DreamLLM-3D: Affective Dream Reliving using Large Language Model and 3D Generative AI
arXiv (Cornell University) February 13, 2025 Pinyao Liu, Keon Ju M. Lee, Alexander Steinmaurer et al.
DreamLLM-3D is an immersive art installation that combines a Large Language Model with text-to-3D generative AI to enable dream re-experiencing. It processes voiced dream reports to identify key dream entities, social interactions, and sentiment, then visualizes these as dynamic 3D point clouds where emotional data influences color and soundscapes. The system also proposes an experiential AI-Dreamworker Hybrid paradigm. This approach could potentially facilitate emotionally engaging dream-reliving, enhancing personal insights and creativity.