Research map / Video / Video generation
GeometryCrafter: Consistent Geometry Estimation for Open-world Videos with Diffusion Priors
Tian-Xing Xu, Xiangjun Gao, Wenbo Hu, Xiaoyu Li, Song-Hai Zhang, Ying Shan
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 point map VAE (replace SVD VAE on normalized disparity) (ablation) | GeometryCrafter · Sintel depth | Reld: 25.2 -> 16.9 | Improved |
| add point map VAE (ablation) | GeometryCrafter · GMU Kitchen depth | Reld: 9.50 -> 8.03 | Improved |
| add point map VAE (ablation) | GeometryCrafter · KITTI depth | Reld: 5.00 -> 5.25 | No clear effect |
| use decoupled point map representation (Eq. 5) instead of cuboid (ablation) | point map VAE · Sintel VAE reconstruction | Relp: 8.25 -> 4.24 | Improved |
| add latent alignment (ablation) | point map VAE · Sintel VAE reconstruction | Relp: 4.48 -> 3.45 | Improved |
| add latent alignment (ablation) | UNet · GMU Kitchen | Relp: 12.9 -> 8.52 | Improved |
| add per-frame geometry prior from MoGe (ablation) | UNet · GMU Kitchen | Relp: 9.51 -> 8.52 | Improved |
| add per-frame geometry prior from MoGe (ablation) | UNet · DDAD | Relp: 14.3 -> 15.0 | No clear effect |
| use deterministic variant instead of diffusion (ablation) | GeometryCrafter · inference speed | FPS: 3.7 -> 4.1 | Mixed |
| replace MoGe prior with DUSt3R prior (ablation) | UNet · GMU Kitchen | Relp: 8.52 -> 12.2 | Mixed |
| use GeometryCrafter instead of DAV | video depth · Sintel | Reld: 35.7 -> 16.9 | Improved |
| use GeometryCrafter instead of UniDepth | point map · Sintel | Relp: 30.9 -> 25.6 | 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.
- DepthCrafter — DepthCrafter: Generating Consistent Long Depth Sequences for Open-world Videos (CVPR 2025) · better on 3 of 3 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
- MotionCrafter: Dense Geometry and Motion Reconstruction with a 4D VAE (CVPR 2026 · similarity 0.90)
- GeoVideo: Introducing Geometric Regularization into Video Generation Model (NeurIPS 2025 · similarity 0.88)
- DepthCrafter: Generating Consistent Long Depth Sequences for Open-world Videos (CVPR 2025 · similarity 0.87)
- Towards Physical Understanding in Video Generation: A 3D Point Regularization Approach (NeurIPS 2025 · similarity 0.86)
- LaGeM: A Large Geometry Model for 3D Representation Learning and Diffusion (ICLR 2025 · similarity 0.86)
- Dora: Sampling and Benchmarking for 3D Shape Variational Auto-Encoders (CVPR 2025 · similarity 0.86)
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.
