Wavelet-Guided Semantic Signal Compensation for Inversion-Free Image Editing

2026-07-02

AI Summary

Existing inversion-free image editing methods struggle to effectively deviate from the source distribution during early-stage generation when performing global attribute editing, often resulting in insufficient semantic modification and distorted background structures. To address this limitation, this work proposes a frequency-aware semantic signal compensation strategy that, for the first time, integrates wavelet-based frequency-domain analysis into an inversion-free editing framework. By differentially enhancing high- and low-frequency components during the initial phase of diffusion flow generation, the method strengthens text-guided semantic signals while preserving background structural consistency. This approach achieves a superior balance between global editing strength and background fidelity, consistently outperforming current state-of-the-art methods across diverse attribute editing tasks.

Abstract

Text-guided image editing aims to modify visual content according to a target prompt while preserving the background. Recent inversion-free image editing frameworks such as FlowEdit have demonstrated strong editing capability without requiring inversion. Empirically, FlowEdit can achieve substantial semantic changes under appropriate hyperparameter settings. However, we observe that under certain global attribute shifts, the editing trajectory may not effectively move away from the source distribution in the early timesteps. Our analysis suggests that in the high-noise regime, the dominant manifold-seeking flow toward the data manifold can reduce the influence of the text-conditioned direction, leading to limited global modification while background structures remain only moderately preserved. Inspired by this observation, we propose an inversion-free, frequency-aware semantic compensation strategy that strengthens the effective signal in the early stage of generation, while maintaining structural consistency in the background. The proposed method improves global editing capacity without sacrificing background fidelity.

Problem

• inversion-free image editing
• global attribute shift
• semantic signal compensation
• background preservation
• text-guided editing

Innovation

• wavelet
• inversion-free
• semantic compensation
• frequency-aware
• image editing