Multimodal Fusion Algorithm for Generating Transmedia Art Installations: A Review

Abstract

With the rapid development of artificial intelligence and digital art, multimodal fusion algorithms bring new opportunities for the creation of cross-media art installations. This paper summarizes the research on the application of multimodal fusion algorithms in the generation of cross-media art installations, firstly, it introduces the types and characteristics of multimodal data, as well as common multimodal fusion algorithms, including the fusion algorithms based on the feature level, decision level and hybrid level. Then the characteristics and creative needs of cross-media art installations are elaborated, and the key role and special requirements of multimodal fusion in them are analyzed. Then we discuss in detail the application of multimodal fusion algorithms in the generation of different types of cross-media art installations, including visual and auditory, text and other modalities, and multimodal interactive art installations. Finally, we summarize the advantages and limitations of multimodal fusion algorithms in generating cross-media art installations, and look forward to the future development direction. The study shows that multimodal fusion algorithms can provide richer contents and forms for the creation of cross-media art installations, but at the same time, they also face the challenges of data processing and algorithm optimization.