Research map / 3D vision / 3D generation
SpatialCrafter: Unleashing the Imagination of Video Diffusion Models for Scene Reconstruction from Limited Observations
Songchun Zhang, Huiyao Xu, Sitong Guo, Zhongwei Xie, Hujun Bao, Weiwei Xu, Changqing Zou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ray embedding camera encoding (ablation) | SpatialCrafter video diffusion · Video diffusion ablation | FVD: 112.843 -> 78.132 | Improved |
| use ray embedding instead of raw value embedding (ablation) | SpatialCrafter video diffusion · Video diffusion ablation | T_err: 1.592 -> 1.014 | Improved |
| add epipolar attention (ablation) | SpatialCrafter video diffusion · Video diffusion ablation | COLMAP_err: 12.2% -> 4.2% | Improved |
| add epipolar attention (ablation) | SpatialCrafter video diffusion · Video diffusion ablation | FVD: 79.547 -> 78.132 | Improved |
| add unified scale alignment across datasets (ablation) | SpatialCrafter video diffusion · Video diffusion ablation | T_err: 1.712 -> 1.014 | Improved |
| use SpatialCrafter instead of ViewCrafter for generation | camera-controlled video generation · Tanks-and-Temples | PSNR: 16.24 -> 16.12 | Mixed |
| use SpatialCrafter instead of ReconX for two-view reconstruction | RealEstate10K two-view · RealEstate10K | LPIPS: 0.088 -> 0.121 | Hurt |
| use SpatialCrafter instead of MVSplat on low-overlap inputs | RealEstate10K two-view small overlap · RealEstate10K (small overlap) | PSNR: 20.353 -> 22.514 | Improved |
| use SpatialCrafter instead of CAT3D on DTU 3-view | sparse-view reconstruction · DTU (3-view) | PSNR: 22.02 -> 27.92 | Improved |
| add video-latent generative priors and depth features to reconstruction (ablation) | feed-forward Gaussian reconstruction · qualitative | geometric detail fidelity | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- ViewCrafter — ViewCrafter: Taming Video Diffusion Models for High-fidelity Novel View Synthesis (TPAMI 2025) · better on 5 of 7 shared results
- CAT3D — CAT3D: Create Anything in 3D with Multi-View Diffusion Models (NeurIPS 2024) · better on 2 of 2 shared results
- MVSplat — MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images (ECCV 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ViVid-1-to-3: Novel View Synthesis with Video Diffusion Models (CVPR 2024 · similarity 0.90)
- UMAMI: Unifying Masked Autoregressive Models and Deterministic Rendering for View Synthesis (NeurIPS 2025 · similarity 0.90)
- Sparse3D: Distilling Multiview-Consistent Diffusion for Object Reconstruction from Sparse Views (AAAI 2024 · similarity 0.89)
- Novel View Synthesis with Pixel-Space Diffusion Models (CVPR 2025 · similarity 0.89)
- Novel View Synthesis from A Few Glimpses via Test-Time Natural Video Completion (NeurIPS 2025 · similarity 0.89)
- ViDAR: Video Diffusion-Aware 4D Reconstruction From Monocular Inputs (NeurIPS 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
