基于机器学习的汉字字体设计探索

目的 为应对智能时代汉字字体设计的困境，思考汉字字体的未来属性，探索机器学习为字体设计思维与方法带来的多种可能。方法 基于Sternberg提出的创意三层次，分别开展初步探索、深入融合与交叉突变三个层级的基于机器学习的汉字字体设计探索。结果 通过三个实验分别实现了颜宋体设计、字体族群设计和枯枝体设计三种具有不同目的与功能的设计结果。结论 基于不同的设计实验与结果发现，机器学习不仅可以通过改变汉字字体设计的流程与方法来提高设计效率，实现传统设计方法不可能完成的字体设计内容，而且可以与设计师协同探索新的汉字字体设计思维方式，形成新的设计形态，满足逐渐多元化的运用情境。

Exploration of Chinese Characters Font Design Based on Machine Learning

The paper aims to deal with the predicament of Chinese characters font design in the intelligent era, consider the future attributes of Chinese characters fonts, and explore the many possibilities that Machine Learning brings to font design thinking and methods. Based on the three levels of creativity proposed by Sternberg, three levels of Machine Learning-based Chinese characters font design exploration were carried out respectively: preliminary exploration, deep integration and crossover mutation. Through three experiments, the design results of Yan Song font design, Typeface Family design and Deadwood font design with different purposes and functions are achieved. From different design experiments and results, it can be found that Machine Learning can not only improve design efficiency by changing the process and methods of Chinese characters font design, and realize font design content that cannot be completed by traditional design methods, but also can collaborate with designers to explore new Chinese characters font design thinking mode, form an new design form and meet the diverse application situation of fonts.