1 paper
Yushe Cao, Xuechao Zou, Xing Xi +3
Although diffusion-based methods have substantially improved the controllability of multimodal face synthesis, their semantic alignment remains suboptimal because most existing app…