From Point Clouds to Digital Narratives: A Generative AI Framework for Immersive Media Reconstruction and Cultural Storytelling of Classical Garden Heritage

Classical gardens, as tangible embodiments of Eastern philosophy and aesthetics, face dual challenges of physical deterioration and cultural disconnection in preservation and transmission. This study proposes a generative AI framework designed to systematically transform high-precision point cloud data into immersive digital narratives. By integrating multimodal data through semantic point cloud segmentation, generative geometry and texture completion, and context-aware narrative generation powered by large language models, the framework constructs culturally rich interactive digital twin scenarios. Using Jiangnan classical gardens as a case study, the experiment demonstrates the method's significant advantages in geometric reconstruction accuracy, semantic scene understanding, and cultural narrative consistency. This approach effectively enhances user immersion and cultural cognition, providing an innovative pathway for in-depth digital interpretation and dissemination of cultural heritage.