Cross-cultural communication is an effective way to enhance national soft power and international competitiveness, but the cultural differences between different countries hinder cross-cultural communication to a certain extent, and the efficiency of communication needs to be further improved. This study is based on the generative adversarial network algorithm in computer vision technology, which is applied to the method of creating cross-cultural digital media content. Taking the digital media content creation of traditional Dong tattoos as an example, the tattoo content created using this paper’s method is compared to the tattoo content created using Diffusion’s stable diffusion model. According to the different cultural and professional backgrounds of the interviewees, the tattoos were analyzed in terms of three evaluation dimensions, namely, the appearance characteristics, the inner performance and the practical value of the tattoos. The interviewees unanimously agreed that the digital media content of Dong traditional tattoos produced by this paper is superior to the comparative model and has more cultural identity. Finally, this paper proposes a specific strategy for using digital media to promote cross-cultural communication.