US20260080636
2026-03-19
Physics
G06T19/006
The patent application describes a system designed to simplify the creation of augmented reality (AR) applications by utilizing artificial intelligence (AI). This system employs a multi-agent framework comprising various agents, including a lens or AR content creator agent, an AR engineer agent, and a designer agent. These agents work collaboratively to produce AR application designs based on user input. The system features a content management system for storing reusable components and asset generators that offer customized visual elements. A user interface facilitates natural language interactions, allowing users to iteratively refine their AR applications.
The system falls within the technical field of automated AR experience creation using AI, specifically through multi-agent language models and generative asset creation techniques. It focuses on computer-implemented AR application development, aiming to lower the barrier to entry for creators with limited technical expertise. The system integrates large language models, computer vision, and 3D modeling technologies to enable rapid prototyping and generation of interactive AR content.
Augmented reality technology has gained popularity through applications like Snapchat® and its Lenses® feature, among others. However, creating customized AR effects often demands specialized knowledge in programming and design tools, making it inaccessible to many users. Current AR development environments are complex and resource-intensive, requiring significant time and effort to build and deploy customized AR experiences. The system addresses this by providing a simplified and intuitive approach, making AR technology more accessible to a broader audience.
The figures included in the patent illustrate various aspects of the AI-powered AR content creation system. They depict system architecture, component interactions, data flows, and user interfaces that facilitate the automated generation of AR experiences. Examples include diagrams of system architecture, tasks and interactions of language model agents, relationships between asset generators and AR content blocks, and various 3D and 2D visual effects.
The disclosed system automates the creation of AR experiences by leveraging a multi-agent Large Language Model (LLM) architecture. This approach simplifies the process and reduces the complexity and time required to develop engaging AR applications. Traditionally, AR development involves multiple specialized skills and expertise, making it challenging and time-consuming. The system addresses these challenges by automating concept generation, asset creation, and interaction design, thus making AR development more accessible and efficient.