jaddai/openbrush-renoir

OpenBrush Renoir是一个从OpenBrush-75K数据集中精选的子集，专门包含印象派画家皮埃尔-奥古斯特·雷诺阿的1400幅艺术作品。数据集提供了每幅画作的结构化视觉语言模型（VLM）生成描述，基于v2结构化模式，涵盖标签、主题、动作、场景、情绪、风格、照明、色彩和构图等9个字段，以及完整的描述文本。这些描述突出了雷诺阿印象派风格的特点，如肤色处理、织物质感和柔和的边缘融合。数据集旨在方便用户直接获取雷诺阿作品，避免下载整个大型数据集。原始艺术作品属于公共领域，结构化描述在MIT许可证下发布。

OpenBrush Renoir is a curated subset of the OpenBrush-75K dataset, containing all Pierre-Auguste Renoir works with structured VLM captions. It includes 1,400 images of Renoirs figure-and-portrait Impressionism paintings, with captions generated using the Qwen3-VL-30B-A3B model. The captions follow a v2 structured schema covering TAGS, SUBJECT, ACTION, SETTING, MOOD, STYLE, LIGHTING, COLOR, and COMPOSITION fields, along with a full caption text. This subset is designed to provide focused access to Renoirs works without needing to download the entire 75,313-image parent dataset. The underlying artworks are in the public domain, and the structured captions are released under the MIT license.