The medium brings new opportunities sure, but also its own challenges. For example, capturing convincing 3D representations of live people is really difficult. Even just motion capture in specially designed studios isn’t easy.
Besides 2D overlays, I suspect that most AR applications will use avatars instead of trying to digitize real people. An unrealistic or cartoonish avatar would also avoid any uncanny valleyness and quite likely would even be part of the appeal
Microsoft has 2 Mixed Reality studios now that do 3D capture, one in SF and the other in Seattle, which they rent out for VR game creation and other purposes.
NVidia also has a 3D video capture tech called Virtual Eye. Rumor has it modern compression brings the video within 20% of 2D video size.
While it's not yet feasible for an AR headset to do such dense capture, the tech is not far off.
