentation. A Definition of Virtual Reality Terry Rowley began the presentation by talking about our senses and how we use them to interact with our immediate environment. He followed this with a definition of VR as being the substitution of sensory inputs. He added that this must be coherent to avoid sickness. The substitution of sensory inputs may be performed in many ways; visually, film or video are familiar media, while through sound, the use of headphones to create a sound image in space is common. Technology has also become available to provide input to the sense of touch, through pneumatic tactile gloves and force-feedback techniques (eg. the steering wheel on a driving simulator). Some difficulties arise with the sense of touch, however. For example how do we simulate the feeling of wetness when dipping our fingers into virtual water. The simulation of temperature is also a problem: how do we make the temperature of the simulated water feel right? Other senses are altogether more problematic. Simulation of smells, although quite possible, is difficult in an interactive environment due to the need to dispose of smells that are no longer applicable in the current situation. More research needs to be done on smell - for example discovering the Primary Elements of smell (analogous to Red, Green and Blue in light) to allow us to create any smell necessary. The sense of taste may also be hard to simulate, but even so: since we usually swallow what we taste, would we actually want to taste a virtual object? History Mr Rowley then went on to discuss some of the history of VR, and his experiences with simulation technology. He lead us from the early experiments with stereo film, wide-angle screens, 3D glasses (and associated headaches - literally) to the development of Flight Simulators, in which he played a part during his work at Marconi Radar Systems Ltd, England. A range of visual simulation techniques were described, from using film and anamorphic lenses to the use of a video camera suspended over a model landscape. This latter approach was very expensive to make, and had its problems in the fact that when the camera moved nearer to the model the end-user's image became out of focus, whereas in reality closer proximity results in greater clarity. Other problems with depth of field and the model landscapes melting under strong lights or being gouged out by mis-navigated cameras meant that this approach was costly and inconvenient. Simulation techniques benefitted from the advent of computer generated images, although in the earlier days these were of low resolution. Mr Rowley showed slides of early military trainers that compromised by providing simple silhouettes of aircraft rather than attempting to display detailed images. When increased computer power arrived in the 1970's and 80's, colour, shading and texture became possible in real time. In the 70's a simulation system developed by Marconi (in Leicester) was capable of rendering 5000 textured polygons/s in real time at a rate of 50Hz. A video clip taken directly from such a system showed a harrier jet performing vertical take off, before disappearing smoothly into the distance. Such a system cost between 2 and 20 million pounds. Mr Rowley pointed out that although these simulations were only available to a few, it was during this time that much of the groundwork was laid down and techniques developed that would be used later. Moving on to more recent times, we were told how such devices as the TMS 34010/20 40Mflops graphics co-processor can be used to create images at high speed, and how LCD colour screens can be used for head-mounted displays (for example VPL eyephones and W Industries headset). Since such devices were capable of being mass-produced, the result can be a large saving in the potential cost of realistic simulation systems. The Rise of W Industries Terry Rowley next turned his attention to the history of the company of which he is Simulation Director: W Industries of Leicester, UK. The early work was due to Jonathan Waldern at Leicester Polytechnic, who worked there with Professor Edmonds in the HCI research unit (which later moved to Loughborough University). A video clip from an early 80's episode of the BBC TV programme "Tomorrow's World" showed some of the early work of Dr Waldern. The "Roaming Caterpillar" (as it was known) was shown being used by presenter Maggie Philbin to examine a virtual room. The display was a large B/W monitor with handles on each side, supported by a flexible arm on castors covered by a concertina of rubber (the "caterpillar"). The image displayed was a wire-frame view of the "contents" of the (actually empty) room. The location of the monitor was sensed by 3 fixed speakers emitting audible clicks in a fixed sequence, with the time taken to reach microphones being used to calculate the distance. It was also possible to detect the position of the user's fingers by wearing tubes on them, with microphones and wires attached to a control unit. By pinching her fingers together in the right place, the presenter picked up the receiver of a virtual telephone and left it suspended in space. A stereoscopic view of the room was also possible by wearing a head- mounted visor with a rotating shutter, a