Qualcomm Technologies Inc. today unveiled a new chip with the Snapdragon XR2+ Gen 2 platform that will power high-resolution 3D graphics in next-generation immersive virtual reality and mixed reality headsets.

The new chip builds on the same technology as the Snapdragon XR2 Gen 2 platform, which was announced in September and is built into the Meta Platforms Inc. Quest 3 VR headset.

The new “+” version represents a performance boost over the XR2 Gen 2 with a 15% higher graphics processing unit frequency and 20% higher central processing unit performance. It was also built with support for powerful on-board artificial intelligence designed to power gesture and body tracking using 12 or more camera sensors.

“Snapdragon XR2+ Gen 2 unlocks 4.3K resolution which will take XR productivity and entertainment to the next level by bringing spectacularly clear visuals to use cases such as room-scale screens, life-size overlays and virtual desktops,” said Hugo Swart, vice president and general manager of XR at Qualcomm. “We are advancing our commitment to power the best XR devices and experiences that will supercharge our immersive future.”

One of the biggest features that the XR2+ Gen 2 builds on is foveated rendering, which allows for the high-quality rendering of 3D graphics only where the user’s gaze is by using eye tracking and low-quality rendering elsewhere. This works because human eyesight is high-quality in only a very narrow vision range in the center, or tunnel, of vision. By only rendering a very small area in that gaze, it can save a lot of GPU and CPU cycles.

“There is no way you can process a row or a native 4K by 4K resolution. So here foveation is very important,” Said Bakadir, senior director of product management at Qualcomm, explained during a press briefing. “We have supported foveation for years now due to our architecture.”

Another key upgrade included an increase in tracking capability that allowed for driving more sensors for face tracking, body tracking and gesture tracking. “What we have also is the ability to drive a lot more cameras,” said Bakadir. “We have the ability to support 12 or more cameras. Now that we have these visuals people are expecting more interactions.”

With this increased number of sensors and cameras that can be enabled at once, explained Bakadir, virtual avatars can be made more lifelike and interactive. For example, when meeting someone in a virtual space, they are represented by a digital representation of themselves, which is sometimes a cartoony version that might have a face, arms and legs. The ability to reflect the movements of eyes, eyebrows, limbs, heads and fingers makes the social experience a lot more lifelike because suddenly body language is added to voice.

For mixed reality capabilities, which differs from virtual reality in that users can see the real world blended with virtual objects, Bakadir said that the chip supports full-color video passthrough with extremely low latency of 12 milliseconds.

Qualcomm also revealed that it’s working in close partnership with Samsung Electronics Co. Ltd. and Google LLC to bring the new chip to market. The company also said it’s working with five customers, but did not reveal their names.

To help accelerate the ability for developers to begin working with the chip, Qualcomm also unveiled a new MR and VR reference design developed by China-based Goertek Inc., with eye-tracking technology from Tobii AB.

Photo: Pixabay