1 paper
Yitong Yang, Yinglin Wang, Tian Zhang +2
While text-driven diffusion models demonstrate remarkable performance in image editing, the critical components of their text embeddings remain underexplored. The ambiguity and ent…