Facebook recently demonstrated the ability to render 3D faces with full real expression but just using a set of mobile VR glasses is enough. To achieve this achievement, programmers have designed a neural network to learn the changes in muscle groups and wrinkles of the skin on the face, thereby optimizing the algorithm to minimize the number of devices. Need to reproduce any face. You can follow the introduction clip to better understand the performance of Facebook.
Specifically, by using two VR glasses, one is conventional, equipped with 3 cameras and a reference glass with 9 cameras, Facebook's neural network was able to optimize the algorithm to recreate the complete Adjust the face and expression with only 3 cameras, while the quality is still very realistic for a glass using 9 cameras.
In fact it took years for the technology to reach the commercial stage, but Facebook has put a lot of ambition into the future of VR society (as evidenced by the acquisition of Oculus VR). Allow users to see and interact with each other at higher levels. Therefore, soon we will experience social networking in a whole new way. What do you think?