separate image being presented to each eye on alternate frames of the video image. Although slow at the time, this system allowed further experimentation. John Waldern was joined by Al Humrich - another colleague with expertise in graphics - and then Richard Holmes (an ex-Rolls Royce employee) and Terry Rowley from Marconi. These four decided to put their money where their mouths were, and in October 1987 founded W Industries. Although the four had to start by constructing equipment in their garages, by January 1989 they acquired and moved into their own premises. They started with a head-mounted display, tethered at the top, with handles on a small screen (LCD?). This version didn't get very far off the drawing board. Their second HMD, another tethered system, was known as the "giraffe". This featured mechanical 2D head-tracking. Some games were written that used the device. Mr Rowley showed a slide featuring a white plastic unit, not entirely unlike a giraffe (!), with the wearer reaching out into virtual space (despite the lack of a hand-tracking device). The legend on the side of the arm read "W Industries 3d vids". The fourth device to be built was much slimmer, featuring a magnetic head-tracker and twin LCD screens. It is of interest to note that the head-mounted part of the device allowed for adjustment of inter- occular spacing. At the time it was thought that the correct adjustment of this distance to suit each viewer would be vital to the stereoscopic effect. Mr Rowley pointed out that it was later found not to be so important, as long as the exit pupil of the optics was made large enough. The fifth head-mounted display/tracker is the current W industries system as seen in many a smoke-filled publicity photograph (see later for details). In early 1989 W Industries received an award for "best emerging technology" which brought a prize of GBP20,000 and, just as important, a good deal of publicity. They were approached by a large leisure firm who bought 75% of their equity. This gave the leisure firm majority control, but the resulting input of 1 million pounds was greatly needed. In November 1990 the "Stand-Up" unit was launched at the Computer Graphics 90 exhibition staged at London's Alexandra Palace. At the same time the parent company went into liquidation, but fortunately W Industries had been sold just prior to this to another company, which also owns the Wembley Stadium in London. The "Sit-Down" unit was launched, at Wembley, in March 1991. With ever-gathering momentum, WI moved into their own premises of 22,000 square feet in an industrial park in Leicester in May 1991. These premises are still only partly occupied and are currently used for development, assembly and testing of the units, known as "Virtuality". Inside the Virtuality System The system comprises a User Interface (sensors, headset etc.), Host Computer, Software and Console. * User Interface The key element of the user interface is the visor (or "visette"). This contains a sensor for 3-dimensional head tracking, along with the colour TFT LCD screens which create the stereo image. The inputs to the screens are simply RGB signals created by the graphics boards. The screens provide a resolution of 276 horizontal by 376 vertical pixels, with the graininess of the image created being removed using a "spacial convolver". The screens are actually mounted on the sides of the user's head, to avoid making the visor front-heavy (balancing of the unit being very important since it has to be able to be used in a Video Arcade environment: safety of users when moving the head quickly from side to side is paramount). This side- mounting technique means that the optical path has to be folded, with the image passing through a collimating lens. The headset also has a built-in microphone and quadraphonic sound with 2 speakers on each side of the head. The sound image in space is "vectored" by the host computer. A headband is provided which is adjustable to allow the user to fix the position of the unit relative to the top of the head, to create the stereo graphical image correctly. A clamp pulls down and fixes the unit on the head at the back: no chin-strap is used for safety. The visor can be quickly released in an emergency by simply tapping on the clamping bar. This unit will, apparently, fit 90% of users. With the possibility of 200 people per day using the visor, hygiene is an important factor. Bacteria, fungus and livestock (eg. head lice) could all be a problem. The Glass-reinforced plastic headset unit is supplied with hygienic paper "wipes" which can be used to clean it between users. During health and safety testing, the headset unit had passed non-transmission tests with many bacteria and virii, including HIV. When not in use, the visor rests on a plastic "head" built onto the main unit. * Host Computer Known as "Expality", this is a custom unit with a CD-ROM drive and floppy disk. Currently the CD-ROM contains only sound-effects and background scenes, while the floppy disks are used to transport the games software. Currently the CD-ROM contains 5 games; a purchaser of the games unit must buy a separate key-disk for each program they wish to run. The computer -- actually an Amiga 3000 motherboard -- handles tracking of