Collaborative Virtual Environments – From Birth to Standardization Chris Joslin, Thomas Di Giacomo, Nadia Magnenat-Thalmann MIRALab - University of Geneva {Joslin, Giacomo, Thalmann}@miralab.unige.ch Abstract Collaborative Virtual Environments, which have been referred to by many different names over the years, have been around since the early 90’s, and some even before as hardware dedicated versions; now they are being standardized by the Motion Picture Experts Group (MPEG). This paper examines most of the significant systems over the past 13 years, discussing some of the most pioneering systems and the main methodologies implied, and whether their standardization is feasible. Introduction Fluid, unrestrictive, and realistic collaboration, between two or more people, in a virtual environment has been the dream of many possibly since computers (and emphasized by the improvement of graphics) were conceived. To be able to interact with people virtually, whilst unable to determine that it really is just bits and bytes, is the an ultimate goal. However to provide such a functionality, the kind of technologies required takes a lot of craft. It is not just a case of virtual reality that is simpler to set up compared to creating a collaborative virtual environment, it is virtual reality plus a whole lot more features: multiple users, interaction, networking and communication, servers and clients, virtual and real spaces, accuracy and multi-modal communication on levels of significant complexity etc. CVE Systems, a History These types of collaborative virtual workspaces come under many different guises: collaborative virtual environments (CVE), networked (NVE or NET-VE), collaborative workspaces (CW) and many more, but they all take on the same generally tasks; albeit for different purposes. Leaving aside some of the simpler systems involving just text based collaboration (such as Multi-User Dungeons or MUDs) we discuss CVE research and development from the early systems through to the latest developments, the trends that have emerged and the concepts that disappeared. We have examined the mass of papers specifically related to CVE systems from the last 13 years, starting around 1990 with a system called “Reality Built for Two” [1]; there is one system in 1987 by Sim et al [2] which can be classified as a CVE, but was built using dedicated hardware for military training purposes. Since then there have been a multitude of systems developed, popularity peaked in 1997 with new developments generally falling off ever since; this is exemplified in Figure 1. It is atypical in many latter papers to claim credit for innovations passed over many times; Area of Interest Management, which has been called many things in the past – but uses the same concept, has been reinvented almost as many times as there have been years. Scene segmentation suffers from similar problems, but here authors claim different methodologies. In this paper we attempt to examine them all, and provide an overview of CVE History as it has taken place over the last 13 years. We conclude the paper with a discussion on MPEG; after so much innovation it is interesting to observe how and in fact what is being standardised and also how it might affect the average user of an MPEG system.