Getty Images has partnered with Nvidia to launch Generative AI by Getty Images, a new tool that allows people to create images using Getty’s library of licensed photos.
Getty Images’ generative AI (yes, that’s a hard name to handle) is trained solely on Getty Images’ vast library, including premium content, providing users with full copyright compensation. This means that anyone who uses the tool and posts the commercially created image will be legally protected, Getty promises. Getty worked with Nvidia to use its Edify model, available in Nvidia’s Picasso generative AI model library.
I was able to take a hands-on look at Getty Images’ Generative AI and was able to play around with the tool a bit. I mainly wanted to see how it generates photographs, rather than illustrations, to test how close it can get to a real Getty watermarked image. And the photos look better than expected. Stock photos already have an artificial, soulless quality, and I wasn’t surprised that some of the first images the tool generated also seemed… soulless. This sentiment is not unique to Getty’s generative AI tool; The photos generated by OpenAI’s upcoming DALL-E 3 made me think the same thing.
Getty’s tool worked well when it came to representing human figures with a sense of realism. I asked him to create a photo of a dancer in an arabesque position (standing on one leg with the other raised behind her) on a stage with a slightly blurred background. The photos I got looked more human than when I tried the same message with Stable Diffusion, and the Getty image fooled my friends when I texted it to them. It is clear that Getty’s model was trained not only in illustrated art but also in real photographs. On the other hand, the tool’s illustration mode only gave me 2D pre-designed representations of the same message.
The company said photos created with the tool will not be included in Getty Images and iStock content libraries. Getty will pay creators if they use their AI-generated image to train current and future versions of the model. It will share the revenue generated by the tool, “allocating a prorated portion with respect to each file and a portion based on traditional licensing revenue.”
“We’ve listened to customers about the rapid growth of generative AI, and we’ve heard both enthusiasm and hesitation, and we try to be intentional about how we develop our own tool,” says Getty Images chief product officer Grant Farhall. it’s a statement.
The Getty tool limits the types of images that users can generate. It wouldn’t let me create a photo of Joe Biden in front of the White House or a cat in the style of Andy Warhol or Jeff Koons. Any message with a real person’s name was banned. Asking for a picture of the president of the United States yielded photographs of men and women, some of whom were people of color, in front of the American flag. The company said The edge the model “doesn’t know who Andy Warhol, Joe Biden, or anyone else in the real world is” because he doesn’t want to manipulate or recreate real-life events.
Customers can access Getty Images’ Generative AI through the Getty Images website. The company said the tool will be priced separately from a standard Getty Images subscription, and pricing will be based on immediate volume. However, I would not specify prices.
Getty says users will get perpetual, worldwide, unlimited rights to the image they created. (That said, the technical status of copyright for AI-generated images is still unclear.) Getty said it’s similar to when customers license content from their library, where the company owns the file but licenses its use. You can write your own message or use the message generator as a guide. Users can also integrate the tool into their own workflows through an API. True to form, Getty marks images created through the tool, identifying the photo as AI-generated.
It’s no surprise that Getty is getting into the AI imaging game; After all, he has one of the largest image libraries out there. But the company has fought back against other text-to-image generative AI developers, suing Stability AI for copyright infringement, alleging that its Stable Diffusion image generator used Getty photos without permission.
By building its own generative AI imaging platform, Getty can undercut other companies that want to use its image libraries to train models. Getty is far from the only company creating AI imaging platforms with its permissioned data. Adobe launched its Firefly model, based on its set of licensed images, in its Creative Suite and Creative Cloud service.
The use of copyrighted material to train large language models and text-to-image systems has been a major concern for many in the creative community. Three artists previously sued Stability AI, Midjourney, and art website DeviantArt for using their art without permission to train their models.
Getty said customers can eventually add their own data to train the model and generate images in their brand’s style. This feature and other services will be available later this year.