CAREER: Mesh Renaissance: From AI Foundations to Interactive 3D Creation

Three dimensional models power modern technologies ranging from engineering and manufacturing to robotics and digital content creation, yet creating and modifying these models remains a difficult and time-consuming process even for experts. Even basic operations, such as selecting or editing parts of a shape, require complex tools and significant effort. As a result, scientists, engineers, and creators cannot easily generate or manipulate 3D content. This project seeks to remove these barriers by developing artificial intelligence technologies that fundamentally transform how people interact with 3D shapes. By enabling users to create and edit 3D objects through intuitive interactions such as language, images, and simple geometric selections, the project aims to make 3D modeling accessible to a much broader community. These capabilities will accelerate innovation across the many fields that depend on geometric data. The project develops new computational foundations for artificial intelligence systems that operate directly on meshes, the dominant representation of geometry in graphics, engineering, and simulation. The research will first establish a mesh foundation model that learns general geometric representations capable of supporting tasks such as segmentation, correspondence, and analysis across diverse shapes. Building on these representations, the project will develop multimodal algorithms that allow users to edit and transform meshes using text descriptions, images, point selections, and other intuitive inputs while preserving geometric structure. Finally, the research will investigate the use of large language models as generative agents that can synthesize, refine, and evaluate 3D data, enabling scalable pipelines for generating training data and accelerating progress in 3D artificial intelligence. Together, these advances will provide new tools for understanding and creating geometric data while training students and researchers in emerging areas of artificial intelligence. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2542757 | Program: 01002930DB NSF RESEARCH & RELATED ACTIVIT,01003031DB NSF RESEARCH & RELATED ACTIVIT,01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Rana Hanocka | Institution: University of Chicago, CHICAGO, IL | Award Amount: $360,549 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2542757 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2542757.html