MultiBanana：一个面向多参考文本到图像生成的挑战性基准
MultiBanana: A Challenging Benchmark for Multi-Reference Text-to-Image Generation
为多参考主体驱动生成提供了覆盖参考数量（最多8张）、跨域、尺度不匹配、罕见概念与多语言文本的评测协议，可直接用于诊断个性化生成模型在异构参考组合下的失败模式。
原文依据 we introduce MultiBanana, which is designed to assess the edge of model capabilities by widely covering problems specific to multi-reference settings
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的文本到图像生成模型已具备多参考生成与编辑能力，即从多张参考图像中继承主体的外观，并在新的场景中重新渲染它们。然而，现有的基准数据集往往只关注使用单张或少量参考图像的生成，这使我们无法衡量模型性能的进展，也无法识别模型在遵循包含更多参考图像的指令时存在的弱点。此外，它们的任务定义仍然模糊，局限于“编辑什么”或“给定多少张参考图像”等维度，因而无法刻画组合异构参考图像所固有的挑战。为弥补这一空白，我们提出了 MultiBanana，它通过广泛覆盖多参考设定特有的问题来评估模型能力的边界：（1）改变参考图像数量（最多 8 张），（2）参考图像之间的域不匹配（例如照片与动漫），（3）参考图像与目标场景之间的尺度不匹配，（4）参考图像包含罕见概念（例如红色的香蕉），以及（5）用于渲染的多语言文本参考。我们对多种文本到图像模型的分析揭示了它们各自的性能、典型失败模式以及有待改进的方面。
英文原文摘要 1297 字符
Recent text-to-image generation models have acquired the ability of multi-reference generation and editing; that is, to inherit the appearance of subjects from multiple reference images and re-render them in new contexts. However, existing benchmark datasets often focus on generation using a single or a few reference images, which prevents us from measuring progress in model performance or identifying weaknesses when following instructions with a larger number of references. In addition, their task definitions are still vague, limited to axes such as "what to edit" or "how many references are given", and therefore fail to capture the challenges inherent in combining heterogeneous references. To address this gap, we introduce MultiBanana, which is designed to assess the edge of model capabilities by widely covering problems specific to multi-reference settings: (1) varying the number of references (up to 8), (2) domain mismatch among references (e.g., photo vs. anime), (3) scale mismatch between reference and target scenes, (4) references containing rare concepts (e.g., a red banana), and (5) multilingual textual references for rendering. Our analysis among a variety of text-to-image models reveals their respective performances, typical failure modes, and areas for improvement.