Luma AI, a developer of 3D models, has launched Dream Machine, an AI model that generates realistic videos from text instructions and images.

Dream Machine is built on a scalable, efficient, multimodal transformer architecture trained directly on videos. It reportedly generates physically accurate, consistent, and action-packed scenes. Dream Machine has three main elements: Text instructions, images, and video generation.

Analyst QuickTake: Dream Machine is the first in Luma’s family of frontier generative models. The company raised USD 43 million in a Series B round earlier this year, with one of the highlighted applications being the development of models capable of generating 3D objects from text descriptions.