Research map / Video / Video segmentation and tracking
Stereo World Model: Camera-Guided Stereo Video Generation
Yang-Tian Sun, Zehuan Huang, Yifan Niu, Lin Ma, Yan-Pei Cao, Yuewen Ma, Xiaojuan Qi
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 |
|---|---|---|---|
| generate stereo views end-to-end instead of converting monocular outputs | StereoWorld · Stereo video benchmark (435 stereo images) | FVD: 96.87 -> 83.04 | Improved |
| use StereoWorld instead of monocular-then-convert baseline (Aether) | Aether + StereoCrafter · Stereo video benchmark (435 stereo images) | FPS: 0.11 -> 0.49 | Improved |
| use StereoWorld instead of SEVA | SEVA + StereoCrafter · Stereo video benchmark (435 stereo images) | Matching pixels (K): 4.49 -> 4.56 | Improved |
| end-to-end stereo generation vs post-hoc conversion | SEVA · Stereo video benchmark (435 stereo images) | FVD-V: 31.10 -> 22.00 | Improved |
| use copy init instead of zero init for camera subspace (ablation) | unified camera-frame RoPE · Camera injection ablation (TartanAir) | RotErr: 1.81 -> 1.16 | Improved |
| use copy init instead of zero init for camera subspace (ablation) | unified camera-frame RoPE · Camera injection ablation (TartanAir) | FVD: 96.62 -> 93.17 | Improved |
| use unified camera-frame RoPE instead of Plucker ray concatenation (ablation) | StereoWorld · Camera injection ablation (TartanAir) | RotErr: 1.52 -> 1.16 | Improved |
| use unified camera-frame RoPE instead of PRoPE (ablation) | StereoWorld · Camera injection ablation (TartanAir) | FVD: 128.32 -> 93.17 | Improved |
| use stereo attention instead of full 4D attention (ablation) | StereoWorld · Attention scheme ablation | FLOPs (x10^10): 3.11 -> 1.56 | Improved |
| use stereo attention instead of full 4D attention (ablation) | StereoWorld · Attention scheme ablation | FPS: 0.34 -> 0.49 | Improved |
| use stereo attention instead of full 4D attention (ablation) | StereoWorld · Attention scheme ablation | CLIP-V: 97.55 -> 97.05 | Hurt |
| use stereo attention instead of full 4D attention (ablation) | StereoWorld · Attention scheme ablation | View synchronization (matching pixels): 4.51 -> 4.52 | No clear effect |
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.
- Aether — Aether: Geometric-Aware Unified World Modeling (ICCV 2025) · better on 4 of 4 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
- StereoWorld: Geometry-Aware Monocular-to-Stereo Video Generation (CVPR 2026 · similarity 0.93)
- ImmersePro: End-to-End Stereo Video Synthesis Via Implicit Disparity Learning (ICML 2026 · similarity 0.90)
- Mono2Stereo: A Benchmark and Empirical Study for Stereo Conversion (CVPR 2025 · similarity 0.88)
- SpatialDreamer: Self-supervised Stereo Video Synthesis from Monocular Input (CVPR 2025 · similarity 0.87)
- Stereo Anywhere: Robust Zero-Shot Deep Stereo Matching Even Where Either Stereo or Mono Fail (CVPR 2025 · similarity 0.87)
- PPMStereo: Pick-and-Play Memory Construction for Consistent Dynamic Stereo Matching (NeurIPS 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.
