AI Paper Atlas

Research map / Image generation / Image restoration

Weak Diffusion Priors Can Still Achieve Strong Inverse-Problem Performance

Jing Jia, Wei Yuan, Sifan Liu, Liyue Shen, Guanyang Wang

ICML 2026 · Image generation › Image restoration · approach: Inverse imaging with learned priors · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
use a 3-step in-domain DDIM prior with initial-noise optimization instead of 1000-step DPSinitial-noise optimization · CelebA inpaintingPSNR: 31.98 -> 33.78Improved
use a bedroom-trained out-of-domain weak prior for CelebA inpainting vs DPSinitial-noise optimization · CelebA inpaintingPSNR: 31.98 -> 32.76Improved
use a 3-step prior with initial-noise optimization instead of DPSinitial-noise optimization · CelebA 4x super-resolutionPSNR: 26.82 -> 31.27Improved
use CelebA-trained prior instead of Church-trained prior (domain mismatch)initial-noise optimization · Church inpaintingPSNR: 24.93 -> 23.56Hurt
use in-domain Bedroom prior instead of CelebA priorinitial-noise optimization · Bedroom inpaintingPSNR: 27.78 -> 28.88Improved
use our optimizer and early stopping instead of DMPluginitial-noise optimization · CelebA inpaintingPSNR: 32.778 -> 33.784Improved
use our method instead of DMPlug on Church inpaintinginitial-noise optimization · Church inpaintingPSNR: 24.957 -> 24.927No clear effect
add HoldoutTopK early stopping instead of variance-based stopping (ablation)Adam · Bedroom inpaintingPSNR: 28.669 -> 29.156Improved
add HoldoutTopK early stopping instead of variance-based stopping (ablation)AdamSphere · Bedroom inpaintingPSNR: 28.559 -> 29.162Improved
add AdamSphere in place of Adam (with HoldoutTopK) (ablation)HoldoutTopK · Bedroom inpaintingPSNR: 29.156 -> 29.162No clear effect
add AdamSphere in place of Adam (with variance stopping) (ablation)variance early stopping · Church Gaussian deblurringPSNR: 22.034 -> 22.048No clear effect
increase box-inpainting masked fractionweak prior vs DPS · Church box inpaintingLPIPSHurt

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy