AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Compression-Aware One-Step Diffusion Model for JPEG Artifact Removal

Jinpei Guo, Zheng Chen, Wenbo Li, Yong Guo, Yulun Zhang

ICCV 2025 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 19 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 CaVE embeddings as prompt instead of empty string (ablation)CODiff · LIVE-1MUSIQ: 62.56 -> 67.13Improved
use CaVE instead of DAPE for prompt embedding (ablation)CODiff · LIVE-1M-IQA: 0.3793 -> 0.4584Improved
use learnable embeddings instead of empty prompt (ablation)CODiff · LIVE-1MUSIQ: 62.56 -> 63.39No clear effect
use CaVE instead of empty prompt (ablation)CODiff · DIV2K-ValMUSIQ: 56.68 -> 61.83Improved
add implicit reconstruction learning to explicit QF learning in CaVE (ablation)CaVE · LIVE-1MUSIQ: 62.39 -> 67.13Improved
add explicit QF learning to implicit-only CaVE (ablation)CaVE · LIVE-1MUSIQ: 64.74 -> 67.13Improved
add DISTS loss to L2 loss (ablation)CODiff · LIVE-1/DIV2K ablationLPIPS: 0.3976 -> 0.2109Improved
add GAN loss to L2 + DISTS (ablation)CODiff · LIVE-1/DIV2K ablationMUSIQ: 72.69 -> 73.34Improved
add GAN loss to L2 + DISTS (ablation)CODiff · LIVE-1/DIV2K ablationLPIPS: 0.2109 -> 0.2113No clear effect
add GAN loss to L2 only (ablation)CODiff · LIVE-1/DIV2K ablationLPIPS: 0.3976 -> 0.3899No clear effect
replace OSEDiff-style DAPE prompt extractor with lightweight CaVEone-step diffusion · Complexity 1024x1024Params (G): 1.40 -> 1.00Improved
one-step CODiff instead of 50-step DiffBIRdiffusion JPEG restoration · Complexity 1024x1024inference time (s): 50.81 -> 0.57Improved

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