OpenAI has officially open sourced Point-E, which is a machine learning system that creates a 3D object given a text prompt, and can produce 3D models in one to two minutes on a single Nvidia V100 GPU, according to a published paper and the code base.
While it doesn’t create 3D objects in the traditional sense, what Point-E does is generate point clouds, or or discrete sets of data points in space that represent a 3D shape, which are easier to synthesize from a computational standpoint, however they do not capture a given object’s finite shape or texture.
While that may seem like a severe limitation of Point-E, the team trained an additional AI system to convert Point-E’s point clouds to meshes, which are collections of vertices, edges and faces that define an object. Meshes are commonly used within the modeling and design in 3D.
Point-E consists of two models: a text-to-image model and an image-to-3D model.
“While our method performs worse on this evaluation than state-of-the-art techniques, it produces samples in a small fraction of the time,” the team wrote. “This could make it more practical for certain applications, or could allow for the discovery of higher-quality 3D object.”