编辑溢出作为探测器：图像编辑模型是否隐式理解世界关系？
Edit Spillover as a Probe: Do Image Editing Models Implicitly Understand World Relations?
AI总结：
研究探讨图像编辑模型中编辑溢出现象，通过系统框架评估不同模型的隐式世界理解能力，发现溢出率差异显著，且语义相关性溢出反映真实理解。
摘要：
遵循指令的图像编辑模型应仅修改指定区域，保持其余部分不变。然而，实践中观察到编辑溢出现象：模型会修改与指定区域语义相关但未指定的内容。这引发核心问题：溢出反映真实隐式世界理解还是注意力泄漏？本文提出EditSpilloverProbe框架，通过系统化方法将编辑溢出作为探测世界知识的自然探针。引入溢出分类学（空间、语义、混合、随机），自动化检测与分类流程，以及基于真实世界中文编辑任务构建的EditSpilloverBench基准数据集。对5种代表性编辑模型的系统评估揭示三个核心发现：（1）溢出率在不同架构间差异显著，从3.49%到11.46%，差异达3.3倍；（2）绝对语义溢出量反映模型的世界理解能力——nano_banana产生最多语义溢出（每张图像27.8），而qwen_2511具有最精确的编辑控制但语义溢出较少（每张图像16.3），揭示编辑控制与世界理解之间的权衡；（3）空间衰减分析显示溢出区域密度随距离呈指数衰减，但语义相关溢出比例保持恒定（40%-58%），直接证明语义溢出反映真实世界理解而非空间扩散。
英文摘要：
Instruction-following image editing models are expected to modify only the specified region while keeping the rest of the image unchanged. However, in practice, we observe a pervasive phenomenon -- edit spillover: models alter semantically related but unspecified content outside the edit region. This raises a fundamental question -- does spillover reflect genuine implicit world understanding, or is it merely attention leakage? We propose EditSpilloverProbe, a systematic framework that repurposes edit spillover as a natural probe for world knowledge in image editing models. We introduce a spillover taxonomy (spatial, semantic, mixed, random), an automated detection-and-classification pipeline, and a benchmark dataset constructed from real-world Chinese text editing tasks, EditSpilloverBench. Systematic evaluation of 5 representative editing models reveals three core findings: (1) spillover rates vary dramatically across architectures, from 3.49% to 11.46%, with a 3.3x ratio; (2) absolute semantic spillover quantity reveals models' world understanding capability -- nano_banana produces the most semantic spillover (27.8 per image), while qwen_2511 has the most precise editing control but lower semantic spillover (16.3 per image), revealing a trade-off between editing control and world understanding; (3) spatial decay analysis shows spillover area density decays exponentially with distance, but the proportion of semantically relevant spillover remains constant (40%-58%), providing direct evidence that semantic spillover reflects genuine world understanding rather than spatial diffusion.
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。
