提出近可逆微分方程求解器，提升图像编辑稳定性与保真度。
Stable and Near-Reversible Diffusion ODE Solvers for Image Editing
- 采用近可逆龙格-库塔方法替代严格可逆求解器
- 在大幅修改时保持输出质量，背景保留更佳
- 适合需要高保真度的文本引导图像编辑任务
扩散模型的反演在图像编辑中起核心作用。代数可逆的常微分方程（ODE）求解器通过消除基于DDIM的编辑流程中的反演误差，为文本引导的图像编辑提供了有吸引力的方案。然而，实验表明仅保证可逆性不足：当编辑涉及较大语义或视觉变化时，可逆求解器常出现不稳定性，导致输出质量急剧下降。本文揭示，精确可逆性与数值稳定性之间的权衡，在实际编辑中表现为背景保留与提示对齐之间的矛盾。为此，我们研究了近可逆的龙格-库塔方法作为更稳定的替代方案。结合向量场平滑策略后，该方法提升了编辑保真度，在大范围编辑下仍保持稳定，并较好保留了可逆求解器的背景保留优势。
原文摘要 · Abstract (English)
The inversion of diffusion models plays a central role in image editing. Algebraically reversible ODE solvers provide an appealing approach to diffusion inversion for text-guided image editing, by eliminating the inversion error inherent in DDIM-based editing pipelines. However, empirical results indicate that reversibility alone is insufficient. As edits require larger semantic or visual changes, reversible diffusion solvers often exhibit instabilities and suffer sharp drops in output quality. In this paper, we show that the trade-off between exact reversibility and numerical stability manifests empirically as a trade-off between background preservation and prompt alignment in image editing. We then investigate the use of near-reversible Runge-Kutta methods as a more stable alternative to exactly reversible diffusion schemes. When combined with a vector-field smoothing strategy, the resulting approach improves edit fidelity, remains stable under large edits, and largely retains the background-preservation benefits of reversible solvers.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。