Research map / 3D vision / Gaussian splatting
MonoFusion: Sparse-View 4D Reconstruction via Monocular Fusion
Zihan Wang, Jeff Tan, Tarasha Khurana, Neehar Peri, Deva Ramanan
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 space-time consistent depth initialization (DUSt3R-aligned MoGe) (ablation) | MonoFusion baseline · held-out views ablation | PSNR: 26.19 -> 29.55 | Improved |
| add feature-metric loss L_feat (ablation) | MonoFusion baseline · held-out views ablation | PSNR / IoU: PSNR 26.19 -> 25.39; IoU 0.60 -> 0.63 | Mixed |
| add feature-metric loss on top of our depth (ablation) | MonoFusion · held-out views ablation | PSNR / IoU: PSNR 29.55 -> 29.31; IoU 0.73 -> 0.75 | Mixed |
| add feature-clustering motion bases (ablation) | MonoFusion · held-out views ablation | PSNR: 29.31 -> 30.40 | Improved |
| use feature-based instead of velocity-based motion bases (ablation) | MonoFusion · 4-camera sparse view | qualitative | Improved |
| use fewer than 20 motion bases (ablation) | MonoFusion · sparse-view reconstruction | qualitative | Hurt |
| increase motion bases beyond 28 (ablation) | MonoFusion · sparse-view reconstruction | qualitative | No clear effect |
| use globally-consistent DUSt3R depth only (MV-SOM-DS) instead of DUSt3R+MoGe alignment (ablation) | MV-SOM · ExoRecon | PSNR: 30.43 -> 28.37 | Hurt |
| improve metric depth estimator in MV-SOM (UniDepthV2) (ablation) | MV-SOM · ExoRecon | PSNR: 26.91 -> 27.65 (still below MonoFusion 30.64) | Improved |
| MonoFusion vs monocular Shape of Motion on held-out views | MonoFusion · Panoptic Studio | PSNR: 17.86 -> 28.01 | Improved |
Later papers that compared with this method
2 later accepted papers report MonoFusion* as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- SparseCam4D: Spatio-Temporally Consistent 4D Reconstruction from Sparse Cameras (CVPR 2026 · better on 9 of 9 shared results)
- UniFusion: Sparse-View 4D Reconstruction via Unified Spatio-temporal Depth Alignment (ECCV 2026 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- ViewCrafter — ViewCrafter: Taming Video Diffusion Models for High-fidelity Novel View Synthesis (TPAMI 2025) · better on 1 of 1 shared result
- DNGaussian — DNGaussian: Optimizing Sparse-View 3D Gaussian Radiance Fields with Global-Local Depth Normalization (CVPR 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
- Reconstruct, Inpaint, Test-Time Finetune: Dynamic Novel-view Synthesis from Monocular Videos (NeurIPS 2025 · similarity 0.89)
- World from Motion: Generative Dynamic Gaussian Reconstruction from Monocular Video (NeurIPS 2026 · similarity 0.89)
- CAT4D: Create Anything in 4D with Multi-View Video Diffusion Models (CVPR 2025 · similarity 0.88)
- MoDGS: Dynamic Gaussian Splatting from Casually-captured Monocular Videos with Depth Priors (ICLR 2025 · similarity 0.88)
- 4D3R: Motion-Aware Neural Reconstruction and Rendering of Dynamic Scenes from Monocular Videos (NeurIPS 2025 · similarity 0.88)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.88)
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.
