I met with Dr Keisuke Suzuki and Dr David Schwarzman, Postdoctoral Research Fellows at the Sackler Centre for Consciousness Science at Sussex University. They were involved in a VR experiment where a 360 film of the environment is presented to the viewer interspersed with a live 360 camera feed, as a way of testing what changes people notice, how accurately and the size of the changes they can detect.
Demoing the Quizzer to them we discussed how it is easier to match a video with a live scene viewed through a camera than to match it with the real world, as the lens looks at the world differently from the eye. (How differently, is one of the things I have been trying to find out.)
David suggested I think about using a cross to help users line up the video and the real world, something the optomotrists I have spoken to told me also. In fact, I usually tell the viewer what aspect of the scene to line up, and it always contains horizontal and vertical lines- in the Fusebox corridor, it is the square window in the door. Perhaps this aspect could first appear accented without the background scene using markerless AR to help the user find the right viewing position. Or maybe it's better to simply highlight the physical world - I could do this is a subtle way using paste-up graffiti.