Research map / Video / Video generation
AutoScape: Geometry-Consistent Long-Horizon Scene Generation
Jiacheng Chen, Ziyu Jiang, Mingfu Liang, Bingbing Zhuang, Jong-Chyi Su, Sparsh Garg, Ying Wu, Manmohan Chandraker
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 RGB-D pre-training (ablation) | AutoScape · nuScenes val (Overall 0-20s) | FID: 47.6 -> 35.1 | Improved |
| add RGB-D pre-training (ablation) | AutoScape · nuScenes val (Overall 0-20s) | FVD: 650.0 -> 359.0 | Improved |
| add data filtering of inconsistent samples (ablation) | AutoScape · nuScenes val (Overall 0-20s) | FID: 43.5 -> 35.1 | Improved |
| add data filtering of inconsistent samples (ablation) | AutoScape · nuScenes val (Overall 0-20s) | FVD: 463.0 -> 359.0 | Improved |
| add warp consistent guidance (ablation) | AutoScape · nuScenes val (Overall 0-20s) | FID: 38.5 -> 35.1 | Improved |
| add warp consistent guidance (ablation) | AutoScape · nuScenes val (Overall 0-20s) | FVD: 380.2 -> 359.0 | Improved |
| add joint depth generation (RGB-D diffusion) (ablation) | AutoScape · nuScenes val (Overall 0-20s) | FID: 39.2 -> 35.1 | Improved |
| add joint depth generation (RGB-D diffusion) (ablation) | AutoScape · nuScenes val (Overall 0-20s) | FVD: 511.4 -> 359.0 | Improved |
| use AutoScape instead of Vista | AutoScape vs Vista · nuScenes val (Overall 0-20s) | FVD: 629.8 -> 359.0 | Improved |
| use AutoScape instead of Vista at 15-20s | AutoScape vs Vista · nuScenes val (15-20s) | FVD: 1614.5 -> 657.4 | Improved |
| use AutoScape instead of Vista on unseen dataset | AutoScape vs Vista · Argoverse2 (no fine-tuning) | FID: 80.4 -> 49.2 | Improved |
Closest papers by meaning
- A 3D Scene is Worth 1K Tokens: 3D-Grounded Representation for Scene Generation at Scale (NeurIPS 2026 · similarity 0.87)
- X-Scene: Large-Scale Driving Scene Generation with High Fidelity and Flexible Controllability (NeurIPS 2025 · similarity 0.87)
- Diffusion4D: Fast Spatial-temporal Consistent 4D generation via Video Diffusion Models (NeurIPS 2024 · similarity 0.87)
- LT3SD: Latent Trees for 3D Scene Diffusion (CVPR 2025 · similarity 0.87)
- Scene Splatter: Momentum 3D Scene Generation from Single Image with Video Diffusion Model (CVPR 2025 · similarity 0.87)
- LongScape: Advancing Long-Horizon Embodied World Models with Context-Aware MoE (NeurIPS 2026 · 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.
