Research map / Point clouds / Shape and surface completion
Fast Point Cloud Generation With Straight Flows
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| replace PVD's stochastic diffusion sampling with one-step straight-flow generation | point cloud unconditional generation · Airplane, Chair, and Car | sampling time (s): 29.9 -> 0.04 | Helped |
| use one-step PSF instead of PVD | unconditional point cloud generation · Airplane | CD: 73.82 -> 71.11 | Helped |
| use one-step PSF instead of PVD | unconditional point cloud generation · Chair | CD: 56.26 -> 58.92 | Hurt |
| use one-step PSF instead of PVD | unconditional point cloud generation · Car | CD: 54.55 -> 57.19 | Hurt |
| use one-step PSF instead of PVD | point cloud completion · Airplane | EMD: 1.030 -> 1.004 | Helped |
| use one-step PSF instead of PVD | point cloud completion · Chair | EMD: 2.939 -> 2.937 | Helped |
| use one-step PSF instead of PVD | point cloud completion · Car | EMD: 2.146 -> 2.194 | Hurt |
| use Chamfer distance rather than ℓ2 for distillation (ablation) | PSF · Airplane, Chair, and Car | 1-NNA, CD: 79.5 -> 71.1; 68.4 -> 59.9; 67.8 -> 57.1 | Helped |
| use Chamfer distance rather than ℓ2 for distillation (ablation) | PSF · Airplane, Chair, and Car | 1-NNA, EMD: 73.5 -> 65.0; 62.0 -> 54.3; 67.0 -> 56.0 | Helped |
| add reflow to reduce trajectory transport cost and make the flow straighter (ablation) | initial ODE flow model · unconditional point cloud generation | sampling quality | Helped |
| add distillation to the reflow model (ablation) | PSF · one-step point cloud generation | sampling quality | Helped |
| use a one-step model instead of the reflow model at up to 20 steps (ablation) | PSF · unconditional point cloud generation | sampling quality | No clear effect |
Closest papers by meaning
- SuperPC: A Single Diffusion Model for Point Cloud Completion, Upsampling, Denoising, and Colorization (CVPR 2025 · similarity 0.89)
- TIGER: Time-Varying Denoising Model for 3D Point Cloud Generation with Diffusion Process (CVPR 2024 · similarity 0.88)
- StraightPCF: Straight Point Cloud Filtering (CVPR 2024 · similarity 0.88)
- Point Cloud Pre-training with Diffusion Models (CVPR 2024 · similarity 0.88)
- PointInfinity: Resolution-Invariant Point Diffusion Models (CVPR 2024 · similarity 0.87)
- Not-So-Optimal Transport Flows for 3D Point Cloud Generation (ICLR 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.