1 paper
Zilyu Ye, Jinxiu Liu, Ruotian Peng +9
Recent image generation models excel at creating high-quality images from brief captions. However, they fail to maintain consistency of multiple instances across images when encoun…