FluxSpace：整流流Transformer中的解耦语义编辑
FluxSpace: Disentangled Semantic Editing in Rectified Flow Transformers
AI总结：
针对整流流模型难以实现图像解耦编辑的问题，本文提出领域无关的FluxSpace方法，利用整流流Transformer的表征空间实现语义可控的多类图像编辑，兼具可扩展性与解耦能力。
摘要：
整流流模型已成为图像生成领域的主流方法，在高质量图像合成方面展现出出色能力。然而，尽管整流流模型在视觉生成中效果显著，却往往难以实现图像的解耦编辑，这一局限导致无法在不影响图像无关属性的前提下，进行精准的特定属性修改。本文提出FluxSpace——一种领域无关的图像编辑方法，它利用一个表征空间来控制由整流流Transformer（如Flux）生成图像的语义。通过利用整流流模型中Transformer模块学习到的表征，我们提出了一组具备语义可解释性的表征，可支撑从细粒度图像编辑到艺术创作的各类图像编辑任务。本工作提供了一种可扩展且有效的图像编辑方法，同时具备解耦编辑能力。
英文摘要：
Rectified flow models have emerged as a dominant approach in image generation, showcasing impressive capabilities in high-quality image synthesis. However, despite their effectiveness in visual generation, rectified flow models often struggle with disentangled editing of images. This limitation prevents the ability to perform precise, attribute-specific modifications without affecting unrelated aspects of the image. In this paper, we introduce FluxSpace, a domain-agnostic image editing method leveraging a representation space with the ability to control the semantics of images generated by rectified flow transformers, such as Flux. By leveraging the representations learned by the transformer blocks within the rectified flow models, we propose a set of semantically interpretable representations that enable a wide range of image editing tasks, from fine-grained image editing to artistic creation. This work offers a scalable and effective image editing approach, along with its disentanglement capabilities.
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。
