Research map / Graph learning / Graph self-supervised learning
Nestwork: Conditional 3D Furnished House Layout Generation through Latent Heterogeneous Graph Diffusion
Shuhan Miao, Biru Cao, Junling Zhuang
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 |
|---|---|---|---|
| unified one-pass generation vs two-stage HouseDiffusion+DiffuScene | Nestwork · HG-FRONT RoomGraph | FID: 41.90 -> 7.26 | Improved |
| unified one-pass vs two-stage | Nestwork · HG-FRONT RoomGraph | collision rate: 9.84 -> 10.91 | Hurt |
| heterogeneous HetGAT backbone vs TripletGCN (ablation) | autoencoder · HG-FRONT full graph | FID: 7.50 -> 7.13 | Improved |
| HetGAT backbone vs SLN-Box (ablation) | autoencoder · HG-FRONT full graph | collision rate: 14.80 -> 11.21 | Improved |
| diffusion latent prior vs IID prior (ablation) | Nestwork · HG-FRONT full graph | FID: 24.64 -> 7.13 | Improved |
| diffusion latent prior vs autoregressive prior (ablation) | Nestwork · HG-FRONT full graph | FID: 61.43 -> 7.13 | Improved |
| diffusion prior vs AR prior (ablation) | Nestwork · HG-FRONT full graph | collision rate: 10.89 -> 11.21 | Hurt |
| add node-level cross-attention (ablation) | Nestwork denoiser · HG-FRONT full graph | FID: 76.35 -> 7.13 | Improved |
| add low-rank relational field (ablation) | Nestwork decoder · HG-FRONT full graph | FID: 20.42 -> 7.13 | Improved |
| add low-rank relational field (ablation) | Nestwork decoder · HG-FRONT full graph | walkability: 81.29 -> 88.58 | Improved |
| less conditioning (Topology vs FullGraph) | Nestwork single model · HG-FRONT | FID: 7.13 -> 7.75 | No clear effect |
Closest papers by meaning
- OptiScene: LLM-driven Indoor Scene Layout Generation via Scaled Human-aligned Data Synthesis and Multi-Stage Preference Optimization (NeurIPS 2025 · similarity 0.88)
- InstructScene: Instruction-Driven 3D Indoor Scene Synthesis with Semantic Graph Prior (ICLR 2024 · similarity 0.88)
- Direct Numerical Layout Generation for 3D Indoor Scene Synthesis via Spatial Reasoning (NeurIPS 2025 · similarity 0.88)
- FreeScene: Mixed Graph Diffusion for 3D Scene Synthesis from Free Prompts (CVPR 2025 · similarity 0.88)
- DeBaRA: Denoising-Based 3D Room Arrangement Generation (NeurIPS 2024 · similarity 0.87)
- DiffuScene: Denoising Diffusion Models for Generative Indoor Scene Synthesis (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.
