Theres no reason why green screen couldnt be applied today. All the device has to do... is have two cameras mounted in front of the eye area.
Lenses can compensate for any positional issues, if thats even needed. (it could also be done in-game)
The stereo images would be overlayed into the game directly... and or converted into something cgi, and imported as such.
Ive already seen software that, with only cameras... can generate full 3d models, with full color, texture, and accurate details.
And as far as Ive heard... motion tracking on the new generation of devices is much more advanced.
The more reading / positional devices... the more accuracy.