NewEpisode

NewEpisode数据集由南洋理工大学创建，专注于评估生成模型在仅使用单个示例故事的情况下生成新故事和新角色的适应性。该数据集通过精炼文本提示和消除角色泄露，确保了数据的质量和一致性。NewEpisode数据集特别适用于视觉故事生成模型，强调新角色与现有故事线的无缝整合，为故事角色定制提供了更细致的方法。此数据集的应用领域包括连续漫画等，旨在解决现有模型在处理新角色时的一致性和适应性问题。

The NewEpisode dataset, developed by Nanyang Technological University (NTU), focuses on evaluating the adaptability of generative models when generating new stories and characters using only a single exemplar story. To guarantee data quality and consistency, this dataset adopts refined text prompts and eliminates character leakage. Specifically designed for visual story generation models, NewEpisode emphasizes the seamless integration of new characters into existing storylines, offering a more fine-grained approach to story character customization. Its application domains include sequential comics and other related fields, aiming to address the consistency and adaptability issues that existing models face when handling newly introduced characters.