Research map / 3D vision / 3D generation
Diffusion-Based Generation, Optimization, and Planning in 3D Scenes
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use diffusion-based scene-conditioned generation instead of a CVAE | Human pose generation · PROX human pose generation | plausible rate: 12.57 -> 24.83 | Helped |
| Add optimization-guided sampling (ablation) | SceneDiffuser · PROX human pose generation | plausible rate: 24.83 -> 49.35 | Helped |
| Add optimization-guided sampling (ablation) | SceneDiffuser · PROX human motion generation without start pose | plausible rate: 24.70 -> 23.53 | Hurt |
| Add optimization-guided sampling (ablation) | SceneDiffuser · PROX human motion generation with start pose | plausible rate: 41.76 -> 42.30 | Helped |
| Add physics-guided optimization (ablation) | SceneDiffuser · MultiDex dexterous grasp generation | depth collision (mm): 17.34 -> 14.61 | Helped |
| Add physics-guided optimization (ablation) | SceneDiffuser · MultiDex dexterous grasp generation | success rate (all): 71.25 -> 69.84 | Hurt |
| Use SceneDiffuser rather than deterministic L2 planning | 3D scene navigation planner · ScanNet unseen-scene navigation | success rate (%): 13.50 -> 73.75 | Helped |
| Use SceneDiffuser rather than deterministic L2 planning | Robot arm planner · Robot arm motion planning | success rate (%): 72.87 -> 78.59 | Helped |
| Replace PointNet with PointNet++ as the object scene encoder (ablation) | SceneDiffuser grasp generator · MultiDex dexterous grasp generation | diversity (rad.): 0.0718 -> 0.1568 | Helped |
| Replace PointNet with PointNet++ as the object scene encoder (ablation) | SceneDiffuser grasp generator · MultiDex dexterous grasp generation | success rate (all): 71.25 -> 66.25 | Hurt |
| Replace cross-attention with self-attention (ablation) | SceneDiffuser grasp generator · MultiDex dexterous grasp generation | success rate (all): 71.25 -> 75.94 | Helped |
| Replace cross-attention with self-attention (ablation) | SceneDiffuser grasp generator · MultiDex dexterous grasp generation | diversity (rad.): 0.0718 -> 0.0535 | Hurt |
Closest papers by meaning
- DiffuScene: Denoising Diffusion Models for Generative Indoor Scene Synthesis (CVPR 2024 · similarity 0.87)
- Diffusion Models for Robotics (AAAI 2025 · similarity 0.87)
- DORSal: Diffusion for Object-centric Representations of Scenes $\textit{et al.}$ (ICLR 2024 · similarity 0.87)
- LT3SD: Latent Trees for 3D Scene Diffusion (CVPR 2025 · similarity 0.87)
- SceneWeaver: All-in-One 3D Scene Synthesis with an Extensible and Self-Reflective Agent (NeurIPS 2025 · similarity 0.87)
- SemCity: Semantic Scene Generation with Triplane Diffusion (CVPR 2024 · similarity 0.87)
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.