In recent years, networked virtual environments have steadily grown to become a frontier in social computing. Such virtual cyberspaces are usually accessed by multiple users through their 3D avatars. Recent scientific activity has resulted in the release of both hardware and software components that enable users at home to interact with their virtual persona through natural body and facial activity performance. Based on 3D computer graphics methods and vision-based motion tracking algorithms, these techniques aspire to reinforce the sense of autonomy and telepresence within the virtual world. In this paper we present two distinct frameworks for avatar animation through user natural motion input. We specifically target the full body avatar control case using a Kinect sensor via a simple, networked skeletal joint retargeting pipeline, as well as an intuitive user facial animation 3D reconstruction pipeline for rendering highly realistic user facial puppets. Furthermore, we present a common networked architecture to enable multiple remote clients to capture and render any number of 3D animated characters within a shared virtual environment.