AIGC驱动下的戏曲IP设计转译与多模态生成研究OA
Research on the Translation Mechanism and Multi-Modal Generation of Traditional Opera IP Design Driven by AIGC
生成式人工智能(AIGC)的演进为非遗戏曲 IP 的数字化生存提供了新动能.本文针对戏曲 IP 开发中"程式化还原难"与"动态转化成本高"的痛点,构建闭环技术落地方案.研究以豫剧《花木兰》为核心,通过构建"垂直领域 LoRA 模型"实现角色设计的精准控制,并利用Live2D 与 Ebsynth 等时序一致性技术,打通从静态角色设计到宣传短视频生成的转译路径.本文系统阐述数据集构建、模型调优、骨架绑定及多模态渲染的实施步骤,为戏曲 IP 的智能化、低成本、高质感传播提供可复用的方法论.
The evolution of Artificial Intelligence Generated Content(AIGC)has provided new momentum for the digital survival of intangible cultural heritage(ICH)opera IPs.Addressing the critical pain points in traditional opera IP development―namely the"difficulty of formulaic restoration"and the"high cost of dynamic transformation"―this paper proposes a closed-loop technical solution.Centered on the Henan Opera Hua Mulan,the research achieves precise control over character design by constructing a"vertical domain LoRA model."Furthermore,it utilizes temporal consistency technologies such as Live2D and Ebsynth to bridge the translation path from static character design to the generation of promotional short videos.This study elaborates on the implementation steps of dataset construction,model fine-tuning,skeletal rigging,and multi-modal rendering,providing a replicable methodology for the intelligent,low-cost,and high-quality dissemination of traditional opera IPs.
游杰
郑州西亚斯学院,河南 郑州 451150
社会科学
AIGC戏曲IP豫剧花木兰LoRA微调
aigctraditional opera iphenan opera hua mulanlora fine-tuning
《鞋类工艺与设计》 2026 (15)
17-19,3
2025年度郑州西亚斯学院省部级培育项目《AIGC驱动豫剧<花木兰>IP形象的数智创生实践研究》(项目编号:2025XKB010)
评论