Getty Images is partnering with Nvidia to launch its own generative AI platform trained on its massive catalog of images. Could this offer a possibility for the music industry?
Generative AI by Getty Images is trained on the Getty library of images, including premium content. Anyone using the tool to generate an image to publish commercially is legally protected, Getty says. The company worked with Nvidia to harness the company’s Edify model, available in Nvidia’s generative AI model library Picasso.
The tool seems capable at rendering human figures in realistic poses, so long as you want the soulless PR quality that most Getty images have. Any photos created with the tool are not included in the Getty Images and iStock content libraries, so those wanting to license actual photos from actual photographers don’t have to worry about that. Getty says it will pay creators if it uses their AI-generated image to train current and future versions of the model.
The concept of unleashing machine learning on a controlled copyright collection is interesting — with a very similar concept rumbling in the music industry. That music-focused model is expected to debut within several weeks, with DMN prepping the story now (stay tuned).
Getty says it will “allocate both a pro rata share in respect of every file and a share based on traditional licensing revenue.”
“We’ve listened to customers about the swift growth of generative AI—and have heard both excitement and hesitation—and tried to be intentional around how we developed our own tool,” adds Getty Images Chief Product Officer Grant Farhall in a statement about the new tool. So what kind of limitations does Getty have baked in? The company is well-known as a repository for stock photos of famous people.
The Verge reports that the tool would not allow their reporter to create a photo of Joe Biden standing in front of the White House. It also would not create a cat in the style of Andy Warhol or Jeff Koons.
Both of these prompts require the use of a creator’s art style or persona (Joe Biden), which is a sticking point for generative AI creations. Prompts with the names of actual people are forbidden and won’t be processed, while the AI itself is not trained to recognize art styles for imitation.