Stability AI launches Stable Fast 3D: 1200x faster 3D image generation from single images

Stability AI has unveiled Stable Fast 3D, a new generative AI technology that can create 3D images from a single input image in just half a second. This represents a massive leap in efficiency compared to its predecessor, Stable Video 3D (SV3D), which took up to 10 minutes for similar tasks, making Stable Fast 3D around 1200 times faster.

The new model is poised to revolutionize industries such as design, architecture, retail, virtual reality, and game development. It is accessible via Stability AI’s Stable Assistant chatbot and API, and is available under a community license on Hugging Face.

Stable Fast 3D is built on the TripoSR model and developed in collaboration with 3D modeling vendor Trip AI. It employs an enhanced transformer network to generate high-resolution triplanes and features a material estimation network that predicts global metallic and roughness values using a novel probabilistic method. This release follows the company's recent launch of Stable Video 4D, capable of generating multiple novel-view videos from a single-view video of an object.