InstructMesh, yapay zekânın ürettiği 3B nesneleri onarmayı kolaylaştırıyor

Generative AI can produce attractive 3D models, but it often fails to understand how a physical object must function. MIT researchers and their collaborators developed InstructMesh to make such designs easier to correct before fabrication. The system combines a 3D generator with a large language model, allowing users to describe a problem in ordinary language, select the relevant part and review proposed geometric changes. In a study, nearly 80 percent of models recreated by the underlying generator contained structural flaws. Novices using InstructMesh were able to identify and repair those problems about 90 percent of the time, according to an expert reviewer. The team also printed customised mugs, glasses, a knee brace and a small robot enclosure. Sliders gave users precise control over details such as size and extrusion. Future versions may add physics simulations, helping designers predict whether an object will survive a fall or which material best suits its intended use.