Advances in Digital Human Modeling and Simulation

The field of digital human modeling and simulation is rapidly advancing, with a focus on creating highly realistic and dynamic models of humans and their interactions with the environment. Recent developments have seen significant improvements in the ability to capture complex facial nuances, model loose clothing, and simulate realistic motion and dynamics. The use of neural radiance fields (NeRFs) and 3D Gaussian Splatting (3DGS) has been particularly noteworthy, enabling the creation of highly detailed and realistic digital humans. Notable papers in this area include: Towards Physically Plausible Video Generation via VLM Planning, which proposes a novel framework for generating physically plausible videos. RealityAvatar, which presents an efficient framework for high-fidelity digital human modeling, specifically targeting loosely dressed avatars.

Sources

Refined Geometry-guided Head Avatar Reconstruction from Monocular RGB Video

Towards Physically Plausible Video Generation via VLM Planning

Learning 3D-Gaussian Simulators from RGB Videos

Diffusion Model-Based Size Variable Virtual Try-On Technology and Evaluation Method

Monocular and Generalizable Gaussian Talking Head Animation

GarmageNet: A Dataset and Scalable Representation for Generic Garment Modeling

RealityAvatar: Towards Realistic Loose Clothing Modeling in Animatable 3D Gaussian Avatars

Built with on top of