Revolutionizing Content Creation: OpenUSD and Generative AI in 3D Product Configurators

The advertising industry faces challenges in creating diverse and engaging content efficiently. Traditional methods are resource-intensive, costly, and potentially environmentally damaging. To overcome these hurdles, organizations are increasingly adopting 3D configurator solutions powered by Universal Scene Description (OpenUSD) and Generative AI. These solutions enable the creation of highly personalized advertising content at scale.

NVIDIA's creative team developed CineBuilder on NVIDIA Omniverse to demonstrate how Generative AI can enhance complex design processes. CineBuilder leverages OpenUSD and Generative AI to simplify the synthesis, generation, and production of high-quality cinematic content, making it ideal for automotive advertising and similar applications. This workflow supports the creation of fully configured 30-second ads, tailored to specific audiences based on lifestyle, preferences, and even real-time environmental conditions.

CineBuilder, built as an extension of Omniverse, facilitates the easy exchange and reconfiguration of digital assets. The OpenUSD-based development process allows teams to segment digital assets into components like models, materials, and environments, each residing on different layers. This modular approach enables seamless swapping of elements within a unified digital environment without disrupting the overall configuration. The highly organized patterns and hierarchies within the OpenUSD workflow ensure that CineBuilder functions as a powerful, fast, and accurate scene-building engine. OpenUSD also allows for quick, non-destructive data manipulation, ensuring that changes do not impact other team members' work.

Integrating Generative AI automates repetitive tasks, enabling the large-scale production and management of customized content. This allows artists to focus on creative endeavors, resulting in faster, more efficient production processes with reduced resource and time requirements. The team integrated Generative AI functionalities throughout the workflow, acting as AI assistants to support tasks such as creating virtual sets, generating supporting assets, and producing lighting scenarios. NVIDIA Edify NIM microservices enable text-to-360 HDRi, allowing artists to create sky domes from simple text prompts, providing limitless lighting possibilities. Text-to-3D capabilities, also powered by Edify, allow artists to efficiently populate scenes with relevant assets. Generative AI also enhances Omniverse Viewport by providing vision and language understanding, enabling voice command control and real-time object detection. Finally, Generative AI is used to customize scripts based on the unique elements of each configuration and target market.

To showcase the power of their configurator solution, NVIDIA's creative team generated 630 unique 30-second automotive ad configurations in 4K resolution at 60 FPS. This amounted to 315 minutes of unique CG-rendered content, equivalent to producing seven feature films at 24 FPS. This demonstrates the scalability and efficiency of the OpenUSD and Generative AI-powered workflow.

Developing a product configurator solution involves several strategic steps, from defining scope and objectives to deployment and integration. Key best practices include: Defining scope and objectives to set clear goals for the configurator's functionality; Establishing a modular process to handle a large number of dynamically exchangeable assets; Integrating with Omniverse to provide a collaborative environment for real-time simulation; Building configurator logic to support user customization and variant generation; Designing an intuitive user interface for ease of use; Rigorously testing and refining the configurator; and Deploying and integrating the configurator into existing marketing workflows.

Workflows using OpenUSD and custom tools enable the efficient creation of unique advertising variations, allowing for easy modification of materials, parts, accessories, and 3D scene environments. Generative AI further enhances these workflows by enabling the creation of limitless lighting scenarios and providing supporting assets from simple text prompts. This combination of technologies offers a powerful solution for creating personalized and engaging advertising content at scale.

To begin developing your own product configurator solutions, refer to the following resources: Product Configurator Reference Architecture, End-to-End Configurator Example Guide, Use Case: 3D Product Configurator, and OpenUSD for Marketing and Advertising, WPP Live.