Full Body Avatar

Full-body avatar research aims to create realistic, animatable digital representations of humans for immersive virtual and augmented reality applications. Current efforts focus on generating these avatars from limited input data, such as single egocentric cameras or sparse sensor readings, employing techniques like Gaussian splatting, neural radiance fields, and diffusion models to reconstruct detailed geometry and appearance, often incorporating parametric body models like SMPL-X. This field is significant for advancing human-computer interaction, virtual telepresence, and animation technologies, with implications for fields ranging from healthcare training to entertainment.