PulseAugur
EN
LIVE 14:26:49

SONIC method optimizes noise for advanced image inpainting

Researchers have developed SONIC, a novel training-free method for image inpainting using existing text-to-image models. The technique optimizes the initial noise sample to reconstruct the unmasked image efficiently. Key innovations include a linear approximation to relate noise to model output without costly unrolling and spectral preconditioning for stable optimization. SONIC reportedly outperforms state-of-the-art methods on various inpainting tasks. AI

IMPACT This method could improve the efficiency and quality of image inpainting tasks by leveraging existing text-to-image models without retraining.

RANK_REASON The cluster contains a research paper detailing a new method for image inpainting. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

SONIC method optimizes noise for advanced image inpainting

COVERAGE [1]

  1. arXiv cs.CV TIER_1 English(EN) · Seungyeon Baek, Erqun Dong, Shadan Namazifard, Mark J. Matthews, Kwang Moo Yi ·

    SONIC: Spectral Optimization of Noise for Inpainting with Consistency

    arXiv:2511.19985v3 Announce Type: replace Abstract: We propose a novel training-free method for inpainting with off-the-shelf text-to-image models. While guidance-based methods in theory allow generic models to be used for inverse problems such as inpainting -- in practice their …