This paper introduces a low-cost framework capable of combining both real-time markerless face and body tracking for faithful avatar embodiment in Virtual Reality (VR). We discuss suitable hardware and software solutions and present a first prototype. This work lays the technological basis for further research on the importance of the appearance and behavioral realism of avatars, e.g., for the illusion of virtual body ownership, for social interactions in VR, as well as for VR entertainment applications (immersive games or movies).