sensors, sound and music (from samples). Rendering of graphical images is performed by two custom graphics boards (one per eye), using TMS 34020/082 at 40Mflops. Mr Rowley told us that the unit could render a maximum of 30,000 polygons/s at 20 screen updates per second. * Console This is simply a glass-reinforced plastic shell that protects the user and the electronics from each other. The driving game shell looks like a car (it has a lift-up lid like a car bonnet giving access to hardware). The stand-up unit has a cushioned "bench" (the electronics are underneath) on which one sits to strap on the visor/electronics. A small unit carrying cabling is tied round the waist and fits in the small of the back. The cables carry video and sensor signals to and from the headset. * Software W Industries call their programs "virtual experiences". These are developed on desktop workstations using the same hardware as found in the virtuality unit itself. The specification of each game is reviewed by "Games Consultants". The specification and creation of these experiences includes - Considering vehicle dynamics - Network interaction (for team games) - Creation of scenery and images (models are digitized with video cameras from many angles) - The roles and behaviour of "Actors", which may be computer-generated or other players. The software must have start/finish sequences to instruct a (probably naive) user how to put on/take off headset etc. Mr Rowley commented further on the developing VR market, suggesting that the stand-up units would be suitable for CAD as well as entertainment, although he doubted the present resolution would be adequate for virtual medical surgery. We were told that the stand-up units are being sold to those educational and research institutions that can afford them. Furthermore, not much "real competition" exists in this area since, according to Mr Rowley, WI had the first real custom-made VR units on the market. Next we were shown a WI promotional video. Proclaiming "10 years of development towards Virtual Reality", the video featured clips of two women using sit-down units (with joy-sticks), followed by a demonstration of the stand-up units by two males. Other clips showed the data glove and "force feedback" glove being used for Desktop CAD (ie. non-immersed). Both these gloves can be worn on the same hand simultaneously. After some more nebulous promises about the "possibilities of Virtual Reality", the video ended with the remark that W Industries' offering represented a "Production VR system at PC prices". In (non-virtual) reality, these prices turn out to be from GBP17,500. A stand-up R&D system with full stereo and touch-glove would cost GBP40,000. When the video had ended, questions were taken. Questions and Answers Q. Criticism that the present software (eg. driving game) doesn't encourage the user to "look around" and so make good use of 360 degree environment. A. It was admitted that this was true to an extent, but a new game due at the end of October 91 was mentioned. This features a 30-ft high slow-moving robot biped. The user controls this using a steering wheel and foot-pedals. Other (networked) users can be shot using head-guided laser beams. We were told that this actively encourages head-movement when tracking prey and when avoiding ambushes. But essentially these games are market-driven, so if the users (or arcade owners) want a driving game, they get one. It was also mentioned that a hang-gliding simulator was produced for the promotion of a new after- shave. This encouraged looking around since more time is available than in a high-speed racing game. Q. What is the development time of the games software? How long to develop an application from scratch? A. Utility software can be bought from WI for fast development of a virtual environment. The general rate of progress is very fast. It was admitted that while a Flight Simulator program had been written in 10 days, it then required several months to get it "polished". The new Walker game was said to have taken around 2 months from start of writing to reach the alpha version being shipped to sample customers. The final version would be released 4 weeks after that. The cost of a "custom experience" was put at GBP6-10,000. An example is the hang-glider simulator created for the after-shave company. Routines were developed for the simulation of seagulls that fly around the hang-glider when the user reaches the sea. It was pointed out that this code can then be re-used -- the seagull routines were later used in a modified form to create a Pterodactyl! Q. What languages are used for software development? A. There exists a high level "virtuality graphics system" (actually libraries) used in-house. Development takes place in "C" or Amiga assembler. Q. Has a new hardware base been considered? A. Maybe. WI are looking at something that would allow a possible increase in power of a factor of 10-100. Q. Would this use custom hardware? A. No -- it's cheaper to develop products using off-the-shelf hardware. Q. What is the time delay between user movement and display update? A. The electromagnetic tracker operates at 30 - 120MHz. Any delay is mostly due to the graphics chip (a 25MHz device (?) using a 33MHz clock). Lag is 60 