Nvidia has introduced a new tool enabling developers to generate AI images by initially crafting them in 3D. The product, known as the Nvidia AI Blueprint for 3D-guided generative AI, is now available for download for users with an Nvidia RTX 4080 GPU or better. This innovative tool integrates Blender’s 3D modeling software with Black Forest Lab’s FLUX.1 image generator.
Developers can set up a scene in Blender utilizing various 3D objects such as buildings, trees, animals, and vehicles, which serve as a reference for creating a corresponding 2D image. This approach allows for greater manipulation of the viewing angle and object positioning, thus enhancing control over intricate details when generating images in comparison to relying solely on textual descriptions.
For instance, if one envisions a detailed image of a city—with specific building shapes, tree placements, and car counts—Nvidia’s workflow allows for the manual crafting of a rough outline in Blender. The FLUX.1 image generator can then create images based on this outline, potentially alleviating the frustration of repeatedly adjusting text descriptions to approximate one’s vision. The 3D objects used in Blender do not require intricate details, as they are primarily intended as a layout reference.
Nvidia characterizes its blueprints as “pre-defined, customizable AI workflows” that assist developers in creating generative AI applications. Specifically, this AI Blueprint for 3D-guided generative AI comes with comprehensive documentation, sample assets, and a preconfigured environment to facilitate user interaction with the tool.
While this workflow has similarities to other offerings, such as Adobe’s “Project Concept” demonstrated at its MAX event in October—allowing users to create 3D scenes for image generation—Adobe’s tool remains in an experimental phase and may not be accessible to the general public.
