Research map / Image generation / Autoregressive and latent generation
HouseDiffusion: Vector Floorplan Generation via a Diffusion Model With Discrete and Continuous Denoising
Mohammad Amin Shabani, Sepidehsadat Hosseini, Yasutaka Furukawa
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 |
|---|---|---|---|
| Use joint continuous noise prediction and discrete coordinate denoising (ablation) | HouseDiffusion · RPLAN | Diversity (FID): 38.8\scalebox{0.8}{$\pm$0.9} -> 9.5\scalebox{0.8}{$\pm$0.1} | Improved |
| Use joint continuous noise prediction and discrete coordinate denoising (ablation) | HouseDiffusion · RPLAN | Compatibility: 2.2\scalebox{0.8}{$\pm$0.0} -> 2.5\scalebox{0.8}{$\pm$0.0} | Hurt |
| Replace continuous-plus-discrete denoising with discrete-only AnalogBits (ablation) | HouseDiffusion · RPLAN | Compatibility: 2.5\scalebox{0.8}{$\pm$0.0} -> 6.0\scalebox{0.8}{$\pm$0.03} | Hurt |
| Add Component-wise Self Attention (ablation) | HouseDiffusion · RPLAN | Compatibility: 2.9\scalebox{0.8}{$\pm$0.0} -> 2.5\scalebox{0.8}{$\pm$0.0} | Improved |
| Add Global Self Attention (ablation) | HouseDiffusion · RPLAN | Diversity (FID): 12.8\scalebox{0.8}{$\pm$0.2} -> 9.5\scalebox{0.8}{$\pm$0.1} | Improved |
| Add Relational Cross Attention (ablation) | HouseDiffusion · RPLAN | Compatibility: 6.8\scalebox{0.8}{$\pm$0.0} -> 2.5\scalebox{0.8}{$\pm$0.0} | Improved |
| Replace the three attention types with more GSA layers (ablation) | HouseDiffusion · RPLAN | Compatibility: 6.5\scalebox{0.8}{$\pm$0.0} -> 2.5\scalebox{0.8}{$\pm$0.0} | Hurt |
| Augment corners with sampled points along their walls (ablation) | HouseDiffusion · RPLAN | Compatibility: 3\scalebox{0.8}{$\pm$0.0} -> 2.5\scalebox{0.8}{$\pm$0.0} | Improved |
| Augment corners with sampled points along their walls (ablation) | HouseDiffusion · RPLAN | Diversity (FID): 10.0\scalebox{0.8}{$\pm$0.3} -> 9.5\scalebox{0.8}{$\pm$0.1} | Improved |
| Use discrete denoising for too few or too many final steps (ablation) | HouseDiffusion · RPLAN | Performance | Hurt |
| Generate vector floorplans with HouseDiffusion instead of House-GAN++ | RPLAN · RPLAN | Diversity (FID): 30.4\scalebox{0.8}{$\pm$4.4} -> 9.5\scalebox{0.8}{$\pm$0.1} | Improved |
| Generate vector floorplans with HouseDiffusion instead of House-GAN++ | RPLAN · RPLAN | Realism: -0.52 -> -0.19 | Improved |
Later papers that compared with this method
2 later accepted papers report HouseDiffusion as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Unified Vector Floorplan Generation via Markup Representation (CVPR 2026 · better on 2 of 3 shared results)
- MaskPLAN: Masked Generative Layout Planning from Partial Input (CVPR 2024 · better on 4 of 4 shared results)
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.
- House-GAN++ — House-GAN: Relational Generative Adversarial Networks for Graph-constrained House Layout Generation (ECCV 2020) · better on 11 of 11 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
- CG-Floor: Centroid-Guided Diffusion for Large-Scale Floorplan Generation (CVPR 2026 · similarity 0.90)
- GSDiff: Synthesizing Vector Floorplans via Geometry-enhanced Structural Graph Generation (AAAI 2025 · similarity 0.89)
- CAGE: Continuity-Aware edGE Network Unlocks Robust Floorplan Reconstruction (NeurIPS 2025 · similarity 0.87)
- DiffuScene: Denoising Diffusion Models for Generative Indoor Scene Synthesis (CVPR 2024 · similarity 0.87)
- DeBaRA: Denoising-Based 3D Room Arrangement Generation (NeurIPS 2024 · similarity 0.87)
- FreeScene: Mixed Graph Diffusion for 3D Scene Synthesis from Free Prompts (CVPR 2025 · similarity 0.86)
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.
