Every year, during the Oculus Connect, Oculus chief scientist Michael Abrash holds a session to discuss his thoughts on virtual reality (VR), its future and how the industry can get there – this is in-depth, but worth seeing. All R & D companies were under Oculus Research until last year, when it was renamed Facebook Research Labs. From time to time it was possible to sneak a peek at what the labs are working on. But now this information will be even more transparent thanks to a new series of blogs.
Abrash made an announcement through the Oculus blog today, reporting that it would be an annual series of messages going deep into the various Facebook reality labs, highlighting another team and what they are working on in the future.
“I expect blog posts to be markers on the way to the future of AR / VR.” In the coming months, you will see deep dives about optics and displays, computer vision, audio, graphics, tactile interaction, the brain / computer interface and eye / hand / face / body tracking, ”Abrash notes.
Revealing all sorts of interesting events and breakthroughs that teams make, the first blog post focuses on the realistic avatars and connections that people create inside the digital world. The project is called Codec Avatars, which is managed by Yasser Sheikh, director of Facebook Reality Labs in Pittsburgh.
In an effort to overcome the problems of distance between people, the project uses a combination of 3D capture technology and AI systems to create realistic user avatars.
FRL Pittsburgh’s Facebook Reality LabsThe team has been working on this issue for several years, Sheikh joined Facebook in 2015. Their work was demonstrated during the F8 2018 with two realistic digital people animated in real time. Since then, the team has completed the creation of two capture objects: one for the face and one for the body. Each of them is designed to reconstruct the structure and measure body movement at an unprecedented level of detail. This achievement allowed the team to take data and build an automated pipeline to create photo-realistic avatars.
Pittsburgh’s FRL capture system is capable of capturing 180 gigabytes of data per second, thanks to hundreds of cameras — each camera captures data at a speed of 1 GB per second. The proprietary algorithm then uses the data to create a unique avatar.
The type of technology that Pittsburgh FRL uses will not be quickly available to the everyday consumer, but the company certainly demonstrates the steps that need to be taken to eventually get to this stage of technological evolution.
As soon as VRcue learns more about technology, we’ll let you know.