1 paper · 1 filter
Mingrui Wu, Lu Wang, Pu Zhao +14
Despite recent progress in text-to-image (T2I) generation, existing models often struggle to faithfully capture user intentions from short and under-specified prompts. While prior…