Research map / Video / Video segmentation and tracking
UniDynamics: Event-RGB Fusion for Unified Future 4D Dynamic Scene Generation
Daikun Liu, Xin Zhan, Teng Wang, Xiaoping Wang, Changyin Sun
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 event input (ablation) | UniDynamics · E-VKitti2 | FID: 46.8 -> 39.1 | Improved |
| add event input (ablation) | UniDynamics · E-VKitti2 | FVD: 577.1 -> 223.6 | Improved |
| add event input under motion blur (ablation) | UniDynamics · E-VKitti2 motion blur | FVD: 617.7 -> 272.8 | Improved |
| use ELE module instead of directly concatenating event features (ablation) | UniDynamics · E-VKitti2 | FVD: 417.6 -> 223.6 | Improved |
| add optical-flow branch (ablation) | UniDynamics · E-VKitti2 | FID/FVD: FID 42.0 -> 39.1; FVD 264.2 -> 223.6 | Improved |
| add optical-flow branch (ablation) | UniDynamics · E-VKitti2 | depth AbsRel: 0.27 -> 0.27 | No clear effect |
| add depth branch (ablation) | UniDynamics · E-VKitti2 | flow EPE: 3.64 -> 3.00 | Improved |
| add PDS task adapters and gated bi-interaction (ablation) | UniDynamics · E-VKitti2 | FID: 50.8 -> 39.1 | Improved |
| add PDS task adapters and gated bi-interaction (ablation) | UniDynamics · E-VKitti2 | flow EPE: 4.60 -> 3.00 | Improved |
| model depth and flow jointly with RGB (vs Vista) (ablation) | UniDynamics · E-VKitti2 | FID: 101.8 -> 46.8 | Improved |
| UniDynamics vs UniFuture-Depth on zero-shot DSEC FID (ablation) | UniDynamics · DSEC (zero-shot) | FID: 46.2 vs 50.9 | Mixed |
Closest papers by meaning
- EA3D: Event-Augmented 3D Diffusion for Generalizable Novel View Synthesis (ICLR 2026 · similarity 0.88)
- Bring Event into RGB and LiDAR: Hierarchical Visual-Motion Fusion for Scene Flow (CVPR 2024 · similarity 0.88)
- Diffusion4D: Fast Spatial-temporal Consistent 4D generation via Video Diffusion Models (NeurIPS 2024 · similarity 0.86)
- Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting (ICLR 2024 · similarity 0.86)
- Motion4D: Learning 3D-Consistent Motion and Semantics for 4D Scene Understanding (NeurIPS 2025 · similarity 0.86)
- 4Diffusion: Multi-view Video Diffusion Model for 4D Generation (NeurIPS 2024 · 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.
