Style Conversion of Zhuang Brocade Patterns Based on Low Rank Adaptive and Deep Learning

As a key bearer of traditional Zhuang culture, Zhuang brocade patterns face challenges such as low design efficiency and a stylistic disconnect from modern demands. Therefore, this study innovatively proposes an integrated model that combines low rank adaptive fine-tuning, stable diffusion model, and generative adversarial network for intelligent pattern generation and style transformation. This model adopts an image-to-image method based on GAN for pattern generation, and introduces an improved synthesis and cascading transformation method to achieve differentiated style conversion of brocade patterns. In performance validation experiments, the model demonstrates clear clustering boundaries in complex pattern feature recognition, with a peak signal-to-noise ratio of 32.5 dB, an initial style transfer score of 78.3, and an average diversity index of 0.78. When processing 400 data samples, the model requires 1123 MB of memory and responds within 122 ms, significantly outperforming comparison models in both performance and efficiency. The results show that the model proposed in the study has advantages in the intelligent generation of Zhuang brocade patterns and style transfer generation quality, style adaptability and comprehensive use of resources. The research results provide an effective solution for intelligent pattern generation and modern style adaptation, with broader potential applications in protecting and innovating intangible cultural heritage.