A masked fine-tuned diffusion model simultaneously completes occluded views and synthesizes novel viewpoints, enabling fast feed-forward 3D reconstruction.
Objaverse: A universe of annotated 3d objects
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
EscherNet++: Simultaneous Amodal Completion and Scalable View Synthesis through Masked Fine-Tuning and Enhanced Feed-Forward 3D Reconstruction
A masked fine-tuned diffusion model simultaneously completes occluded views and synthesizes novel viewpoints, enabling fast feed-forward 3D reconstruction.