The prestigious stock media licensing agency Getty Images has joined the AI race by launching its own AI image generator today.
Generative AI by Getty Images, built in collaboration with NVIDIA, is a text-to-image generator trained solely with Getty Images’ content and produces more ethical, less biased, and commercially safe AI-generated images with a full royalty-free license.
Read on for all the details!
The new AI image generator by Getty Images was announced today and is developed on the advanced AI generative model Edify, available on NVIDIA Picasso, which –thanks to the tech expert’s computing capabilities– enables it to be used at scale without compromising speed and stability.
Plus, it was trained solely with licensed images from Getty’s extensive library, including premium and exclusive content.
The tool is available on the Getty Images website via a demo request; it will later be included in Getty’s online creative suite and through an API service for at-scale commercial use.
While pricing has not yet been disclosed, we know it will be priced separately from traditional image licensing, and price points will be based on the volume of prompts.
Samples show that it uses a text prompt field and adds presets like photo or illustration style, options for aspect ratio, and color and mood, and the tool will create four variations of an image based on that input. Interestingly, it also includes a prompt builder that guides your hand in writing a good description to synthesize images.
According to a report from The Verge, the model produces very good results in terms of human figures and is said to be better in this aspect than other popular AI image generators out there.
While Getty is later than other stock photo agencies to join the AI frenzy, they have been vocally involved in the discussion around visual generative AI and the need for it to respect artists’ intellectual property and livelihoods and developed responsibly to avoid problematic uses. You may remember Getty Images is suing Stability AI for alleged copyright infringement on their model Stable Diffusion and recently signed an open letter calling for AI copyright regulations, among other important actions.
With this mindframe, it is unsurprising that the firm’s native AI image generator is designed to be as ethical and responsible as possible. For one, the fact it’s been trained with authorized content from the agency only is a big step in the right direction. But there’s more.
The company has purposely excluded terms from the training dataset for their AI-generative model, so it’s unable to produce images of famous people or replicate real-life artists’ styles, for example, to impede users from creating “deepfakes” or copycat content. Those who tried it also mentioned it seems to do better than other tools regarding harmful bias regarding gender and race, another pressing problem of AI-generated media. And the firm stated they will not add your ai-generated images to their licensable libraries, either Getty Images or its daughter microstock company iStock –they will use the synthesized imagery and the prompts to retrain the model further, though.
Finally, the agency stated their model compensates Getty Images’ contributors for using their original work in training the software and synthesizing new images based on them.
Another highlight of this new AI-image generator is that all the images it produces are licensed under Getty Images’ royalty-free license. This means you don’t get copyright ownership of the visuals, but you do get a perpetual and worldwide right to use them, including in commercial projects with few restrictions.
Furthermore, it also means they’re under Getty’s uncapped indemnification regarding copyright claims, as the agency fully backs its legal validity against intellectual property, likeness, and name issues.
All of which makes them much safer commercially than other, more popular online alternatives.
Let us know how it works for you!