State-of-the-art generative AI models will allow Shutterstock customers to generate 3D assets with simple text prompts for running metaverse simulations, generating digital twins, and creating new 3D context.
Shutterstock is working with NVIDIA to advance the world of 3D by way of AI. NVIDIA’s contributions to the advancements of artificial intelligence, the industrial metaverse, and AI-accelerated computing are transforming the world’s largest industries across a broad range of use cases.
The NVIDIA Omniverse, for example, is a 3D development platform that enables teams to connect their favorite 3D tools, generate synthetic data, and develop applications for a variety of industries.
As a result of this partnership with NVIDIA, text-to-3D generative content will soon be available from Shutterstock for commercial use. NVIDIA Edify-3D models will be trained with images and video from Shutterstock and 3D assets from its subsidiary TurboSquid, the world’s largest marketplace for 3D content.
Once models are trained with Shuterstock’s data, its service will use NVIDIA Picasso, a foundry for custom generative AI models. Users will be able to use the Shutterstock service to convert text prompts into high-fidelity 3D assets based initially on the University Scene Descriptions (USD) file framework.
Thanks to these advancements in AI and 3D, work across several industries will become more efficient, as animators will be able to generate textured and structured 3D models for any project with just a few prompts. This will soon help teams build digital 3D models and run simulations faster in order to predict outcomes.
For example, grocery store chains will be able to more quickly mock up digital check-out lanes and test them to find efficiencies. City planners could rapidly simulate traffic patterns and public transit routes, in order to better prepare for special events like concerts, ball games, or parades. Shutterstock’s massive data set of ethically sourced content, along with NVIDIA’s advanced technology for 3D and industrial metaverse simulations, make it all possible.
Here, we talk with Greg Estes, Vice President of Developer Programs at NVIDIA, and Dade Orgeron, the Vice President of 3D Innovation at Shutterstock, about why this partnership is not only exciting—it’s groundbreaking for city planning, manufacturing, retail, and so many other industries.
How does this collaboration power the future of 3D?
Similarly to text-to-image generative AI, text-to-3D AI creates brand new content based on what is typed into a text field. Shutterstock’s partnership with NVIDIA brings the companies’ technologies together, while Shutterstock’s massive resources of metadata and 3D content are combined with NVIDIA Edify foundation models in order to expand digital creativity and problem solving.
“Creating 3D models is complex, time consuming, and requires both technical knowledge and creative skill,” said Estes. “With increasing demand for virtual content in product design, simulation, and digital twins, Shutterstock and NVIDIA’s AI research teams have been looking at ways to make that content faster and easier, democratizing the 3D model creation process.”
The capabilities brought forth by this technology have huge potential for creative thinkers, technical problem solvers, and specialists across nearly limitless industries.
“NVIDIA AI and 3D help run digital twins faster to determine how to boost safety in the real world, and generative AI capabilities can enable designers to bring more physical accuracy to their work. 3D allows them to run thousands of simulations in one day. Normally, it would take years and years to discover those things through trial and error,” said Orgeron.
Advancements from this partnership will assist professionals across a multitude of industries:
- 3D Visual Effects and Game Design: Generate assets and features quickly to iterate on more creative ideas.
- Architecture and Interior Design: Create floor plans, explore architectural styles faster, and design collaboratively with clients.
- Fashion and Product Design: Inspire unique design concepts and explore how they move in simulation.
- Graphic Design and Marketing: Create innovative new elements and reusable motifs easily to supercharge social media content production.
- Manufacturing: Accelerate design of industrial parts and explore structures and solutions to build safer factories and warehouses.
In addition, these new NVIDIA-powered generative AI capabilities will allow professionals to create content for commercial use. Text-to-3D features will be offered on Shutterstock’s Creative Flow tool, Turbosquid, and within the NVIDIA Omniverse platform for building and operating 3D industrial metaverse applications.
Building ethical AI for business use cases
One of the most exciting aspects of this partnership is that the content generated will be commercially available for brands, businesses, and individuals to use. This is due to Shutterstock’s commitment to building ethically sound AI. Because AI models are trained on content from the Shutterstock library, contributors who create this content are compensated through our Contributor Fund. This ethical approach is a shared focus for Shutterstock and NVIDIA.
“NVIDIA shares many values with Shutterstock, including a commitment to innovation, support of artists’ work, and responsible AI development,” said Estes.
“From a technical perspective, Shutterstock offers a vast library of 3D content and metadata, and rich experience in licensing and distribution,” said Estes. “Data is the fuel of AI, and Shutterstock’s content is an excellent complement to the work of NVIDIA’s AI researchers.
Orgeron agrees, stating that Shutterstock has successfully stepped up to the challenge of providing high quality, ethically sourced data for NVIDIA applications.
“AI-based simulation is one of those fields where a massive amount of content is needed in order to do these types of simulations,” said Orgeron. “Shutterstock meets that demand.”
What does the future of text-to-3D generative AI hold?
Generative AI is transformative. The advancements resulting from this partnership will make it possible for enterprises and creators to save enormous amounts of time and effort when making 3D assets. Just as text-to-image AI will speed up processes and workflows, text-to-3D has great potential to assist both technical and creative teams.
“Training a custom Shutterstock (3D) model with the NVIDIA Picasso generative AI cloud services will enable developers to build tools that can automate much of the tedious work for artists, freeing them to spend more time exploring new concepts and refining their ideas,” said Estes.
Complementing human work is one of generative AI’s biggest benefits, no matter what type of content it creates. By freeing up time for technical artists and creative thinkers, these professionals will now have more space to work on other important, higher-value tasks.
Reducing time spent on 3D model creation will allow teams across city planning, manufacturing, entertainment, retail, and more to benefit from generative AI’s speed and high-fidelity outputs. The future certainly looks promising, thanks to these advancements within the generative AI and 3D space.
Discover even more insights into content creation for today: