Abstract: We address the challenge of generating dynamic 4D scenes from monocular multi-object videos with heavy occlusions and introduce GenMOJO, a novel approach that integrates rendering-based ...