1 paper · 1 filter
Suchae Jeong, Inseong Choi, Youngsik Yun +1
Text-to-Image models, including Stable Diffusion, have significantly improved in generating images that are highly semantically aligned with the given prompts. However, existing mo…