arxivcs.CV2026-06-26
MindFlow: Harmonizing Cognitive Semantics and Acoustic Dynamics for Facial Animation Generation in Dyadic Conversations
Hejia Chen, Haoxian Zhang, Xu He, Xiaoqiang Liu, Pengfei Wan, Shoulong Zhang, et al.
Generating lifelike facial animation for dyadic conversations requires reconciling high-level cognitive intent with precise low-level motor reflexes, yet existing methods fall short in the semantic understanding of dialogue context and in precise dynamic control. In this paper, w…