Dual-Scale Condition-Shift Editing for Robust Real-World Image Demoir\'eing

Abstract

Robust demoir\'eing of screen-recaptured images remains challenging: moir\'e varies in frequency, color, and spatial support across display-camera pairs and capture conditions. Dedicated restoration networks trained on finite paired data struggle with this diversity, especially when low-frequency color banding and high-frequency stripes co-occur. We present Demoir\'eEdit, a content-preserving editing framework that adapts the natural-image prior of a pretrained flow-matching editor to demoir\'eing. First, an endpoint-supervised training objective supplements velocity matching with pixel, perceptual, and frequency-domain restoration losses. Second, a dual-scale condition-shift strategy performs global correction at low resolution, then refines the image at full resolution by shifting the denoising condition from the first-stage output to the original input for detail recovery. We also introduce DailyMoir\'e130, a benchmark of challenging moir\'e in everyday screen photography, comprising 130 smartphone captures across diverse displays with low-frequency color banding, fine high-frequency stripes, and mixed-frequency interference. On the challenging DailyMoir\'e130 benchmark, our method achieves the lowest residual moir\'e area and receives 71.3% overall-quality preference in a blind Best-of-7 study with 31 participants. We also evaluate on paired FHDMi/UHDM, reducing residual moir\'e confidence by 47.1%/9.3% and increasing MUSIQ by 4.3%/2.0% over the best baseline for each metric.