Meta and Nvidia Announce AI Tech at Siggraph 2024

Nvidia has announced significant advancements in its Omniverse cloud platform, emphasizing the importance of OpenUSD. The company introduced new microservices aimed at simplifying the use of USD and enhancing applications like USD Search. Notably, Foxconn is leveraging Omniverse and Nvidia's NIMs to train robots at its Guadalajara factory, utilizing a digital twin powered by Nvidia's AI training suite, Isaac. Additionally, WPP, a leading advertising agency, has integrated USD Search and USD Code NIMs into its workflows, employing generative AI for content creation for clients such as Coca-Cola. At the recent Siggraph conference, Meta unveiled the SAM 2 open-source AI model, accompanied by the SA-V dataset designed for training general-purpose segmentation models. This dataset includes a variety of real-world video content, which could be pivotal for companies and countries looking to develop their own AI models. The high costs associated with AI model development and the scarcity of quality datasets have led many to rely on Meta's offerings to enhance their domestic AI capabilities. Meta's recent release of Llama 3.1 further underscores its commitment to producing efficient AI models. During a discussion at Siggraph, Meta CEO Mark Zuckerberg and Nvidia CEO Jensen Huang explored how creators could utilize tailored AIs to generate unique content for their audiences, potentially monetizing these services. Both companies' contributions to the graphics industry were highlighted, showcasing their ongoing influence and leadership in AI development. The commercialization of Meta's SAM 2 model is anticipated, particularly for its capabilities in masking, a sought-after feature in video editing applications. As both Nvidia and Meta continue to innovate, the future of AI in creative industries looks promising.