Research map / Image generation / Fast diffusion sampling
Geometry-Correct Diffusion Posterior Sampling with Denoiser-Pullback Curvature Guidance and Manifold-Aligned Damping
Seunghyeok Shin, Minwoo Kim, Dabin Kim, Hongki Lim
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 variance-preserving Langevin transition instead of ODE-style update (ablation) | CLAMP · FFHQ/ImageNet inverse problems | reconstruction quality | Improved |
| add prior-aligned rank-one damping (ablation) | CLAMP · FFHQ/ImageNet inverse problems | reconstruction quality | Improved |
| keep the denoiser pullback J_F^T in the curvature model (ablation) | CLAMP · FFHQ/ImageNet inverse problems | accuracy | Improved |
| use full curvature including forward J_F instead of one-sided (ablation) | CLAMP · FFHQ/ImageNet inverse problems | accuracy vs runtime | No clear effect |
| use GMRES instead of CG for the nonsymmetric system (ablation) | CLAMP · FFHQ/ImageNet inverse problems | stability | Improved |
| CLAMP vs SITCOM on pixel FFHQ motion deblurring | CLAMP · FFHQ motion deblurring | run-time: 4.14x speedup | Improved |
| CLAMP vs Latent DAPS in latent space | Latent CLAMP · latent-space inverse problems | run-time: 2.4x faster | Improved |
| CLAMP vs ReSample in latent space | Latent CLAMP · latent-space inverse problems | run-time: 9x faster | Improved |
| CLAMP vs best baseline on MRI x4 | CLAMP · SKM-TEA MRI x4 | PSNR: 32.21 -> 34.05 | Improved |
| CLAMP vs best baseline on MRI x8 | CLAMP · SKM-TEA MRI x8 | PSNR: 30.29 -> 32.27 | Improved |
| CLAMP vs DAPS on ImageNet phase retrieval | CLAMP · ImageNet phase retrieval | PSNR: 22.354 -> 19.680 | Hurt |
| CLAMP vs SITCOM on FFHQ Gaussian deblurring | CLAMP · FFHQ Gaussian deblurring | PSNR: 29.359 -> 29.010 | Hurt |
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.
- SITCOM — SITCOM: Step-wise Triple-Consistent Diffusion Sampling For Inverse Problems (ICML 2025) · better on 9 of 12 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
- Improving Diffusion Posterior Samplers with Lagged Temporal Corrections for Image Restoration (NeurIPS 2026 · similarity 0.90)
- Divide-and-Conquer Posterior Sampling for Denoising Diffusion priors (NeurIPS 2024 · similarity 0.89)
- Provable Posterior Sampling with Denoising Oracles via Tilted Transport (NeurIPS 2024 · similarity 0.88)
- Analyzing and Guiding Zero-Shot Posterior Sampling in Diffusion Models (NeurIPS 2026 · similarity 0.88)
- A Variational Perspective on Solving Inverse Problems with Diffusion Models (ICLR 2024 · similarity 0.87)
- Decomposed Diffusion Sampler for Accelerating Large-Scale Inverse Problems (ICLR 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.
