Aiming at the problem that digital narration is not perfect for users' cognitive construction in immersive communication environment, particularly within application scenarios involving smart textiles and virtual fashion content, this paper puts forward a multimodal evolution mechanism of digital narration in immersive communication based on MICU. Firstly, a multimodal evolution path model is constructed and compiled as a systematic framework with four core variables: modal density (M)-interaction frequency (I)-contextual resonance (C)-Depth of Cognitive Construction (U). Then, a controlled user experiment was designed, and the interactive experience and cognitive feedback of 40 users were collected through three typical immersive media platforms: visual short film, multimodal interactive drama and AI-driven narrative generation system, which simulates digital interactive environments of textiles and garments with varying levels of complexity, and an analysis data set was constructed. Finally, multiple linear regression and structural equation are used to establish the path model of evolution mechanism and make empirical verification. The results show that M (β = 0.351, p = 0.001), I (β = 0.279, p = 0.009) and C (β = 0.414, p < 0.001) all have significant positive effects on U, and the contextual resonance is between M and U (indirect effect = 0.154, CI). 0.216]), and there is a significant hierarchical distinction in user cognitive construction. The average U value of the generation-oriented users (0.54) is significantly higher than that of the reproduction-oriented users (0.22). This study provides a quantitative model for optimizing the interaction design of smart textiles and virtual garments, enabling more effective stimulation of users' higher-order cognition and value creation.

AI, immersive communication, digital narrative, user cognitive construction, smart textiles