How Did Getty Learn to Make Human Figures Feel? The Case for A Getty Image of the White House and the U.S.
“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,” says Getty Images chief product officer Grant Farhall in a statement.
Craig Peters will be at the Code conference in September. You can apply here to attend Code in person or go here for virtual tickets.
Getty’s tool did well at rendering realistic-feeling human figures. I forced it to create a photo of a ballerina in an arabesque position on a stage with a blurry background. The photos I got made me feel more human than when I tried Stable Diffusion and the picture on the screen fooled my friends. It’s clear Getty’s model trained not just on illustrated art but on actual photos. The tool had an illustration mode that only gave me 2D, clip-arty depictions of the same prompt.
The Getty tool does limit what types of images 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 Jeff Koons. Any prompt with the name of an actual person was prohibited. Asking for an image of the president of the United States yielded pictures of both men and women, some of whom were people of color, in front of the US flag. The company told The Verge the model “doesn’t know who Andy Warhol, Joe Biden, or any other real-world person is” because it doesn’t want to manipulate or recreate real-life events.
Proceedings Third International Conference on Artificial Intelligence (Code) and Proceedings of the 11th epsilon Tech Conference (EpiTech2023)
The event that brings together the most influential voices in tech is back for 2023. Code returns with new hosts at The Ritz-Carlton. Apply to attend here or grab a virtual ticket.
The use of copyrighted material to train large language models and text-to-image systems has been a big 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 its models.
Customers can add their own data to the model and create images for their brand. This feature and other services will be available later this year.
Earlier this year, there was a lawsuit brought against Stability by the stock photo service provider, and they claimed that the misuse of more than 12 million photos was a result of training their photo generation tool Stable Diffusion.
Getty’s plunge into the AI photo pool raises questions about the ethics of training an AI model on more than two decades’ worth of photographers’ images and how companies exploring this business model will ultimately pay those photographers.
