Nvidia has taken generative AI to a new level by partnering with Shutterstock to create 3D models using AI technology. This collaboration was announced at Siggraph 2024, showcasing the potential of generative AI in the world of 3D modeling.
While there are existing AI tools that can generate 3D models, Shutterstock’s approach stands out as a significant advancement in this field. The generated models will be housed in Shutterstock’s TurboSquid library, providing users with access to a wide range of 3D assets. Users can input text and images as prompts to generate new assets, which can then be edited in various applications and exported in popular file formats.
Nvidia promises that designers can see a preview of their generated asset in just 10 seconds, giving them the option to convert it into a high-quality 3D asset with physically-based rendering materials. Additionally, users can create 360-degree HDRi backgrounds to enhance virtual scenes with dynamic lighting effects.
Similar to Nvidia’s partnership with Getty Images, Shutterstock’s 3D model generator is trained on ethically sourced data, ensuring that the assets are safe for commercial use. However, the exact sources of the dataset remain undisclosed, as is common with generative AI tools.
Get your weekly breakdown of the latest in PC gaming technology
In addition to the Shutterstock partnership, Nvidia has enhanced its generative AI features with Getty Images, boasting faster image creation and improved quality. The service is now capable of generating images at twice the speed with higher precision.
Shutterstock’s 3D feature is currently in beta and accessible only to enterprise customers. There is no confirmation yet on whether the program will be extended to non-commercial users in the future. At Siggraph, the integration of the tool with Blender is showcased, enabling direct generation of 3D assets within the application. HP is also displaying physical prototypes of 3D models produced using Shutterstock’s tool.
