arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2509.00541cs.GRcs.AIcs.CV
收藏

LatentEdit:用于一致性语义编辑的自适应潜空间控制

LatentEdit: Adaptive Latent Control for Consistent Semantic Editing

Southern University of Science and Technology, China(南方科技大学,中国);Pengcheng Laboratory, China(鹏城实验室,中国)

Siyi Liu, Weiming Chen, Yushun Tang, Zhihai He

AI总结:

本研究提出轻量即插即用的自适应潜空间融合框架LatentEdit,通过动态融合潜码实现可控图像编辑,在PIE-Bench上8-15步即达SOTA水平,无反转变体更提升实时部署效率。

摘要:

基于扩散的图像编辑近年来取得了显著成功。然而,在保持背景相似性的同时实现高质量图像编辑,且不牺牲速度或内存效率,仍然是一项挑战。在本研究中,我们提出了LatentEdit,这是一种自适应潜空间融合框架,可将当前潜码与从源图像反转得到的参考潜码动态结合。通过在高相似度、语义重要区域选择性保留源特征,同时在目标提示引导下在其他区域生成目标内容,LatentEdit可实现细粒度、可控的编辑。关键在于,该方法无需修改内部模型或复杂的注意力机制,是一种轻量的即插即用解决方案,兼容基于UNet和基于DiT的架构。在PIE-Bench数据集上的大量实验表明,我们提出的LatentEdit在保真度和可编辑性之间实现了最优平衡,仅需8-15步即可优于当前最优方法。此外,其无反转变体进一步将神经函数评估次数减半,且无需存储任何中间变量,大幅提升了实时部署效率。

英文摘要:

Diffusion-based Image Editing has achieved significant success in recent years. However, it remains challenging to achieve high-quality image editing while maintaining the background similarity without sacrificing speed or memory efficiency. In this work, we introduce LatentEdit, an adaptive latent fusion framework that dynamically combines the current latent code with a reference latent code inverted from the source image. By selectively preserving source features in high-similarity, semantically important regions while generating target content in other regions guided by the target prompt, LatentEdit enables fine-grained, controllable editing. Critically, the method requires no internal model modifications or complex attention mechanisms, offering a lightweight, plug-and-play solution compatible with both UNet-based and DiT-based architectures. Extensive experiments on the PIE-Bench dataset demonstrate that our proposed LatentEdit achieves an optimal balance between fidelity and editability, outperforming the state-of-the-art method even in 8-15 steps. Additionally, its inversion-free variant further halves the number of neural function evaluations and eliminates the need for storing any intermediate variables, substantially enhancing real-time deployment efficiency.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