Abstract
Until recently, most content on the Internet has not been explicitly tied to specific people, places or things. However, content is increasingly being geo-coded and semantically labeled, making explicit connections between the physical world around us and the virtual world in cyberspace. Most augmented reality systems simulate a portion of the physical world, for the purposes of rendering a hybrid scene around the user. We have been experimenting with approaches to terra-scale, heterogeneous augmented reality mirror worlds, to unify these two worlds. Our focus has been on the authoring and user-experience, for example allowing ad-hoc transition between augmented and virtual reality interactions for multiple co-present users. This form of ubiquitous virtual reality raises several research questions involving the functional requirements, user affordances and relevant system architectures for these mirror worlds. In this paper, we describe our experiments with two mirror world systems and some lessons learned about the limitations of deploying these systems using massively multiplayer and dedicated game engine technologies.