Research map / 3D vision / 3D generation
TabletopGen: Tabletop Scene Generation and Interactive Simulation for Robotic Manipulation
Ziqian Wang, Yonghao He, Licheng Yang, Wei Zou, Hongxuan Ma, Liu.Liu Liu.Liu, Wei Sui, Yuxin Guo, Hu Su
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 Differentiable Rotation Optimizer (vs MLLM-only rotation) (ablation) | TabletopGen · Image-to-scene | Col_S (%): 16.67 -> 7.69 | Improved |
| add DRO (ablation) | TabletopGen · Image-to-scene | DINOv2: 0.8261 -> 0.8383 | Improved |
| add Top-View Spatial Alignment (vs MLLM-only translation and scale) (ablation) | TabletopGen · Image-to-scene | Col_S (%): 61.54 -> 7.69 | Improved |
| add TSA (ablation) | TabletopGen · Image-to-scene | LPIPS: 0.4799 -> 0.4483 | Improved |
| remove both DRO and TSA vs removing TSA only (ablation) | TabletopGen · Image-to-scene | Col_S (%): 61.54 -> 62.82 | No clear effect |
| TabletopGen vs best image-driven baseline MIDI | MIDI · Image-to-scene | Col_S (%): 98.72 -> 7.69 | Improved |
| TabletopGen vs MIDI | MIDI · Image-to-scene | GPT-4o average: 4.33 -> 6.19 | Improved |
| TabletopGen vs MesaTask on text-to-scene | MesaTask · Text-to-scene | GPT-4o average: 4.12 -> 6.27 | Improved |
| TabletopGen vs Holodeck-table collisions | Holodeck-table · Text-to-scene | Col_S (%): 0.00 -> 6.67 | Hurt |
| TabletopGen vs MIDI under mean-view protocol | MIDI · Image-to-scene | DINOv2 gain: +0.1313 -> +0.1473 | Improved |
| TabletopGen vs MIDI in user study | MIDI · User study | overall preference (%): 11.61 -> 83.13 | Improved |
| TabletopGen vs SceneGen/FlashSculptor on stability | FlashSculptor · Image-to-scene | DSS (%): 6.42 -> 92.13 | Improved |
Later papers that compared with this method
2 later accepted papers report TabletopGen as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- One Video, One World: Turning Monocular Video into Physical 4D Scenes (ECCV 2026 · better on 5 of 5 shared results)
- PRISM: Programming Interactive Scenes from Monocular Images for Embodied Simulation (NeurIPS 2026 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- MIDI — MIDI: Multi-Instance Diffusion for Single Image to 3D Scene Generation (CVPR 2025) · better on 9 of 9 shared results
- MesaTask — MesaTask: Towards Task-Driven Tabletop Scene Generation via 3D Spatial Reasoning (NeurIPS 2025) · better on 1 of 1 shared result
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
- MesaTask: Towards Task-Driven Tabletop Scene Generation via 3D Spatial Reasoning (NeurIPS 2025 · similarity 0.89)
- GENMANIP: LLM-driven Simulation for Generalizable Instruction-Following Manipulation (CVPR 2025 · similarity 0.88)
- Simulation-Ready Compositional 3D Scene Reconstruction from a Single Image (NeurIPS 2026 · similarity 0.87)
- Atlas3D: Physically Constrained Self-Supporting Text-to-3D for Simulation and Fabrication (NeurIPS 2024 · similarity 0.86)
- DynScene: Scalable Generation of Dynamic Robotic Manipulation Scenes for Embodied AI (CVPR 2025 · similarity 0.85)
- Interactive3D: Create What You Want by Interactive 3D Generation (CVPR 2024 · similarity 0.85)
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.
