This paper puts forward a new deep learning based system for automatic generation and style transfer of qipao patterns which can solve the problem of how to combine traditional chinese aestetics and fasion design. System uses GANs along with the neural style transfer technique to come up with new innovative qipao designs that keep the cultural heritage alive but with modern artistic elements We adopt an updated GAN model with multi-scale discriminator and self-attention mechanism as well as a dataset of high-quality traditional Chinese textiles to generate different qipao patterns. The next one is a Convolutional neural network using the style transfer module to get the desired art style applied to the generated pattern so the different aesthetic can be combined. Experiment proved, our method got FID score reaching 9.5 points, which greatly surpassed other baseline model scores, and the SSIM remained over 0.8 all the way through every artistic style. System can successfully create visually beautiful qipao pattern that combine Chinese culture traditional aesthetic elements and a variety, it offers support for fashion design, and for cultural heritage protection.