Shape Repair

Shape repair research focuses on automatically reconstructing broken 3D objects, aiming to generate accurate and printable restoration shapes. Current approaches leverage deep learning, employing implicit shape representations like occupancy functions, signed distance functions, and normal fields within neural network architectures to infer complete shapes from fractured inputs. This field is driven by the need for robust and scalable methods, fueled by the creation of large-scale datasets of real-world fractured objects, moving beyond reliance on synthetic data. The resulting advancements have significant implications for digital fabrication, cultural heritage preservation, and various manufacturing applications.