As a national intangible cultural heritage, Ningbo Mud-Gold Color Paint, with its exquisite craftsmanship and historical and cultural values, occupies an important position in Chinese traditional crafts, and also provides rich inspirations for modern clothing design. This paper constructs a 3D reconstruction model based on Transformer-MVSNet and proposes a fast garment migration network method with multi-style fusion, so as to realize the digital reconstruction of Ningbo Mud-Gold Color Paint images and the innovative design of garments. The study shows that the accuracy difference and overall error of the Transformer-MVSNet model are 0.312mm and 0.339mm, respectively, which is better than other methods in terms of overall error and second only to CVP-MVSNet in terms of accuracy difference, with better comprehensive performance. Moreover, under the same resolution, the memory consumption and running time of the Transformer-MVSNet model are always less than those of other models, realizing high accuracy under low consumption. In addition, the clothing style migration model in this paper outperforms other comparative models in five metrics, including Inception Score (IS), Mode Score (MS), Frechet Inception Distance (FID), Wasserstein Distance, and Maximum Mean Difference (MMD). This paper provides an example for digital reconstruction and innovative design of intangible cultural heritage.
