The MPEG Avatar Representation Format (ARF) an Interoperable Container and Animation Framework for Avatars

Avatars are becoming key for identity and presence in immersive communication, from social virtual reality to remote collaboration and mixed-reality experiences. Although most avatar technologies remain closed in proprietary ecosystems, limiting reuse of a user's digital persona across applications and complicating deployment of avatar-based services over 5 G networks. The MPEG Avatar Representation Format (ARF), standardized as ISO/IEC 23090-39 (MPEG-I Part 39), addresses these limitations by introducing an extensible, and application-independent framework for representing, packaging, exchanging, and animating 3D avatars. This article presents a system-level view of ARF and its use for avatar communication over 5 G. We outline the motivation and relation to H-Anim and VRM, describe the base avatar model, ISOBMFF and ZIP-based container formats, and the animation stream, show how ARF integrates with MPEG-I Scene Description and 3GPP's IP Multimedia Subsystem, and highlight key features, including the reference client, authentication, and tensor formats.