NVIDIA Edify Powers 3D Generative AI & Fine-Tuning AI Image Creation

NVIDIA recently showcased new 3D generation capabilities in its multimodal generative AI model Edify. These capabilities are already used in new generative AI features from partners Shutterstock and Getty Images and will soon power similar functions in Adobe Firefly, among others. 

NVIDIA Edify applications include 3D modeling for set dressing and prototyping and greater creative control of AI image generation. 

By loading the video, you agree to Vimeo's privacy policy.
Learn more

Load video

Always unblock Vimeo

NVIDIA Edify Brings 3D Asset Generation to Shutterstock Service

Yesterday, at the NVIDIA GTC event, which is taking place this week in California, the company presented the latest capability in its multimodal generative AI model, NVIDIA Edify. The model is now capable of 3D asset generation—one of the frontiers visual AI generation will conquer this year. 

Furthermore, it was disclosed that stock media licensing company Shutterstock is already giving early access via API to a new Edify-powered feature that lets creators generate 3D scenes and objects from text prompts or reference images. The new functionality is part of Shutterstock’s custom AI generative platform, based on NVIDIA foundation models and trained solely with Shutterstock content, deeming it legally safer. 

This tool intends to help 3D modelers with set dressing and scene planning so that they can focus on the main character and object design. 

Arguably more interesting is that Shutterstock and HP presented a new collaboration where HP 3D printers can bring Shutterstock-made 3D models into physical form, advancing and expanding prototyping possibilities for product developers. 

One renowned brand working with Shutterstock’s 3D generation resources is Mattel, which is employing it to let toy designers create 3D models for their ideas. 

Getty Images Adds Fine-Tuning for AI-Generated Images Through NVIDIA Edify

Another relevant stock media firm exploiting NVIDIA Edify is Getty Images. They have a Generative AI feature built on NVIDIA models, allowing their enterprise customers to feed their own visual data and obtain brand-focused outputs. They also provide generative AI image manipulation features like inpainting (modifying elements within an image) and outpainting (expanding and filling canvas). It’s commercially safer to use as it’s only trained with Getty Images’ data. 

In a few months, they will enable clients to custom fine-tune the Edify model with their own datasets, allowing brands to generate even more tailored imagery. They will also add new features to give users greater control over the generated images. Namely, the chance to use a sketch as a prompt to create an image, to extract a reference image’s composition and depth to apply to a newly generated one, and to segment parts of an image to add, remove, or modify elements in it. 

Notably, some huge names are already using Getty’s AI service to explore new creative horizons in marketing. Marketing firm WPP and Coca-Cola are collaborating to use Getty Images Generative AI to create custom visuals that meet brand style and guidelines, for example. 

More Companies Using NVIDIA Edify 

Another exciting partnership announced now is with Adobe, which is said to be working to bring Edify’s 3D capabilities to Adobe Firefly. The software giant is becoming known for taking longer to release new generative AI products but making a big impact in little time once they do. 

BRIA, a startup providing a platform for responsible AI tech development, also uses NVIDIA foundation models in its service. 

One can only repeat it so often, but these are exciting times to be in the visual creative industry. We get to watch NVIDIA (and other firms) develop technology to fuel practical and fun tools. What do you think of NVIDIA Edify? Let us know!

THE AUTHOR

Ivanna Attie

All About Ivanna

I am an experienced author with expertise in digital communication, stock media, design, and creative tools. I have closely followed and reported on AI developments in this field since its early days. I have gained valuable industry insight through my work with leading digital media professionals since 2014.

• NETWORKING