XR (VR/AR/MR) applications with full-bodied avatars for the mass end-user are focused on low-cost motion capture equipment, which either provides relatively high speed control of the avatar (mainly a helmet with 2 hand controllers) - but with a loss of accuracy in movements transfer (the most problematic, as a rule, are displayed lower body movements), or, conversely, provides higher accuracy in displaying movements of all body parts (for the low-budget category these are mainly front-facing cameras) - but then there is not enough computing power to be able to accurately recognize a large video stream in real time, which is necessary for online avatar control. The technology PAO-XR allows XR applications developers to solve this dilemma based on preliminary training of the avatar by user's personalized movements on the frontal video stream, followed by personalized control of the avatar in online mode using more faster MOCAP equipment.
コメント