EscherNet++: Simultaneous Amodal Completion and Scalable View Synthesis through Masked Fine-Tuning and Enhanced Feed-Forward 3D Reconstruction
EscherNet++ is a unified diffusion model that simultaneously performs amodal completion and novel view synthesis. It uses hierarchical masking—input-level and feature-level—to handle occluded input views robustly. This allows the model to generate complete objects from various query poses while seamlessly integrating with fast 3D reconstruction models.