Research map / 3D vision / 3D generation
R$^3$L: Reasoning 3D Layouts from Relative Spatial Relations
Zhifeng Gu, Yuqi Wang, Bing WANG
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 invariant spatial decomposition and consistent spatial imagination (ablation) | R3L relation inference · Ablation (physical losses disabled) | instruction following: 7.8 -> 9.1 | Improved |
| add invariant spatial decomposition (versus imagination only) (ablation) | R3L · Ablation (physical losses disabled) | instruction following: 8.5 -> 9.1 | Improved |
| add consistent spatial imagination (versus decomposition only) (ablation) | R3L · Ablation (physical losses disabled) | out-of-bound ratio %OR: 6.3 -> 1.6 | Improved |
| add consistent spatial imagination (versus decomposition only) (ablation) | R3L · Ablation (physical losses disabled) | collision ratio %CR: 3.1 -> 1.0 | Improved |
| semantic grouping instead of invariant spatial decomposition (ablation) | R3L · Ablation (physical losses disabled) | realism: 8.0 -> 5.3 | Hurt |
| use R3L instead of LayoutGPT | 3D layout generation · 9 scene types average | collision and out-of-bound ratios: 4.3% and 8.1% -> 0.0% and 0.0% | Improved |
| add global-to-local pose re-parameterization to optimization (ablation) | supportive spatial optimization · Optimization convergence | convergence speed: 1.8x speedup | Improved |
| add decomposition to reduce reference-frame transformations (ablation) | R3L · Placement error by hop count | placement error rate | Improved |
| use R3L instead of Holodeck | 3D layout generation · 9 scene types average | instruction following: 3.0 -> 9.1 | Improved |
| use R3L instead of LayoutVLM | 3D layout generation · 9 scene types average | realism: 4.9 -> 7.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Holodeck — Holodeck: Language Guided Generation of 3D Embodied AI Environments (CVPR 2024) · better on 4 of 4 shared results
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
- LayoutVLM: Differentiable Optimization of 3D Layout via Vision-Language Models (CVPR 2025 · similarity 0.92)
- Thinking in Structures: Evaluating Spatial Intelligence in Constraint-Governed Spaces (ICML 2026 · similarity 0.90)
- Direct Numerical Layout Generation for 3D Indoor Scene Synthesis via Spatial Reasoning (NeurIPS 2025 · similarity 0.90)
- S$^2$-MLLM: Boosting Spatial Reasoning Capability of MLLMs for 3D Visual Grounding with Structural Guidance (CVPR 2026 · similarity 0.89)
- 3ViewSense: Spatial and Mental Perspective Reasoning from Orthographic Views in Vision-Language Models (ICML 2026 · similarity 0.88)
- SpatialRGPT: Grounded Spatial Reasoning in Vision-Language Models (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.
