Adobe, much like every tech company, is getting into AI — and it actually looks pretty cool.
Adobe Firefly is a forthcoming suite of generative AI models with an array of potential uses.
For now, users can sign up for stand-alone beta versions of its image and stylized text generators, but all tools will eventually be integrated into Adobe’s apps.
How they work
The text generator lets users manipulate text; a demo features letters that look like they’re made of melting chocolate and toasted bread.
The image generator, like DALL-E or Midjourney, returns images based on text prompts. But:
- Generated images can be manipulated using Adobe’s other editing tools. Check out this demo from UI/UX designer Jesse Showalter.
- Artists will be able to use AI tools to create variations of their work.
Perhaps most importantly…
… Adobe said its tools won’t scrape the web to steal artists’ styles.
Image generators that do threaten artists’ livelihoods and can lead to copyright issues. Getty Images has banned AI-generated art for that reason.
Adobe said its models are trained on images that are public domain, licensed for training, or in the Adobe Stock library.
“We can generate high quality content and not random brands’ and others’ IP because our model has never seen that brand content or trademark,” Alexandru Costin, VP of generative AI and Sensei at Adobe, told The Verge.
