MegaStyle：通过一致性文本到图像风格映射构建多样化可扩展的风格数据集
MegaStyle: Constructing Diverse and Scalable Style Dataset via Consistent Text-to-Image Style Mapping
April 9, 2026
作者: Junyao Gao, Sibo Liu, Jiaxing Li, Yanan Sun, Yuanpeng Tu, Fei Shen, Weidong Zhang, Cairong Zhao, Jun Zhang
cs.AI
摘要
本文提出MegaStyle——一种新颖且可扩展的数据构建流程，能够创建内部风格一致、风格间多样化且高质量的风格数据集。我们通过利用当前大型生成模型具备的文本到图像风格映射一致性能力实现这一目标，该模型可根据给定风格描述生成相同风格的图像。基于此，我们构建了包含17万风格提示词和40万内容提示词的多样化平衡提示库，并通过内容-风格提示词组合生成了大规模风格数据集MegaStyle-140万。基于该数据集，我们提出风格监督对比学习来微调风格编码器MegaStyle-Encoder，以提取具有表现力的风格特异性表征，同时训练了基于FLUX的风格迁移模型MegaStyle-FLUX。大量实验证明了保持风格内一致性、风格间多样性和高质量数据对风格数据集的重要性，以及所提MegaStyle-140万数据集的有效性。经MegaStyle-140万训练后，MegaStyle-Encoder和MegaStyle-FLUX可提供可靠的风格相似度度量与泛化性强的风格迁移效果，为风格迁移领域作出重要贡献。更多结果详见项目网站https://jeoyal.github.io/MegaStyle/。
English
In this paper, we introduce MegaStyle, a novel and scalable data curation pipeline that constructs an intra-style consistent, inter-style diverse and high-quality style dataset. We achieve this by leveraging the consistent text-to-image style mapping capability of current large generative models, which can generate images in the same style from a given style description. Building on this foundation, we curate a diverse and balanced prompt gallery with 170K style prompts and 400K content prompts, and generate a large-scale style dataset MegaStyle-1.4M via content-style prompt combinations. With MegaStyle-1.4M, we propose style-supervised contrastive learning to fine-tune a style encoder MegaStyle-Encoder for extracting expressive, style-specific representations, and we also train a FLUX-based style transfer model MegaStyle-FLUX. Extensive experiments demonstrate the importance of maintaining intra-style consistency, inter-style diversity and high-quality for style dataset, as well as the effectiveness of the proposed MegaStyle-1.4M. Moreover, when trained on MegaStyle-1.4M, MegaStyle-Encoder and MegaStyle-FLUX provide reliable style similarity measurement and generalizable style transfer, making a significant contribution to the style transfer community. More results are available at our project website https://jeoyal.github.io/MegaStyle/.