FreeDiff: Progressive Frequency Truncation for Image Editing with Diffusion Models
This addresses the problem of precise image editing for users of diffusion models, offering a versatile tool without complex modifications, though it appears incremental as it builds on existing frequency insights.
The paper tackles the misalignment problem in text-to-image diffusion models where editing guidance affects broader areas than intended, by introducing FreeDiff which uses progressive frequency truncation to refine guidance without fine-tuning. The method achieves comparable results to state-of-the-art methods across various editing tasks and diverse images.
Precise image editing with text-to-image models has attracted increasing interest due to their remarkable generative capabilities and user-friendly nature. However, such attempts face the pivotal challenge of misalignment between the intended precise editing target regions and the broader area impacted by the guidance in practice. Despite excellent methods leveraging attention mechanisms that have been developed to refine the editing guidance, these approaches necessitate modifications through complex network architecture and are limited to specific editing tasks. In this work, we re-examine the diffusion process and misalignment problem from a frequency perspective, revealing that, due to the power law of natural images and the decaying noise schedule, the denoising network primarily recovers low-frequency image components during the earlier timesteps and thus brings excessive low-frequency signals for editing. Leveraging this insight, we introduce a novel fine-tuning free approach that employs progressive $\textbf{Fre}$qu$\textbf{e}$ncy truncation to refine the guidance of $\textbf{Diff}$usion models for universal editing tasks ($\textbf{FreeDiff}$). Our method achieves comparable results with state-of-the-art methods across a variety of editing tasks and on a diverse set of images, highlighting its potential as a versatile tool in image editing applications.