Research map / 3D vision / 3D generation
VideoWeaver: Multimodal Multi-View Video-to-Video Transfer for Embodied Agents
George Eskandar, Fengyi Shen, Mohammad Altillawi, Dong Chen, Yang Bai, Liudi Yang, Ziyuan Liu
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 |
|---|---|---|---|
| inject 4D point cloud (Pi3) into latent space (ablation) | VideoWeaver multi-view · Droid test | Met3R: 0.481 -> 0.533 | Improved |
| inject 4D point cloud (Pi3) into latent space (ablation) | VideoWeaver multi-view · Agibot test | Met3R: 0.597 -> 0.612 | Improved |
| add heterogeneous-timestep training (conditional on one view) (ablation) | VideoWeaver multi-view · Droid test | Met3R: 0.533 -> 0.578 | Improved |
| add heterogeneous-timestep training (conditional on one view) (ablation) | VideoWeaver multi-view · Agibot test | Met3R: 0.612 -> 0.624 | Improved |
| fuse depth and sketch with MoE instead of sketch only (ablation) | VideoWeaver single-view · Bridge test | Depth siRMSE: 0.245 -> 0.158 | Improved |
| fuse depth and sketch with MoE instead of sketch only (ablation) | VideoWeaver single-view · Bridge test | VBench: 0.833 -> 0.837 | No clear effect |
| fuse depth and sketch with MoE instead of depth only (ablation) | VideoWeaver single-view · Bridge test | Edge-F1: 0.250 -> 0.393 | Improved |
| drop sketch at inference with MoE model (ablation) | VideoWeaver single-view · Bridge test | Edge-F1: 0.393 -> 0.171 | Hurt |
| use multi-view model instead of single-view (ablation) | VideoWeaver · Droid test | Depth siRMSE: 0.362 -> 0.347 | Improved |
| use multi-view model instead of single-view (ablation) | VideoWeaver · Droid test | JEDi: 0.384 -> 0.509 | Hurt |
| use VideoWeaver single-view instead of VACE | VACE · Droid test | VBench: 0.729 -> 0.732 | No clear effect |
Closest papers by meaning
- Seeing Realism from Simulation: Efficient Video Transfer for Vision-Language-Action Data Augmentation (ICML 2026 · similarity 0.90)
- DiT4DiT: Jointly Modeling Video Dynamics and Actions for Generalizable Robot Control (NeurIPS 2026 · similarity 0.89)
- Action Images: End-to-End Policy Learning via Multiview Video Generation (NeurIPS 2026 · similarity 0.88)
- RoboAlign-R1: Distilled Multimodal Reward Alignment for Robot Video World Models (NeurIPS 2026 · similarity 0.88)
- Seeing Across Views: Benchmarking Spatial Reasoning of Vision-Language Models in Robotic Scenes (ICLR 2026 · similarity 0.88)
- Learning an Actionable Discrete Diffusion Policy via Large-Scale Actionless Video Pre-Training (NeurIPS 2024 · similarity 0.88)
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.
