Stable Fast 3D Generates High-Quality Images in Half a Second

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More

Not long ago, it was hard to create 3D images. Now, Stability AI has developed a new technology called Stable Fast 3D that can generate 3D images from a single image quickly.

Stable Fast 3D is fast – it takes just half a second to create a 3D image. This is much faster than previous models, which took minutes to achieve similar results.

The technology has many practical uses across industries such as design, architecture, retail, virtual reality, and game development. It’s available through Stability AI’s chatbot and API.

Stable Fast 3D builds on previous work by Stability AI with the TripoSR model. Researchers at the company have detailed the innovative methods used in a research paper.

The system combines several techniques to quickly reconstruct high-quality 3D meshes from single images. It uses an enhanced transformer network to generate triplanes, which are 3D volumetric representations, from input images.

The researchers also detail an approach to material and illumination estimation, as well as the ability of the model to combine multiple elements required for a 3D image into a compact, ready-to-use asset.

Stability AI is known for its Stable Diffusion technology, which generates 2D images. The company has been working on 3D since November 2023 and has released several technologies, including Stable Video 3D, which allows basic camera panning. Just last week, the company announced Stable Video 4D, which adds time to short 3D video generation.

Source: https://venturebeat.com/ai/stability-ai-speeds-up-3d-image-generation-with-stable-fast-3d/