Using artificial intelligence to generate content for augmented reality

The application of AI, particularly through generative models, is rapidly advancing. This thesis examines the use of generative models in creating digital content for AR applications, an area experiencing significant growth supported by advancements in localization systems, including VPS. This thesis focuses on the generation of digital content for AR applications, with emphasis how such content is customized to specific geographic locations. It involves integrating location-based data—such as environmental context, points of interest, and address information—into a text prompt. This prompt then serves as input to the generative model that creates 3D models. These models are then loaded into an AR scene. In addition to analyzing generative AI, the thesis examines crucial localization services for AR applications and explores visual enhancements that could improve appearance of digital content in the real-world environment view. A significant outcome of this research is the design and the development of a system that dynamically generates models based on the surrounding environment. The system architecture comprises three main components: the AR application, a computation server, and a communication server with a database where all models are stored. This structure not only does support the generation of new models but also enables users to access models created by others. The thesis demonstrates the potential of using generative models to enhance AR applications with context-sensitive digital content.