Virtual Reality – revolution or rhubarb? (October 1998)

PDM’s editor Neil McNaughton spent someconsiderable time wearing the funny glasses at last month’s SEG conference andlistened to the evangelists. He came away not entirely convinced.

At one moment, alongside a crowd of fellow virtual explorers immersed in the sub-salt section of the Gulf of Mexico I almost believed that maybe there was something beyond the hype, that just perhaps, virtual reality (VR) did have something to offer the E&P community. But that was an ephemeral conversion, after all a good sales pitch is a good sales pitch – there is probably an individual down at your local Chrysler outlet who would at least come near to convincing you that you actually need something as improbable as a utility vehicle to drive down to the mall.

As soon as I took the specs off and began to listen to the increasingly improbable claims from the proponents of VR, I returned quickly to my old skeptical self. Lets first consider what is VR really is. VR should allow the user to perceive computer generated objects in a realistic manner, and to interact with them equally realistically. Note that I say perceive and not ‘view’. Complete VR should ideally involve all of our senses, but technology is such that while computer screens are commonplace, real-time generation of smells is not yet available. Most VR implementations just involve 3D viewing, perhaps with some motion/force feedback device. One product actually offers only 2D viewing, but claims VR-ness because the viewer is inside a hemispherical screen. I have achieved a similar effect by wearing two pairs of (regular) glasses and pushing my nose close to my (2D) monitor.

VR may be imitative and attempt to recreate reality, such as the cockpit of a 777 in a flight simulator. VR can equally be analogue, where we can perceive and interact with objects to which we would not ordinarily be exposed. Most E&P applications are in the analogue domain, it is hard to imagine what we would perceive if we were actually down at 5000 m below sea level. Uncomfortable pressure and heat mostly, and of course light doesn’t travel too far through rock. But we are already used to the transparency paradigm that allows us to see partly through 3D volumes of the workstation, or the layered object approach where well bores and selected horizons, faults and so on can be isolated and visualized.

What about the interactivity. I have thought about how to interact with an analogue of a rock, and I must admit that I have not got very far. Hit it with a virtual hammer? Nah. Stroke it to see how rough it is? Nah again. Thrust your hand into the heaving shales. Nope. In fact the only ‘interaction’ that is actually on offer is the walk through and look around. You move within the body of your data, behind fault planes, looking up well bores and so on. This is really very limited ‘VR’ when you think about it, little beyond our current technology. In fact the few ‘feedback’ devices in evidence seem to be used as very unwieldy 3D mice, for pointing and clicking at hard-to-reach floating menus. Some vendors do propose a rather amusing sound interactivity; as we slide down the borehole, a Telstar-like warble modulates as a function of the sonic log velocity.

The crucial interactivity is that as you move and turn your head, the scenery adapts in real time and presents you with the required view. Now this is all very well if it were free, but it ain’t. The kit required to perform this sort of exercise is extremely costly. What is worse is that the compute effort involved in adapting to the turning head is so great that it stretches supercomputers to the limit – and I would argue, beyond. In fact to present even a rudimentary 3D moving picture, considerable image quality has to be sacrificed. A good sized workstation screen presents around a million pixels to the viewer (a megapixel). A large – several meter wide - VR display actually makes do with little more than this, that’s all it can handle. Contrast this with the 100 megapixel-up resolution of the lowly inkjet printer and you see that this is really yet another example of the dumbing-down of our data in the cause of IT ‘efficiency’. Additional sacrifices are necessary in terms of color depth, further diluted by the need to feed binocular viewers.

To what end we ask? Actually we did ask, we emailed all the VR vendors and asked them the simple question, "what can you do with VR that was impossible without?" We have not exactly been overwhelmed by the response. In fact, at the SEG’s Visualization workshop, the most telling example of ‘use’ of high powered visualization was the application of very dubious science to what amounted to a grandiose management presentation. And that is unfortunately where most use of VR probably lies – in presentation and sales. Soon it may be de-rigeur to show your prospects to management in this way.

Going back to the sales pitch for VR itself I found one claim, of a reduction in cycle time for a 3D survey from 4 weeks to 3 days completely over-the-top. I mean, anyone who could reduce data loading for 3D interpretation by that amount would have the data management equivalent of the philosopher’s stone in their hands. Could we actually be talking about ‘interpreting’ data that has already been .. interpreted?