Research map / Video / Video generation
SV4D 2.0: Enhancing Spatio-Temporal Consistency in Multi-View Video Diffusion for High-Quality 4D Generation
Chun-Han Yao, Yiming Xie, Vikram Voleti, Huaizu Jiang, Varun Jampani
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 3D attention over merged view and spatial dimensions (ablation) | SV4D 2.0 · ObjaverseDy NVVS | FV4D: 283.10 -> 256.84 | Improved |
| add 3D attention (ablation) | SV4D 2.0 · ObjaverseDy NVVS | LPIPS: 0.115 -> 0.105 | Improved |
| add 3D attention (ablation) | SV4D 2.0 · ObjaverseDy NVVS | FVD-F: 548.14 -> 473.90 | Improved |
| add improved ObjaverseDy++ data curation (ablation) | SV4D 2.0 · ObjaverseDy NVVS | FV4D: 316.91 -> 256.84 | Improved |
| add improved ObjaverseDy++ data curation (ablation) | SV4D 2.0 · ObjaverseDy NVVS | LPIPS: 0.112 -> 0.105 | Improved |
| add improved ObjaverseDy++ data curation (ablation) | SV4D 2.0 · ObjaverseDy NVVS | FVD-F: 576.66 -> 473.90 | Improved |
| upgrade SV4D to SV4D 2.0 for novel-view video synthesis | SV4D · ObjaverseDy NVVS | FV4D: 458.34 -> 256.84 | Improved |
| upgrade SV4D to SV4D 2.0 for novel-view video synthesis | SV4D · ObjaverseDy NVVS | LPIPS: 0.122 -> 0.105 | Improved |
| upgrade SV4D to SV4D 2.0 for 4D optimization | SV4D · ObjaverseDy 4D | FV4D: 699.04 -> 532.55 | Improved |
| upgrade SV4D to SV4D 2.0 for NVVS | SV4D · Consistent4D NVVS | FVD-F: 677.68 -> 553.43 | Improved |
| upgrade SV4D to SV4D 2.0 for 4D optimization | SV4D · Consistent4D 4D | LPIPS: 0.118 -> 0.117 | No clear effect |
| use SV4D 2.0 instead of L4GM for 4D optimization | L4GM · Consistent4D 4D | FVD-F: 691.87 -> 692.68 | Hurt |
Later papers that compared with this method
1 later accepted paper reports SV4D 2.0 as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Ani3DHuman: Photorealistic 3D Human Animation with Self-guided Stochastic Sampling (CVPR 2026 · better on 8 of 8 shared results)
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.
- SV4D — SV4D: Dynamic 3D Content Generation with Multi-Frame and Multi-View Consistency (ICLR 2025) · better on 7 of 7 shared results
- Consistent4D — Consistent4D: Consistent 360° Dynamic Object Generation from Monocular Video (ICLR 2024) · better on 2 of 2 shared results
- STAG4D — STAG4D: Spatial-Temporal Anchored Generative 4D Gaussians (ECCV 2024) · better on 2 of 2 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
- SV4D: Dynamic 3D Content Generation with Multi-Frame and Multi-View Consistency (ICLR 2025 · similarity 0.94)
- Diffusion4D: Fast Spatial-temporal Consistent 4D generation via Video Diffusion Models (NeurIPS 2024 · similarity 0.92)
- 4Diffusion: Multi-view Video Diffusion Model for 4D Generation (NeurIPS 2024 · similarity 0.91)
- Diffusion$^2$: Dynamic 3D Content Generation via Score Composition of Video and Multi-view Diffusion Models (ICLR 2025 · similarity 0.91)
- CAT4D: Create Anything in 4D with Multi-View Video Diffusion Models (CVPR 2025 · similarity 0.90)
- Consistent4D: Consistent 360° Dynamic Object Generation from Monocular Video (ICLR 2024 · similarity 0.90)
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.
