FDS: Frequency-Aware Denoising Score for Text-Guided Latent Diffusion Image Editing
This work addresses a specific issue in text-guided image editing for users of T2I models, offering an incremental improvement by focusing on frequency-band optimization to reduce unintended modifications.
The paper tackled the problem of text-guided image editing in latent diffusion models, which often introduces unintended modifications like loss of detail and color changes, by introducing a frequency-aware denoising score method that selectively optimizes specific frequency bands in localized regions, resulting in high-quality and precise edits as demonstrated through quantitative evaluations and user studies.
Text-guided image editing using Text-to-Image (T2I) models often fails to yield satisfactory results, frequently introducing unintended modifications, such as the loss of local detail and color changes. In this paper, we analyze these failure cases and attribute them to the indiscriminate optimization across all frequency bands, even though only specific frequencies may require adjustment. To address this, we introduce a simple yet effective approach that enables the selective optimization of specific frequency bands within localized spatial regions for precise edits. Our method leverages wavelets to decompose images into different spatial resolutions across multiple frequency bands, enabling precise modifications at various levels of detail. To extend the applicability of our approach, we provide a comparative analysis of different frequency-domain techniques. Additionally, we extend our method to 3D texture editing by performing frequency decomposition on the triplane representation, enabling frequency-aware adjustments for 3D textures. Quantitative evaluations and user studies demonstrate the effectiveness of our method in producing high-quality and precise edits.