Researchers at Purdue University have outlined a new software solution to the limitations of underpowered smartphone-driven VR.

You could argue that one of the great technological divides of today is the gap between smartphone-powered VR and computer-powered VR. There’s a noticeable difference between the two, but its logical that smartphone driven VR experiences are likely to be the mainstream future of the tech.

It’s only a small leap to say we all have a smartphone. Whereas it’s no leap at all to say most of us do not know a single person to have shelled out hundreds of dollars on a dedicated VR system.

Charlie Hu, professor of electrical and computer engineering at Purdue University, Indiana, is acutely aware of this. Co-author of a paper outlining a software-based approach to addressing the smartphone power bottleneck, Hu explains:

“We have performed a systematic design study of the ‘elephant in the room’ facing the VR industry: Is it feasible to enable high-quality VR apps on untethered mobile devices such as smartphones?”

Succinctly, the answer is no. The report’s findings indicate that current mobile tech and wireless infrastructure are 10 times too slow for high quality VR on your smartphone.

Instead of waiting for hardware to catch up, the research paper proposes a three-step software solution as the answer to the problem, now. Through the Furion software platform, they claim mobile VR titles achieve a “quality of experience” akin to the high-end VR offered by PC-powered setups.

How Does Furion Work?

According to the research, for a VR experience to be good enough (passing a high enough “quality of experience” benchmark), each frame displayed to you must render within 16 milliseconds — approximately 60 frames-per-second.

The chips inside a modern smartphone, while increasingly powerful, cannot keep up with such a graphically intensive task. Not least one that offers an experience one would consider high-quality.

Whatever is powering the VR experience, it is constantly rendering everything in the scene, including background objects and scenery that likely will not move. This inefficiency is one thing the paper identifies as a place to offload the rendering.

Hu reasons “the user’s position doesn’t change randomly, you move continuously and in a very predictable way. So that means we can predict how the background will change based on the user’s position and pre-render the background.”

Furion, on a separate PC, picks up this background rendering, and streams it in sync to the game back to the smartphone device. In essence it creates a panoramic photo that is streamed to the device in four parts. “Pre-fetched” and loaded on across the smartphone’s many processor cores, this stream of pre-rendered imagery is dynamically kept pre-fetched in anticipation of the user’s movements.

Thanks to the method, the researchers claim an improvement from 111 milliseconds pre-Furion, to just 14 milliseconds render time with the help of Furion. This improvement was noted using the same Google Pixel XL smartphone.

While a higher frame rate is always a welcome thing, we’re intrigued to see if the technology could free up computational resources to up the graphical quality of smartphone-driven VR. Presented during the October ACM MobiCom, the researchers have already filed a provisional patent.

Source: Purdue University