Nvidia patents AI method to build 3D scenes from text prompts
Nvidia describes a method where a user types a description of a 3D scene, and the first AI model creates a detailed list of objects and their layout. A second AI model then turns that list into a visual image, which is automatically adjusted to fix object positions before being shown to the user.
In the filing’s own words
In various examples, methods for generating images using generative artificial intelligence models include receiving an input prompt specifying a three-dimensional scene to be generated using one or more generative artificial intelligence models; generating, using a first generative artificial intelligence model, a description of the three-dimensional scene, the description including at least an enumeration of objects to be included in the scene and a description of an arrangement of the objects to be included in the scene; generating, using a second generative artificial intelligence model and the description of the three-dimensional scene, an image representing the three-dimensional scene; iteratively adjusting positioning of one or more objects in the generated image; and outputting the generated image including the adjusted positioning of the one or more objects.
What exactly they claim (claim 1)
1. A processor-implemented method, comprising: obtaining an input prompt specifying a three-dimensional scene to be generated using a plurality of generative artificial intelligence (AI) models; generating a description of the three-dimensional scene using a first generative AI model of the plurality of generative AI models, the description including at least an enumeration of objects and an arrangement of the objects to be included in the scene; generating, using a second generative AI model of the plurality of generative AI models and based at least on the description of the three-dimensional scene, an image representing the three-dimensional scene; adjusting positioning of one or more objects in the generated image; and outputting the generated image including the adjusted positioning of the one or more objects.
Facts
- Company
- Nvidia (NVIDIA Corporation)
- Official title
- TEXT-BASED SCENE GENERATION USING ARTIFICIAL INTELLIGENCE MODELS
- Publication number
- US20260330789A1
- Application number
- 19355871
- Filed
- Oct 10, 2025
- Published
- Oct 22, 2026
- Inventors
- 9
- Claims
- 20
Classified under
G06T 19/20 Physics & computingG06T 17/00 Physics & computingG06T 2219/2016 Physics & computingMore from Nvidia in AI & software
- Nvidia patents AI coach that analyzes your gameplay to suggest strategies Oct 13, 2026
- Nvidia patents AI system that writes and voices character conversations Sep 10, 2026
- Nvidia patents system that uses AI to find bugs in game videos Sep 17, 2026
- Nvidia patents AI method to make digital characters move realistically Sep 8, 2026
- Nvidia patents AI system that builds image generators from text prompts Sep 15, 2026
Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.