Research map / Image generation / Fast diffusion sampling
Triadic Dynamics Aware Diffusion Posterior Sampling for Inverse Problems: Optimizing Guidance and Stochasticity Schedules
Junseo Bang, Dong Ju Mun, Hoigi Seo, Seongmin Hong, Se Young Chun
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 |
|---|---|---|---|
| schedule DC, CFG and stochasticity jointly instead of keeping all fixed (ablation) | TriPS (SD3.5-M) · FFHQ SR x8 | LPIPS: 0.155 -> 0.101 | Improved |
| schedule only DC guidance (others fixed) (ablation) | TriPS · FFHQ SR x8 | LPIPS: 0.155 -> 0.125 | Improved |
| schedule only CFG (others fixed) (ablation) | TriPS · FFHQ SR x8 | PSNR: 29.04 -> 29.05 | No clear effect |
| schedule only stochasticity (others fixed) (ablation) | TriPS · FFHQ SR x8 | LPIPS: 0.155 -> 0.133 | Improved |
| add GRPO schedule optimization on top of template search | TriPST · FFHQ SR x8 (flow matching) | FID: 26.66 -> 22.18 | Improved |
| add GRPO schedule optimization on top of template search | TriPST · FFHQ SR x8 (flow matching) | PSNR: 29.03 -> 28.55 | Hurt |
| use distortion-weighted GRPO reward (w_dist 0.9) (ablation) | TriPS_G · FFHQ Gaussian deblurring | PSNR: 30.33 -> 30.62 | Improved |
| use perception-weighted GRPO reward (w_perc 0.7) (ablation) | TriPS_G · FFHQ Gaussian deblurring | LPIPS: 0.070 -> 0.067 | Improved |
| transfer schedule optimized on SR x8 to Gaussian deblurring | TriPS_G · FFHQ 100 samples | PSNR: 27.74 -> 28.90 | Improved |
| transfer schedule optimized on SR x8 to SR x12 | TriPS_G · FFHQ 100 samples | LPIPS: 0.148 -> 0.099 | Improved |
| apply TriPS schedules to SD1.5 diffusion backbone vs TReg | TriPS · FFHQ SR x8 (50 NFE) | PSNR: 26.34 -> 28.02 | Improved |
| raise CFG scale early in sampling (ablation) | posterior sampling · DIV2K SR x8 | residual norm decay | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- FlowDPS — FlowDPS : Flow-Driven Posterior Sampling for Inverse Problems (ICCV 2025) · better on 4 of 4 shared results
- FlowChef — FlowChef: Steering of Rectified Flow Models for Controlled Generations (ICCV 2025) · better on 2 of 2 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)
- SITCOM: Step-wise Triple-Consistent Diffusion Sampling For Inverse Problems (ICML 2025 · similarity 0.90)
- Learning to Sample From Diffusion Models via Inverse Reinforcement Learning (NeurIPS 2026 · similarity 0.90)
- A Variational Perspective on Solving Inverse Problems with Diffusion Models (ICLR 2024 · similarity 0.90)
- Tempered Guided Diffusion (NeurIPS 2026 · similarity 0.89)
- Coupled Data and Measurement Space Dynamics for Enhanced Diffusion Posterior Sampling (NeurIPS 2025 · similarity 0.89)
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.
