Driving VR locomotion forward through the discovery of rectangular gain
The developer community has collectively faced many hurdles in pushing the current generation of VR from beguiling concept to consumer-ready medium. And one challenge has weighed more than most; simulation sickness. Feeling unpleasant while donning a VR headset is almost always a result of user movement – or more accurately 'locomotion' – where a disconnect between physical and virtual motion can cause a good degree of nausea.
Diego Montoya and Daniel Sproll worked together at VR and AR content production agency RE’FLEKT. A commission from Audi would teach them a great deal about that thorn in the side of virtual realty design; user locomotion.
As a developer and UX designer, Montoya and Sproll would see Audi take them on a journey through various ineffective solutions to solving user locomotion in VR, and lead them to conceive a new approach that virtual reality content creators should consider.
In theory, at least, Audi asked for something fairly simple. The German performance car manufacturer wanted to let its clientele explore a 1:1 scale model of a given automobile in VR, exactly set-up to an individual customer's requirements, right down to the paint finish and interior features. It needed users to be able to move around each car freely, and, with Audi attracting customers looking for a premium experience, RE’FLEKT's virtual showroom had to be polished, suitable for a first-time VR user, require no controller, and feel entirely comfortable.
“Comfort was our biggest design goal of all," said Sproll, who now serves as CXO & co-founder at realities.io. "We had to have no sickness, and maintain spatial representation. No sickness is a no-brainer. Maintaining spatial representation means while the user moves around the room, they should feel really like they are moving.”
The experience also needed to squeeze a four-metre by two-metre vehicle into a four-metre by four-metre capture area, while additionally providing scalability.
All of that meant many traditional solutions for VR user locomotion wouldn't work. Teleportation and blink movement, where users are carried in a flash to the final destination of their intended route, would feel unnatural. The same could be said of the cloudstep approach – where a user travels in a series of small jumps. Tunneling – where only a small central portion of a user field of view communicates movement – would also be to counterintuitive for the intended audience. Gamepads and keyboards were clearly off the list, and where redirected walking would be too complex, a one-to-one mirroring of virtual and physical movement would simply not respect capture volume restrictions.
Experimenting With Gain
With every conventional solution to comfortable VR movement unsuitable, Montoya and Sproll had to get creative, and prototyped several of their own solutions, mostly exploring the potential of 'gain', where a user's real movements are exaggerated in the virtual realm. With gain, if a user moves a yard in reality, they may pass through four feet of space in the virtual realm, for example.
Thus, the duo toyed with their own concepts, sometimes tweaking existing experimental appraoches. They conceived and extended approaches with names like 'walking gain', 'tangential gain', 'complex gain' and 'simple gain', each of which either placed too much demand on the user, or simply failed when a user would, for example, kneel down in reality, only to descend far further in VR.
Each of those approaches can be seen in the video of Montoya and Sproll's superb Vision Summit presentation, including the curious circular conveyer belt method of 'tengential gain' illustrated below, where a user can step on and off a rotating ringed pathway surrounding the Audi in question.
Ultimately, though, they arrived with a final solution they labeled 'rectangular gain', illustrated in the images below. In a rectangular gain set-up, users have a square 'safe zone' around them; an area with no gain, where their real-world movements translate in VR at a one-to-one scale.
This let Audi customers kneel, crane necks and strike any pose they wished without discomfort. But beyond the safe zone, communicated to the user with visual clues – and marked in red in the visual above – was an area where gain was applied, allowing them to move fast and loose around a virtual car without colliding with the limitations of a small capture zone. Rectangular gain can been seen in action in-editor below, from a top-down perspective (left) and from the user view-point (right).
Gain + Design
The pair also added a smart trick – described with clarity in their Vision Summit video – to allow a user to find and sit on a real world bench without needing to remove their HMD, which would teleport them into the seat of the car.
“Sometimes we have to teleport," says Montoya with a knowing smile.
All of that let first time VR users explore their dream Audi without discomfort. But what if they moved in a way as to collide with the virtual car, passing their head through its digital form in a way impossible in reality? To Audi, that would seem too much like a bug for its intended customer experience. That kind of clipping was impossible to prevent, but unacceptable to allow.
So Montoya and Sproll applied what they called ‘the X-ray effect’, where the car becomes translucent as the user moves into its 'physical' form. This gave the customer the sense they could really explore every element of their coming purchase both inside and out, with RE’FLEKT essentially using some good old-fashioned artistry to turn a bug into a feature.
Ultimately, their journey to creating rectangular gain VR locomotion saw Montoya and Sproll learn a great deal. And in the spirit on which the VR community is founded, they are only too happy to share their learnings.
“I can really recommend, whenever you develop something, try to think of those things – the edge cases where things break – and create a narrative to help the user to understand why things are happening, and what is happening," offers Sproll, explaining their solution for users who couldn't resist putting their head through a virtual steering wheel shaft.
“Navigation can be unrealistic, even if you’re trying to replicate a realistic experience," suggests Montoya, speaking more generally of how to best approach VR locomotion problems. "You can navigate in a realistic way, but it has to be coherent. So if I’m moving in one direction, and applying gain to my movement, the gain has to be in the same direction as my movement, or mostly in the same direction. That can help keep the experience feeling real.”
The pair also recommend VR developers consider the firm 'dos and don'ts' rules of VR content creation as a moving feast, and assert that often the simplest solution is the best one to pursue. If there must be trade offs; cut what is least important, Montoya suggests. And test often with as wide a pool of users as possible.
Sproll also concludes with a snippet of advice which, in a few short words, captures what makes developing for VR so wildly exciting.
“Try things that should not work," he implores, with glint in his eye.