GAN-Based New Media Art Design for the Metaverse Using Artificial Intelligence and the Internet of Things

In the current metaverse scene, new media art creation faces several pain points. These include shallow integration of Artificial Intelligence (AI) and Internet of Things (IoT) technology, a lack of immersive interactive experience for users, and the absence of independent dynamic iterative capability in artistic works. To address these issues, this study constructs a four-stage integrated technical system, “perception-generation-interaction-evolution.” This system integrates multi-source sensor data fusion algorithms with Generative Adversarial Networks (GAN). Together, they form a complete technical framework capable of quantifying both artistic creation and user experience. The comparative experiment, ablation experiment and interpretable analysis are carried out based on the open multimodal dataset. The results show that the quality index of artistic content in this model is improved by about 1.5 points compared with the traditional scheme, the end-to-end interaction delay is controlled within 25ms, and the collaborative correlation coefficient of the system is R=0.837. The comprehensive performance is significantly better than many mainstream existing algorithms, and the creative paradigm of artistic works is transformed from static single generation to dynamic independent evolution. The analysis of experimental data proves that there is a strong positive correlation between the collaboration level of system modules and the user's immersion experience, and verifies the effectiveness of the tripartite collaborative operation mechanism of “technology-art-user”. In this study, the systematic design framework of metaverse new media art adapted to the scene of virtual and real integration is built completely, which can provide theoretical basis and practical paradigm for related creative research in the same field.