Research map / Image generation / Autoregressive and latent generation
From Patches to Images: A Nonparametric Generative Model
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 |
|---|---|---|---|
| Add image-specific internal clusters to external eDP (HDP vs eDP) (ablation) | HDP-Grid · BSDS-68 sigma=25 | PSNR: 28.76 -> 28.82 | Helped |
| Add image-specific internal clusters to eDP (ablation) | HDP-Grid · classic-12 sigma=25 | PSNR: 29.47 -> 29.68 | Helped |
| Add image-specific internal clusters to eDP (ablation) | HDP-Grid · classic-12 sigma=10 | PSNR: 33.77 -> 33.99 | Helped |
| Use external clean-image clusters (eDP) instead of internal-only (iDP) (ablation) | DP-Grid · BSDS-68 sigma=50 | PSNR: 25.10 -> 25.75 | Helped |
| Combine internal and external clusters (HDP) instead of internal-only iDP (ablation) | HDP-Grid · classic-12 sigma=25 | PSNR: 29.02 -> 29.68 | Helped |
| Bayesian nonparametric eDP (K=449) instead of EPLL (K=200 fixed) | DP-Grid · classic-12 sigma=25 | PSNR: 29.39 -> 29.47 | Helped |
| Bayesian nonparametric eDP instead of EPLL | DP-Grid · BSDS-68 sigma=50 | PSNR: 25.72 -> 25.75 | Helped |
| HDP-Grid vs BM3D | HDP-Grid · BSDS-68 sigma=25 | PSNR: 28.55 -> 28.82 | Helped |
| HDP-Grid vs BM3D | HDP-Grid · classic-12 sigma=50 | PSNR: 26.55 -> 26.42 | Hurt |
| HDP-Grid vs LSSC | HDP-Grid · classic-12 sigma=10 | PSNR: 34.05 -> 33.99 | Hurt |
| HDP-Grid vs BM3D | HDP-Grid · classic-12 sigma=50 | SSIM: 0.7425 -> 0.7366 | Hurt |
| Use larger 12x12 patches in LSSC | LSSC · sigma=50 | PSNR: 0.2 dB | Helped |
Closest papers by meaning
- Patched Denoising Diffusion Models For High-Resolution Image Synthesis (ICLR 2024 · similarity 0.88)
- ZoomLDM: Latent Diffusion Model for Multi-scale Image Generation (CVPR 2025 · similarity 0.86)
- Image Inpainting via Tractable Steering of Diffusion Models (ICLR 2024 · similarity 0.85)
- The Crystal Ball Hypothesis in diffusion models: Anticipating object positions from initial noise (ICLR 2025 · similarity 0.85)
- Improving Progressive Generation with Decomposable Flow Matching (NeurIPS 2025 · similarity 0.85)
- Hierarchical Patch Diffusion Models for High-Resolution Video Generation (CVPR 2024 · similarity 0.85)
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.