Architecture

The diagram below illustrates the USD Code API architecture, providing a visual representation of the high-level architecture and design used in USD Code API.

This diagram shows how the Large Language Model (LLM), Embedding models and Retrieval-Augmented Generation (RAG), Mixture of Experts and code evaluator components work together to generate Python USD Code.

USD Code Helper Functions

Among the experts shown in this diagram, USD Code Helper Functions provide high-level abstractions leveraging the Python-USD API, enabling you to efficiently manage complex tasks such as creating, modifying, and querying USD scene graph structures. These abstractions facilitate workflows by simplifying the management of attributes, relationships, layers, and automating intricate operations across the USD environment. Through USD Code Helper Functions, you can integrate higher-level logic into your workflows, enabling USD applications to adapt to evolving needs and fostering innovation in digital content pipelines.