现有三维雕刻模型生成研究中注重模型网格的生成，而少涉及雕刻材质生成。本文结合现实中雕刻作品的制作步骤，以及雕刻作品中细节的表现手法，提出通过分离雕刻模型主体与材质的方式优化雕刻模型的生成。利用雕刻模型的材质部分表现雕刻的细节，对三维网格部分生成后进行优化以减小生成模型的数据量。实验结果表明，可根据文物拓片和线稿图像，生成细节表现更加丰富且数据量更小的三维雕刻模型。

Existing research on 3D carving generation focuses on the generation of model grids, and less on the generation of carving materials. So this article combines the production steps of carving works in reality and the expression techniques of details in carving works, and proposes to optimize the generation of carving models by separating the main body and materials of the carving model. The material part of the sculpting model is used to represent the details of the sculpting, and then the 3D mesh part is optimized to reduce the data volume of the generative model. The experimental results show that this article can generate 3D carving models with richer details and smaller data volume based on cultural relic rubbings and line draft images.