“Temporarily Forgetting Stress”: Exploring the Multimodal Features and Audience Experience of AI-Generated Psychological Healing Videos

This study examines the content characteristics, audience interaction, and experience feedback of artificial intelligence (AI)-generated psychological healing videos on Douyin, a popular short-video platform in China. It integrates a multimodal analysis framework with stress recovery theory to conduct content analysis of 354 AI-generated healing videos on Douyin and semi-structured interviews with 30 viewers. The study finds that AI-generated videos typically employ hand-drawn animation and warm color tones to establish a safe atmosphere, combined with fantastical visual elements such as biological transformation and material substitution to create stylized, fixed content patterns. Videos that abandon complex narratives and directly construct simple emotional atmospheres through abstract concepts and atmospheric rendering receive significantly higher audience interaction. Users' viewing behaviors have clear functional intentions, often using the videos proactively during times of anxiety or insomnia and relying on platform algorithms to recommend matching content for brief, controllable emotional relief. These findings suggest the emotion-regulation potential of AI-generated psychological healing videos as an accessible form of digital stress-relief media content.