Abstract:
Realistic digital humans can be generated through 3D scanning and have been shown to be well suited for representing us as embodied avatars in virtual reality applications. These avatars are typically represented as surface meshes and are therefore inherently limited when it comes to simulating volumetric behavior, such as realistic character animation or body weight changes. In this talk I will show how to estimate a person's anatomical details from just a surface scan and how the resulting bones, muscles, and soft tissue lead to more accurate animation and simulation results.