Visual Semiotics-Based Emotional Expression and Cross-Cultural Visual Narrative Construction in AI Text-to-Image Generation
—A Case Study of “Loneliness” and “Warmth”

With the rapid advancement of multimodal generative artificial intelligence, Text-to-Image technology is transitioning from simple pixel imitation to advanced semantic and emotional generation. This study selects two contrasting emotional themes, “loneliness” and “warmth”, as experimental samples to compare the performance of international models (Leonardo.Ai) and Chinese domestic counterparts (Doubao, Kling) in visualizing abstract emotions. Grounded in Roland Barthes’ theory of visual semiotics, this paper establishes an evaluation framework encompassing Signifier-Signified Matching Degree and Cultural Adaptability and Innovation to systematically analyze how different models translate abstract emotions into concrete visual symbols (color, lighting, and composition). The findings reveal that domestic large models exhibit superior performance in high-context decoding and localized symbol construction within Chinese cultural settings, though they still face challenges in balancing photorealistic perfection with feature diversity. Finally, this paper discusses the shift in aesthetic intent of human creators and the pathways of human-AI co-creation in the AIGC era, aiming to provide theoretical and empirical references for understanding human-AI emotional interaction and digital visual narration.