The field of human and garment modeling is rapidly advancing, with a focus on developing innovative methods for sim-to-real transfer, dynamic garment deformation, and virtual try-on. Researchers are exploring the use of diffusion models, physics-based simulations, and keypoints to improve the accuracy and generalizability of these models. A key direction in this field is the development of compact and shape-agnostic representations of cloth states, which can be applied to various applications such as semantic labeling and planning. Notable papers in this area include:

• DiSRT-In-Bed, which proposes a sim-to-real transfer framework for in-bed human mesh recovery, and
• D-Garment, which introduces a physics-conditioned latent diffusion model for dynamic garment deformations. These advancements have the potential to significantly improve the realism and accuracy of virtual try-on, garment modeling, and human mesh recovery, with applications in healthcare, entertainment, and gaming.