Both have headsets (used to be a cost issue, gradually getting solved) and play the same thing.

Or one has AR glasses and can peek into what the other is doing and communicate with them.

An internal camera captures the face under the headset and presents it on the outside with correct parallax (could be presented with AR or with auto stereo display), AI in the headset can tell when someone is wanting attention etc. and use other cameras with depth reconstruction or depth sensors to fade them into the VR player's world.