面向复杂图像编辑的鲁棒序列分解
Towards Robust Sequential Decomposition for Complex Image Editing
把复杂指令式编辑拆成多步序列执行，并用合成数据管线构造带高质量分解序列的大规模编辑数据集，再与真实编辑数据联合训练实现 sim-to-real 迁移；对做指令编辑、组合式编辑的研究者，这套分解范式与数据构造方式可直接复用。
原文依据 we examine editing behaviors of different paradigms under a unified in-context editing framework
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期视觉生成模型的进展使得由人类指令引导的高保真图像编辑成为可能。然而，这些模型往往难以处理涉及组合式编辑操作或步骤间依赖关系的复杂指令。这一困难源于两种典型范式的局限：(1) 单轮编辑试图一次性施加所有指令编辑，往往无法准确解析复杂指令并造成非期望的编辑；(2) 序列编辑虽可将任务分解为更简单的步骤，却受制于顺序执行引入的误差累积，导致结果保真度偏低。为得到复杂图像编辑的鲁棒解决方案，我们在统一的上下文编辑框架下考察不同范式的编辑行为，并研究如何在序列分解带来的收益与其误差累积的弊端之间取得平衡。我们进一步开发了一条合成数据管线，用于构造指令复杂度各异的编辑任务，从而整理出带有高质量分解序列的大规模编辑数据集。通过在合成数据上微调，我们发现：在恰当设计的编辑范式下，即便任务复杂度上升，序列分解仍能带来稳健的性能提升。此外，通过与真实世界编辑数据联合训练，从合成任务中学到的分解技能可以迁移到真实图像上，展现出仿真到真实泛化在更广泛领域处理复杂图像编辑方面的潜力。
英文原文摘要 1568 字符
Recent advances in visual generative models have enabled high-fidelity image editing guided by human instructions. However, these models often struggle with complex instructions involving combinatorial editing operations or inter-step dependencies. This difficulty stems from the limitations of two canonical paradigms: (1) single-turn editing, which attempts to apply all instructed edits in one pass, often fails to parse the complex instruction accurately and causes undesired edits; and (2) sequential editing can decompose the task into simpler steps but suffers from compounding errors introduced by the sequential execution, leading to low-fidelity results. To derive a robust solution for complex image editing, we examine editing behaviors of different paradigms under a unified in-context editing framework, and study how the benefits of sequential decomposition can be balanced against its error-accumulation drawbacks. We further develop a synthetic data pipeline that constructs editing tasks of varying instruction complexity, allowing us to curate a large-scale editing dataset with high-quality decomposed sequences. By finetuning on synthetic data, we discovered that with properly designed editing paradigms, sequential decomposition yields robust improvements even as task complexity increases. Furthermore, the decomposition skills learned from synthetic tasks can transfer to real images by co-training with real-world editing data, demonstrating the promise of sim-to-real generalization for tackling complex image editing across broader domains.