RainMirror: An Immersive VR System for Dance Learning and Real-Time Co-dancing with Digital Dancer

This paper presents RainMirror, an immersive virtual reality dance system that bridges dance education and artistic experience through psychologically informed design. Inspired by the iconic rain scene in Singin’ in the Rain (1952), RainMirror integrates motion capture, cinematic aesthetics, and a three-phase learning framework to guide users from passive observation to embodied expression. Grounded in the principles of emotional engagement, progressive learning, and low-interference immersion, our system enables non-expert users to develop kinesthetic confidence and expressive freedom in a non-evaluative virtual environment. Users wear a Meta Quest Pro HMD and interact via handheld controllers, engaging with a virtual world built in Unreal Engine. The system architecture comprises three modules, including motion capture & processing, virtual scene construction & rendering, and interaction & logic control. In addition, user experience design emphasizes the flow of interaction, role transformation from learner to performer, and environmental influence on user’s behavior By blending real-time motion-driven dancer with affective audiovisual design, RainMirror transforms dance into an accessible, emotionally resonant, and self-directed experience, with potential applications in education, cultural heritage, and well-being. Future work will explore multi-user co-presence and adaptive feedback to enhance immersion and inclusivity.