A better way to turn 2D designs into 3D models for rapid prototyping

Researchers at MIT have created a new approach to a real engineering challenge: getting AI to reliably convert two-dimensional design sketches into accurate three-dimensional computer models. As MIT News reports, the bottleneck has always been training data—existing AI models simply haven't seen enough high-quality examples. Rather than spending months labeling new data by hand, the team developed a framework called GIFT that lets the AI improve itself by learning from its own mistakes. When the model generates code that's almost correct, GIFT saves it along with the correct version and feeds both back into the model's learning loop. The result: better performance using only about twenty percent of the computing power of competing approaches. For engineers designing aircraft parts, automobiles, and appliances, this could make prototyping faster and far less expensive.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton