CVPR2025
EditSplat: Multi-View Fusion and Attention-Guided Optimization for View-Consistent 3D Scene Editing with 3D Gaussian Splatting
Dong In Lee, Hyeongcheol Park, Jiyoung Seo, Eunbyung Park, Hyunje Park, Hadam Baek, Sangheon Shin, Sangmin Kim, Sangpil Kim
摘要
Source Views "face" "Make his face resemble that of a marble sculpture" "Turn him into a harry potter" "Make the bonsai snowy" "Change the bonsai flowers into autumn leaves" Source Views "bear statue" "Turn the bear statue into a polar bear" "Turn the bear statue into a wild boar" Source Views "lego bonsai" Figure 1 . Result of EditSplat. EditSplat enables flexible and high-quality editing of pre-trained 3D Gaussian Splatting models guided solely by textual instructions. Through its design focused on multi-view consistency, efficient optimization, and precise semantic local editing, Our approach demonstrates robust performance, producing realistic and fine-grained 3D scene modifications.