Search papers, labs, and topics across Lattice.
Leveling3D enhances feed-forward 3D Gaussian Splatting (3DGS) reconstruction by integrating it with a geometry-aware diffusion model for simultaneous reconstruction and generation. A key component is the geometry-aware leveling adapter, which aligns the diffusion model's internal knowledge with the geometric prior from the feed-forward model to address artifacts in extrapolated novel views. The method also incorporates a palette filtering strategy for training and a test-time masking refinement to improve generation diversity and boundary quality, ultimately leading to state-of-the-art performance in novel-view synthesis and depth estimation.
By combining feed-forward 3D reconstruction with a geometry-aware diffusion model, Leveling3D fills in the gaps in extrapolated novel views, leveling up both 3D reconstruction and generation.
Feed-forward 3D reconstruction has revolutionized 3D vision, providing a powerful baseline for downstream tasks such as novel-view synthesis with 3D Gaussian Splatting. Previous works explore fixing the corrupted rendering results with a diffusion model. However, they lack geometric concern and fail at filling the missing area on the extrapolated view. In this work, we introduce Leveling3D, a novel pipeline that integrates feed-forward 3D reconstruction with geometrical-consistent generation to enable holistic simultaneous reconstruction and generation. We propose a geometry-aware leveling adapter, a lightweight technique that aligns internal knowledge in the diffusion model with the geometry prior from the feed-forward model. The leveling adapter enables generation on the artifact area of the extrapolated novel views caused by underconstrained regions of the 3D representation. Specifically, to learn a more diverse distributed generation, we introduce the palette filtering strategy for training, and a test-time masking refinement to prevent messy boundaries along the fixing regions. More importantly, the enhanced extrapolated novel views from Leveling3D could be used as the inputs for feed-forward 3DGS, leveling up the 3D reconstruction. We achieve SOTA performance on public datasets, including tasks such as novel-view synthesis and depth estimation.
