Junfan Lin

In-Context Model Predictive Generation: Open-Vocabulary Motion Synthesis from Language Models to Physics

Xiaomeng Fu, Junfan Lin, Yang Liu, Yaowei Wang, Guanbin Li, Liang Lin, et al.

Synthesizing human motion from textual descriptions is essential for immersive digital applications, yet existing methods face a persistent trade-off between semantic fidelity and physical realism. Large language model (LLM)-based approaches can interpret diverse open-vocabulary…Read more