1 paper
Bin Hu, Zipeng Qi, Guoxi Huang +6
Single-view reference-to-video methods often struggle to preserve identity consistency under large facial-angle variations. This limitation naturally motivates the incorporation of…