The Zoom guesswork-based imitation is pretty good, and it seems to be optimized for a single person's movement. It gets confused when there are more actors like a child or dog entering from stage left.
And where can I find the OBS implementation of this work?