Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2509.23258 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914074944602112 |
|---|---|
| author | Topaloglu, Atakan Li, Kunyi Niemeyer, Michael Navab, Nassir Tekalp, A. Murat Tombari, Federico |
| author_facet | Topaloglu, Atakan Li, Kunyi Niemeyer, Michael Navab, Nassir Tekalp, A. Murat Tombari, Federico |
| contents | Sparse-view novel view synthesis is fundamentally ill-posed due to severe geometric ambiguity. Current methods are caught in a trade-off: regressive models are geometrically faithful but incomplete, whereas generative models can complete scenes but often introduce structural inconsistencies. We propose OracleGS, a novel framework that reconciles generative completeness with regressive fidelity for sparse view Gaussian Splatting. Instead of using generative models to patch incomplete reconstructions, our "propose-and-validate" framework first leverages a pre-trained 3D-aware diffusion model to synthesize novel views to propose a complete scene. We then repurpose a multi-view stereo (MVS) model as a 3D-aware oracle to validate the 3D uncertainties of generated views, using its attention maps to reveal regions where the generated views are well-supported by multi-view evidence versus where they fall into regions of high uncertainty due to occlusion, lack of texture, or direct inconsistency. This uncertainty signal directly guides the optimization of a 3D Gaussian Splatting model via an uncertainty-weighted loss. Our approach conditions the powerful generative prior on multi-view geometric evidence, filtering hallucinatory artifacts while preserving plausible completions in under-constrained regions, outperforming state-of-the-art methods on datasets including Mip-NeRF 360 and NeRF Synthetic. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_23258 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | OracleGS: Grounding Generative Priors for Sparse-View Gaussian Splatting Topaloglu, Atakan Li, Kunyi Niemeyer, Michael Navab, Nassir Tekalp, A. Murat Tombari, Federico Computer Vision and Pattern Recognition Sparse-view novel view synthesis is fundamentally ill-posed due to severe geometric ambiguity. Current methods are caught in a trade-off: regressive models are geometrically faithful but incomplete, whereas generative models can complete scenes but often introduce structural inconsistencies. We propose OracleGS, a novel framework that reconciles generative completeness with regressive fidelity for sparse view Gaussian Splatting. Instead of using generative models to patch incomplete reconstructions, our "propose-and-validate" framework first leverages a pre-trained 3D-aware diffusion model to synthesize novel views to propose a complete scene. We then repurpose a multi-view stereo (MVS) model as a 3D-aware oracle to validate the 3D uncertainties of generated views, using its attention maps to reveal regions where the generated views are well-supported by multi-view evidence versus where they fall into regions of high uncertainty due to occlusion, lack of texture, or direct inconsistency. This uncertainty signal directly guides the optimization of a 3D Gaussian Splatting model via an uncertainty-weighted loss. Our approach conditions the powerful generative prior on multi-view geometric evidence, filtering hallucinatory artifacts while preserving plausible completions in under-constrained regions, outperforming state-of-the-art methods on datasets including Mip-NeRF 360 and NeRF Synthetic. |
| title | OracleGS: Grounding Generative Priors for Sparse-View Gaussian Splatting |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2509.23258 |