Continuous Iterative Image Editing Based on Diffusion Models
Xiaoyu Hu, Chang Wan, Zhigang Wang · 2025
Recently, there have been great advances in image editing. In this paper, we study the text-driven guided image continuous editing based on the diffusion model to achieve higher- quality editing images. Using the existing model to achieve text extraction mask and combined with the diffusion editing model, to achieve accurate position editing. At the same time, through the characteristics of the diffusion model, the latent space is used to achieve high-quality continuous editing of the image, reducing the impact of noise and artifacts generated by continuous editing. Finally, high-frequency image features are extracted and combined with the image features in the backbone network to enhance the detailed effect of image editing.