Chenxu Zhang

Title: Chenxu Zhang’s Patents in AI-Generated Media

Introduction

Chenxu Zhang is an inventor associated with Los Angeles, California. His patent work includes techniques for generating music-driven dance videos and creating animatable three-dimensional avatars. His portfolio includes 2 patents.

Latest Patents

Zhang’s patent on music-driven dance video generation describes a system that combines a reference image with a music sequence to create animated video frames. The system generates music embeddings and pose tokens, uses a transformer model to predict successive poses, and applies motion guidance and a diffusion model to produce frames conditioned on the reference image.

His work on creating three-dimensional avatars presents a coarse-to-fine method for generating high-quality, animatable human representations from text. The approach creates a representation in a canonical space, deforms it to a target pose using Skinned Multi-Person Linear parameters, and iteratively optimizes it across canonical and deformed spaces using Score Distillation Sampling loss.

Career Highlights

Zhang has been associated with Lemon Inc. and Beijing Zitiao Network Technology Co., Ltd. His inventions address generative media, connecting artificial intelligence techniques with image, motion, and 3D avatar creation.

Collaborations

His listed coworkers include Jianfeng Zhang and Huichao Zhang, reflecting a collaborative context for his work in generative media technologies.

Conclusion

Chenxu Zhang’s patents focus on AI-based methods for turning music and text into animated visual content, spanning dance video generation and three-dimensional avatar creation.