StoryDB

该数据集包含10000组以角色为中心的图像-文本对，每组图像展现同一角色在不同场景下穿着一致的服装，总计有100000张图像。每组包含5至12张图像及相应的提示，还有共享提示元素索引和角色遮罩图像。该数据集支持Storynizor的训练，并为未来在故事生成和IP一致性内容创作方面的研究提供了资源。规模上，数据集跨越10000组，总计100000张图像。其任务旨在故事生成以及在图像生成中保持角色的一致性。

This dataset consists of 10,000 character-centric image-text pairs, with a total of 100,000 images across all sets. Each set displays images of the same character wearing consistent outfits across diverse scenarios, and includes 5 to 12 images along with their corresponding prompts, shared prompt element indices, and character mask images. This dataset supports the training of Storynizor, and serves as a valuable resource for future research on story generation and IP-consistent content creation. The core tasks enabled by this dataset focus on story generation and maintaining character consistency in image generation.