Quantitative analysis of visibility determinations for networked virtual environments

► We formulate visibility models in terms of visible relevance among users and objects. ► Our modified precision/recall metrics reflect dynamics in moving environments. ► We present the fundamental limitation of user-centric AOI filtering models. ► Our new filtering method achieves optimality in terms of visible relevance. ► We present accurate analytical AOI filtering models. The Area-Of-Interest (AOI) model is a simple and popular technique used in many applications to determine the region which needs to be considered and processed for each entity (e.g., user). One example application is object visibility determination around user-representing avatars in virtual environments or networked games. There exist a number of variations of the AOI model and in our prior work we have demonstrated how object-oriented visibility determination is more suitable for networked virtual environments than conventional user-oriented visibility determination. Here we extend our work to study a unified and comprehensive analytical model that reveals fundamental properties about the different visibility determination techniques under a variety of virtual environment settings. We also present what the best operational scenarios are for each different approach. Although our discussion and analytical results are focused on the visibility domain, the arguments and conclusions can be extended to various applications or services where spatial attributes are required.