or 70ms. Mr Rowley admits that one has to compromise over the limitations of hardware in some situations. Q. What compatibility is there with existing CAD systems? A. A DXF file transfer facility is available. Q. How many systems are being used in non-leisure applications? A. One organisation in France intends to use WI units with Silicon Graphics workstations for computer-generated puppetry. In Italy systems are being used for the rehabilitation of children after hand surgery, using datagloves. Another unit is being used by the University of Delft in the Netherlands. Q. Will a cheap "home version" be available? A. Moves are afoot to run a version off a home computer. W Industries are "looking into it" Demonstration For the last 50 minutes of the evening Chris Yewdall supervised use of the "stand-up" games unit they had brought along. The rush to "have a go" was overwhelming to say the least, and with the length of each game being 3 minutes (this is hard-wired into the code!), progress seemed slow. While waiting, the eager crowd had chance to chat further with Mr Yewdall. WI apparently have a non-disclosure agreement with "a chip-set manufacturer" regarding some new graphics chips to be used in the future. Yewdall said that WI were the first to put a TMS34020 in "a product" and as such "ended up doing a lot of debugging" for TMS. Mr Yewdall also said that WI are working closely with Matsushita who supply the LCD screens: better resolution displays could be easily substituted since the whole visor is a modular design, with the edge connectors allowing quick replacement of devices. Impressions Due to lack of time I only had chance to use the head-unit briefly. The colour and rendering seemed fairly smooth, with the frame rate being quite adequate. The main disappointment was the field of view: at only 70 degrees the unit gives a feeling of looking down a tube. Apparently the field of view is limited to 70 degrees by the combination of screens and optics used: spreading the image any wider would distort the image unacceptably. Game Description Several games have been introduced by WI for use with their sit-down units in video arcades: these include a flight simulator and a driving game. The unit demonstrated after the presentation -- a stand-up system including a "free" joy-stick -- can also be used for games. One of these is described below. Having donned the visor and pressed both the joy-stick's trigger buttons the game starts with one of its many digital samples: "Time to Die" from Blade Runner. The user "walks" (actually moves using the joy-stick: not actually walking) around the play area, which is a set of grey platforms suspended in outer space -- a suitable background of stars is provided. Staircases join lower platforms to higher ones. Looking at the joy-stick through the visor it appears as a gun. Extending one's arm shows a virtual arm rendered in bright pink polygons. To add a competitive element, the platforms are patrolled by a lean gun-slinger (looking like a renegade from the Dire Straits "Money for nothing" video) who walks by and suddenly turns to shoot, spinning and bending his knees as he fires. Fortunately the bullet moves slow enough to be able to avoid it, eg. by ducking. Standing on one of the yellow triangles found at edge of a platform is equivalent to calling the elevator: a tray hovers over from a nearby platform making a PING sound when it arrives. Walking onto the tray causes it to return to its platform, taking the user with it. The idea is to shoot the gun-slingers before they get you. To add further complication to this otherwise undemanding task, a more sinister enemy awaits. After firing 9 shots of the gun a ghoulish voice announces: "Birdy's hungry!". After the 10th shot, a large green pterodactyl ("Birdy") flies overhead -- accompanied by wing- beating sounds -- and tries to grab the player. If Birdy is shot at the last minute at close range you will be successful. Otherwise you can expect to be picked up in Birdy's talons, carried to a virtual height of 200ft above the platform and dropped on your head. When this happens, users see themselves falling from a third person's viewpoint. An interesting point about all WI games is that due to worries about the psychological effects of "being killed" virtually in the game, the user is taken to an out-of-body viewpoint to remind them that they are not really there. For example when a car crashes in the driving game, the driver flies up in the air looking down at the wreck. They can then see the car piecing itself back together before finally being flown back down into the driving seat to start again. Conclusions An interesting and thought-provoking evening for all concerned. A testament to the amount of interest in this area is that a larger venue had to be found at short notice to accomodate the number of people attending. It was evident during the presentation that this industry is still at the embryonic stage, and that advances in technology will soon bring usable VR systems within the grasp of many of those who currently find the cost prohibitive. By placing their initial emphasis on entertainment-oriented mass-produced units, W Industries will be one of the companies responsible for bringing VR to the people.