Research map / 3D vision / 3D generation
CamGeo: Sparse Camera-Conditioned Image-to-Video Generation with 3D Geometry Priors
Xuanyi Liu, Deyi Ji, Liqun Liu, Lanyun Zhu, Xuhang Chen, Qianxiong Xu, Peng Shu, Huan Yu, Jie Jiang, Feng Gao, Siwei Ma
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 cross-frame smoothness (ablation) | CamGeo SVD · RealEstate10K 1/2 | TransError: 5.05 -> 4.89 | Improved |
| add cross-frame smoothness (ablation) | CamGeo SVD · RealEstate10K 1/2 | FVD StyleGAN: 103.5 -> 95.9 | Improved |
| add warm-up of geometric losses (ablation) | CamGeo SVD · RealEstate10K 1/3 | TransError: 5.12 -> 4.76 | Improved |
| add warm-up of geometric losses (ablation) | CamGeo SVD · RealEstate10K 1/3 | FVD StyleGAN: 99.2 -> 95.3 | Improved |
| add long-range smoothness beyond adjacent frames (ablation) | CamGeo SVD · RealEstate10K 1/4 | TransError: 5.61 -> 4.57 | Improved |
| use proposed curriculum instead of linear schedule (ablation) | CamGeo CogVideoX · RealEstate10K 1/2 | TransError: 4.86 -> 4.89 | No clear effect |
| use proposed curriculum instead of linear schedule (ablation) | CamGeo CogVideoX · RealEstate10K 1/2 | FVD StyleGAN: 86.4 -> 83.4 | Improved |
| use both camera and depth modalities instead of camera only (ablation) | CamGeo CogVideoX · RealEstate10K 1/3 | TransError: 5.52 -> 4.88 | Improved |
| use both camera and depth modalities instead of depth only (ablation) | CamGeo CogVideoX · RealEstate10K 1/3 | TransError: 5.19 -> 4.88 | Improved |
| add dynamic weighting (ablation) | CamGeo CogVideoX · RealEstate10K 1/4 | TransError: 5.03 -> 4.86 | Improved |
| add CamGeo geometry distillation to SVD-Base | SVD · MannequinChallenge OOD 1/2 | TransError: 13.21 -> 8.02 | Improved |
| add CamGeo geometry distillation to CogVideoX-Base | CogVideoX · RealEstate10K 1/2 | FVD StyleGAN: 89.5 -> 83.4 | Improved |
Closest papers by meaning
- AC3D: Analyzing and Improving 3D Camera Control in Video Diffusion Transformers (CVPR 2025 · similarity 0.87)
- MVSplat360: Feed-Forward 360 Scene Synthesis from Sparse Views (NeurIPS 2024 · similarity 0.86)
- VideoMDM: Towards 3D Human Motion Generation From 2D Supervision (NeurIPS 2026 · similarity 0.86)
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.86)
- Training-free Camera Control for Video Generation (ICLR 2025 · similarity 0.86)
- Taming Video Diffusion Prior with Scene-Grounding Guidance for 3D Gaussian Splatting from Sparse Inputs (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.
