1 paper · 1 filter
Mingshu Cai, Miao Zhang, Chenghe Yang +3
In recent years, training-free video generation has progressed remarkably. However, when handling complex textual instructions, existing methods still suffer from semantic ambiguit…