Research map / 3D vision / 3D generation
Scaling4D: Pushing the Frontier of Video Novel View Synthesis through Large-Scale Monocular Videos
Hongrui Cai, Junjie Luo, Zhihong Fu, Shengnan Zhu, Jiawei Wen, Wanquan Feng, Songtao Zhao, Qian He
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 real monocular video data (SpatialVID) to synthetic-only training (ablation) | Scaling4D · Panda-70M single-view (400 videos) | FVD: 472.62 -> 411.17 | Improved |
| add synthetic data to real-only training (ablation) | Scaling4D · Panda-70M single-view (400 videos) | TransErr: 308.92 -> 286.77 | Improved |
| add synthetic data to real-only training (ablation) | Scaling4D · Panda-70M single-view (400 videos) | FVD: 409.16 -> 411.17 | No clear effect |
| add double reprojection (inpainting-style) data (ablation) | Scaling4D · Panda-70M single-view (400 videos) | RotErr: 6.48 -> 6.27 | Improved |
| add double reprojection (inpainting-style) data (ablation) | Scaling4D · Panda-70M single-view (400 videos) | FVD: 411.17 -> 439.77 | Hurt |
| replace point-map inpainting with correspondence-guided generation | Scaling4D vs TrajectoryCrafter · Panda-70M single-view (400 videos) | FID: 68.94 -> 62.83 | Improved |
| replace point-map inpainting with correspondence-guided generation | Scaling4D vs TrajectoryCrafter · iPhone multi-view | LPIPS: 0.539 -> 0.468 | Improved |
| increase training data volume from 1k to 3000k videos | Scaling4D · Panda-70M single-view (400 videos) | FID/FVD/CLIP-V | Improved |
| increase training data from 100k to 3000k | Scaling4D · Panda-70M single-view (400 videos) | RotErr/TransErr | No clear effect |
| train with flow-based correspondence rather than depth-based | Scaling4D · 100 static videos | EPE: EPE 1.18 px; vec-corr 0.986 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GEN3C — GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control (CVPR 2025) · better on 5 of 5 shared results
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 0.90)
- One View Is Enough: In-the-Wild Monocular Pretraining for Novel View Generation (NeurIPS 2026 · similarity 0.89)
- SV4D: Dynamic 3D Content Generation with Multi-Frame and Multi-View Consistency (ICLR 2025 · similarity 0.89)
- CAT4D: Create Anything in 4D with Multi-View Video Diffusion Models (CVPR 2025 · similarity 0.89)
- Dynamic View Synthesis as an Inverse Problem (NeurIPS 2025 · similarity 0.88)
- MVGenMaster: Scaling Multi-View Generation from Any Image via 3D Priors Enhanced Diffusion Model (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.
