DecFlowEdit: Self-Localized Flow-based Image Editing via Guidance Decoupling

2026-09-28

AI Summary

This study addresses the inherent trade-off between background leakage and editing capability in flow-based image editing by proposing a training-free, inversion-free method for precise local editing. The core innovation lies in a novel guidance decoupling mechanism that separates the optimal guidance scales for localization and editing. By aggregating velocity field discrepancies via classifier-free guidance (CFG) to extract prior information, and incorporating a reweighting strategy to update the generation process, the approach effectively balances background preservation with editing fidelity without requiring external masks or attention manipulation. Evaluated on the PIE-Bench benchmark, the proposed method reduces the background structure distance by 61%–73% and the LPIPS metric by 68%–80%, significantly outperforming existing approaches such as FlowEdit.

Abstract

Flow-based image editing (FlowEdit) enables inversion-free semantic changes through the difference between source and target velocities. In this paper, we observe that FlowEdit's default classifier-free guidance (CFG) configuration, with asymmetric source and target scales, causes substantial background leakage. Matching these guidance scales, for example by removing CFG, improves edit-relevant localization but severely degrades editability. To get the best of both worlds, we propose DecFlowEdit, which decouples the optimal guidance scales for localization and for editing in flow-based generative models. In particular, DecFlowEdit first extracts an edit-relevant prior by temporally aggregating velocity differences evaluated without CFG, and then uses this prior to reweight the original updates under default CFG. Our method remains training-free and inversion-free, requiring neither external spatial masks nor attention manipulation. Experiments on PIE-Bench across FLUX, SD3, and SD3.5 show that DecFlowEdit improves background preservation, reducing structure distance by approximately 61 to 73 percent and background LPIPS by 68 to 80 percent relative to FlowEdit at comparable editing fidelity.

Problem

• Flow-based image editing
• Background leakage
• Classifier-free guidance
• Edit localization
• Editability

Innovation

• Flow-based Image Editing
• Guidance Decoupling
• Classifier-Free Guidance
• Training-free
• Background Preservation