The Reality Mixer client could support a "full body tracking" mode, and then it could use ARKit to perform full body tracking and send the position and orientation of the joints to the app running on visionOS (in addition to the camera pose updates).
The iOS client wouldn't receive a Mixed Reality video in this mode, and the visionOS library could surface the joints to the application so it could be used by the user to interact with the virtual environment.
The Reality Mixer client could support a "full body tracking" mode, and then it could use ARKit to perform full body tracking and send the position and orientation of the joints to the app running on visionOS (in addition to the camera pose updates).
The iOS client wouldn't receive a Mixed Reality video in this mode, and the visionOS library could surface the joints to the application so it could be used by the user to interact with the virtual environment.