Research map / Point clouds / Shape and surface completion
GridFlow: Structured Latent Flow for Seamless City-Scale 3D Point Cloud Generation
Xinyu Wang, Muhammad Ibrahim, Atif Mansoor, Ajmal Mian
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 semantic segmentation maps as condition (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | Chamfer Distance: 0.00350 -> 0.00272 | Improved |
| add semantic segmentation maps as condition (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | JSD: 0.0340 -> 0.0098 | Improved |
| add semantic segmentation maps as condition (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | COV-CD: 59.2% -> 51.8% | Hurt |
| add DSM as condition (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | Chamfer Distance: 0.00340 -> 0.00272 | Improved |
| add DSM as condition (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | COV-CD: 23.0% -> 51.8% | Improved |
| add semantic and DSM to satellite-only conditioning (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | Chamfer Distance: 0.00389 -> 0.00272 | Improved |
| add semantic and DSM to satellite-only conditioning (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | F-Score: 0.8244 -> 0.8840 | Improved |
| add classifier-free guidance at inference (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | Chamfer Distance: 0.00362 -> 0.00272 | Improved |
| add classifier-free guidance at inference (ablation) | GridFlow · City3D-MultiGen Melbourne (conditioning ablation) | 1-NNA-CD: 76.0% -> 46.5% | Improved |
| generate a different city (zero-shot from Melbourne to London) | GridFlow · HoliCity London | Chamfer Distance: 0.00272 -> 0.00394 | Hurt |
| GridFlow versus PVD, the strongest baseline | GridFlow · City3D-MultiGen Melbourne test | Chamfer Distance: 0.00382 -> 0.00272 | Improved |
| GridFlow versus PVD on London zero-shot | GridFlow · HoliCity London (zero-shot from Melbourne) | Chamfer Distance: 0.00542 -> 0.00394 | Improved |
Closest papers by meaning
- BuildAnyPoint: 3D Building Structured Abstraction from Diverse Point Clouds (CVPR 2026 · similarity 0.88)
- Modeling Urban Scenes From Pointclouds (ICCV 2017 · similarity 0.88)
- Building3D: A Urban-Scale Dataset and Benchmarks for Learning Roof Structures from Point Clouds (ICCV 2023 · similarity 0.88)
- Generative Gaussian Splatting for Unbounded 3D City Generation (CVPR 2025 · similarity 0.87)
- CymbaDiff: Structured Spatial Diffusion for Sketch-based 3D Semantic Urban Scene Generation (NeurIPS 2025 · similarity 0.86)
- CityDreamer: Compositional Generative Model of Unbounded 3D Cities (CVPR 2024 · 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.
