Is there a way to do Mocap without marker balls?

…in 2020, is there a way? I remember that during the 3090 card release, Nvidia showed off a “doing a green screen without a green screen” thing, where a guy in a room in a video had the background (the room) autoremoved by AI!! Surely, after the guy has been isolated, to look at the video and figure out which way his limbs and facial muscles are moving - that algorithm/s already EXISTS??
Note that the camera is like, a webcam, NOT a Kinect/ToF sensor/LIDAR scanner and the like.

Can it be done?

1 Like

Thanks, splodo :slight_smile: