Research map / 3D vision / 3D generation
Taming LLMs for Codematic Indoor Scene Generation
yixun liang, Qianyi Wu, Chuan Fang, Rui Chen, Jiahang Liu, Jianfeng Zhang, Ping Tan
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 CMI training recipe (ablation) | Qwen3-4B-Instruct finetune · Component ablation | Instruction Following: 6.371 -> 6.988 | Improved |
| add CMI training recipe (ablation) | Qwen3-4B-Instruct finetune · Component ablation | Layout Fidelity & Completeness: 6.557 -> 6.806 | Improved |
| add planning phase on top of CMI recipe (ablation) | SceneSpinner · Component ablation | Instruction Following: 6.988 -> 7.631 | Improved |
| add planning phase on top of CMI recipe (ablation) | SceneSpinner · Component ablation | Layout Fidelity & Completeness: 6.806 -> 7.157 | Improved |
| add both planning phase and CMI recipe (ablation) | Qwen3-4B-Instruct finetune · Component ablation | Instruction Following: 6.371 -> 7.631 | Improved |
| use SceneSpinner instead of HoloDeck | HoloDeck · Codematic scene generation | Instruction Following: 5.722 -> 7.852 | Improved |
| use SceneSpinner instead of HoloDeck | HoloDeck · Codematic scene generation | Layout Fidelity & Completeness: 6.037 -> 7.191 | Improved |
| use SceneSpinner instead of HoloDeck | HoloDeck · Codematic scene generation | CLIP Score: 0.259 -> 0.260 | No clear effect |
| add planning stage (ablation) | SceneSpinner · attention analysis (bunk bed prompt) | attention on instruction | Improved |
| train without CMI regulation (ablation) | SceneSpinner · training analysis | KL divergence / instruction following | Hurt |
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 3 of 3 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
- Direct Numerical Layout Generation for 3D Indoor Scene Synthesis via Spatial Reasoning (NeurIPS 2025 · similarity 0.93)
- Hierarchically-Structured Open-Vocabulary Indoor Scene Synthesis with Pre-trained Large Language Model (AAAI 2025 · similarity 0.93)
- InstructScene: Instruction-Driven 3D Indoor Scene Synthesis with Semantic Graph Prior (ICLR 2024 · similarity 0.92)
- OptiScene: LLM-driven Indoor Scene Layout Generation via Scaled Human-aligned Data Synthesis and Multi-Stage Preference Optimization (NeurIPS 2025 · similarity 0.92)
- SceneWeaver: All-in-One 3D Scene Synthesis with an Extensible and Self-Reflective Agent (NeurIPS 2025 · similarity 0.90)
- SceneCraft: Layout-Guided 3D Scene Generation (NeurIPS 2024 · similarity 0.89)
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.
