Qualcomm launches the XR2 5G, a very promising processor for stand-alone AR/VR headsets
At the 2019 Snapdragon Tech Summit, Qualcomm announced the new processor: Snapdragon XR2. It is Qualcomm’s second chip (the Snapdragon XR1 was released in May 2018).
This new chip dedicated to virtual and augmented reality headsets and glasses promises significant capabilities compared to the previous generation of processors.
To date, the fastest system on an XR headset chip is the Microsoft Hololens 2 Snapdragon 850. However, Qualcomm supplies dozens of the latest AR/VR devices with the majority of current XR headsets using a Snapdragon 835 or XR1. The XR2 is a derivative of Snapdragon 865, with modifications and additions relevant to the XR.
Qualcomm reports an improvement with a 2:1 ratio over Snapdragon 835 performance, as well as 4x higher video bandwidth, 6x higher native resolution support, and 11x higher AI performance. Qualcomm combines the XR2 5G with its latest Snapdragon X55 modem. It could, therefore, provide cellular connectivity to some VR and AR devices.
The Snapdragon XR2 5G is not designed to replace the XR1, but to complement it as a higher-level product. Video capabilities are improved, including decoding 8K 60 FPS content as well as 4K 120 FPS content, which means supporting display refresh rates up to 120 Hz. The XR2 5G also supports up to 3K x 3K per eye. Besides, it will support HDR color spaces, including HDR10 and HDR10+, for dynamic range and improved color.
The Snapdragon XR2 5G is also designed to support 7 simultaneous cameras. It explains why many manufacturers can adopt XR2 in their new generation of helmets.
The ability to have 7 simultaneous cameras can improve user interaction and immersion at levels never before seen in mobile XR.
Helmet manufacturers will no longer have to choose between functions such as position tracking, eye tracking, hand tracking, face tracking, body tracking, camera tracking, or controller tracking. Manufacturers could theoretically do all these things simultaneously with this chipset.
Features such as object detection, occlusion, and recognition will depend heavily on the capabilities of the chip’s AI inside the helmet.
Other features such as controller and manual tracking already use artificial intelligence to determine where the controllers are or where they could go without interrupting the user’s immersion.
Even features such as object occlusion and 3D reconstruction probably depend on the performance of the AI, which can give XR2 a competitive advantage over other platforms.
Current XR headsets generally have no connectivity other than Wi-Fi and Bluetooth, which directs them to home or office use cases. The ability to connect an XR headset to a 5G network should give operators more flexibility in the way they offer XR devices.