1 paper
Qinghe Wang, Baolu Li, Xiaomin Li +4
Recent advances in text-to-image models have opened new frontiers in human-centric generation. However, these models cannot be directly employed to generate images with consistent…