AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Jumping the Landing Phase: Noise Variance Matching Enables Accurate Few-Step Inversion

Yang Luo, Zhineng Chen, Ya Gao, Xieping Gao, Yu-Gang Jiang

ECCV 2026 · Image generation › Fast diffusion sampling · approach: Diffusion posterior sampling · 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
add NVM variance correction instead of global scaling (ablation)DDIM inversion · HPDv2 (5-step)CLIP-I: 77.45 -> 84.64Improved
add NVM variance correction to DDIM inversion (ablation)DDIM inversion · HPDv2 (5-step)CLIP-I: 67.18 -> 84.64Improved
add NVM variance correction to DDIM inversion (ablation)DDIM inversion · HPDv2 (5-step)LPIPS: 0.4687 -> 0.2159Improved
add global scaling of predicted noise to DDIM inversion (ablation)DDIM inversion · HPDv2 (5-step)CLIP-I: 67.18 -> 77.45Improved
add attention-guided spatial mask to NVM (ablation)NVM · HPDv2 (5-step)PSNR: 20.53 -> 21.04Improved
add attention-guided spatial mask to NVM (ablation)NVM · HPDv2 (5-step)LPIPS: 0.2159 -> 0.2692Hurt
add one ReNoise refinement iteration after NVM+mask (ablation)NVM · HPDv2 (5-step)CLIP-I: 84.48 -> 88.81Improved
add second ReNoise refinement iteration (ablation)NVM · HPDv2 (5-step)CLIP-I: 88.81 -> 93.55Improved
replace ReNoise with NVMSDXL inversion · HPDv2 (5-step)CLIP-I: 88.97 -> 93.55Improved
replace ReNoise with NVMSDXL inversion · MS-COCO (5-step)CLIP-I: 80.32 -> 90.27Improved
apply NVM instead of ReNoise on SD1.5SD1.5 inversion · HPDv2 (5-step)CLIP-I: 85.77 -> 94.10Improved
apply NVM instead of ReNoise on SDXL TurboSDXL Turbo inversion · HPDv2 (5-step)CLIP-I: 81.18 -> 80.35Mixed

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