Timestep Rescheduling in Diffusion Inversion

2026-06-13

AI Summary

This work addresses the issue of cumulative reconstruction error in diffusion inversion caused by suboptimal time-step selection. The authors uncover a parabolic distribution pattern of inversion error across time steps and propose a plug-and-play, non-uniform time-step scheduling strategy that incurs no additional parameters or computational overhead. By integrating global scaling with local dynamic programming-based rescheduling, the method optimizes computational resource allocation within accelerated inversion frameworks such as DDIM. Experimental results demonstrate that the proposed approach significantly enhances the accuracy of existing inversion techniques in both image reconstruction and editing tasks, exhibiting strong generalizability and effectiveness.

Abstract

Diffusion inversion, which maps images back to the Gaussian latent space of a diffusion model, is a critical task for image reconstruction and editing. While DDIM enables fast deterministic inversion, it inherently introduces deviations that accumulate into noticeable inversion errors. Existing methods often address this by solving a fixed-point problem but largely overlook how the selection of the diffusion timestep in the noise scheduler influences inversion fidelity. In this work, we reveal that the deviation scale in diffusion inversion is strongly dependent on the timestep size, and exhibits a parabolic trend, with larger errors concentrated at both small and large timesteps. Based on this finding, we propose a simple yet effective nonuniform timestep scheduler that integrates a global rescaling with a local dynamic programming based rescheduling, enabling a strategic allocation of computational effort that minimizes the overall inversion error and preserves higher inversion accuracy. Our method serves as an off-the-shelf enhancement for existing inversion techniques and requires no extra parameters or computational overhead. Through extensive experiments, we verify that integrating our scheduler consistently boosts the performance of existing inversion methods, achieving superior results in image reconstruction and editing.

Problem

• diffusion inversion
• timestep scheduling
• inversion error
• noise scheduler
• image reconstruction

Innovation

• diffusion inversion
• timestep rescheduling
• nonuniform scheduler
• dynamic programming
• inversion error minimization