ZZEdit: ZigZag Trajectories of Inversion and Denoising for Zero-shot Image Editing

Li Maomao, Yu Li, Yunfei Liu, Dong Xu

Keywords: ZigZag Trajectories，Zero-shot ，Image Editing

Abstract: Editability and fidelity are two essential demands for text-driven image editing, which expects that the editing area should align with the target prompt and the rest should remain unchanged separately. The current cutting-edge editing methods usually obey an ''inversion-then-editing'' pipeline, where the input image is first inverted to an approximate Gaussian noise $z_T$ with $T$ steps, based on which a sampling process is performed using the target prompt. Nevertheless, we argue that \textit{it is not a good choice to use a near-Gaussian noise as a pivot for further editing since it almost lost all structure fidelity.} To verify this, we conduct a pilot experiment and find that the target prompt has different guiding degrees towards those latents on the inversion trajectory. Thus, a structure-preserving while sufficient-for-editing point is a more suitable pivot. Based on this, we propose a novel editing paradigm dubbed ZZEdit, which first locates such a pivot during the inversion trajectory and then mildly strengthens target guidance via the proposed ZigZag process. Concretely, our ZigZag process fulfills denoising and inversion iteratively, which gradually approaches the target while still holding background fidelity. Afterwards, to achieve the same number of inversion and denoising steps, we perform a pure sampling process under the target prompt. Extensive experiments highlight the effectiveness of our ZZEdit paradigm in diverse image editing scenarios compared with the existing ''inversion-then-editing'' pipeline.

Primary Area: generative models

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 1725