arXiv 论文提出 OverReal 基准：八个文生图模型都会"过度实现"提示词中不该出现的目标
When Chekhov's Gun Fires Anyway: Over-Realization in Text-to-Image Models
做文生图的朋友值得看看这篇：你写"画面里没有猫"或者角色想象一只鸟，八个主流模型全都会硬画出来，连 CLIPScore 都测不出这个问题。论文给了 698 条测试提示词和一个自动检测裁判，可以直接拿来测自己用的模型。
arXiv 论文《When Chekhov's Gun Fires Anyway》定义了文生图中的 over-realization 失败模式：提示词以否定、人物想象、比喻喻体或被遮挡视角等方式提到某目标时，模型仍把它画进图像。作者基于心理空间理论构建 OverReal 基准，包含 OverReal-Gen（698 条覆盖四种情境的提示词）和 OverReal-Det（2,841 张人工标注图像）两部分，用于测量各生成器的 over-realization 率并验证自动裁判的可靠性。
对八个文生图系统的评测显示，所有系统都存在过度实现问题，而 CLIPScore 等常规对齐与偏好指标反而会奖励这一失败。提示词扩写只能部分缓解。机制层面的发现是：扩散骨干网络对"被提到"目标的注意力与"被要求"目标几乎相当，而本应抑制目标出现的提示词线索词获得的注意力低于普通词。
此前研究只深入探讨过否定情境，四种情境合起来的系统性测量尚属首次。对开发者而言，论文指出现有评测指标与人类对图像正确性的判断存在结构性偏差，做提示词工程或评测集设计时需要把 over-realization 率单独纳入考量。
When Chekhov's Gun Fires Anyway: Over-Realization in Text-to-Image Models
In text-to-image generation, a target that the prompt mentions often appears in the generated image in a way the prompt does not call for. We call this failure over-realization. Mental-spaces theory points to the situations where it can happen: under a negation, inside what a character believes or imagines, as the vehicle of a figurative comparison, or behind a viewpoint that hides the target. Prior work has studied only negation in depth, and the conventional alignment and preference metrics do not penalize the failure. We build OverReal to measure it: OverReal-Gen, 698 prompts covering the four situations, on which we measure each generator's over-realization rate; and OverReal-Det, 2,841 generated images human-labeled for whether and how the target is over-realized, on which we verify the automatic judge that produces this rate. Our evaluation of eight text-to-image systems reveals that all of them over-realize, and that conventional metrics such as CLIPScore reward the failure rather than penalize it. Prompt expansion only partly mitigates the failure. Inside the generator, the diffusion backbone attends to a mentioned target almost as much as to a requested one. The cue words that should hold the target back receive less attention than ordinary words.
This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include text-to-image, OverReal, CLIPScore, 扩散模型.
Source attribution: curated from arXiv cs.AI (@Jiahao Huang, Kon Woo Kim, Song-Duo Ma, Yuka Kitamura, Felix Wagner, Yunshan Huang, Sunisth Kumar, Xanh Ho, Yun-Ang Wu, Zhijie Chen, Yunjia Xu, Junfeng Jiang, Zefan Yu, Fei Cheng, Akiko Aizawa).