Adobe is launching its own AI-powered image generator dubbed Firefly. The company announced a family of creative generative AI models called Adobe Firefly and released the first two tools that take advantage of them.
One of these tools works like DALL-E or Midjourney, allowing users to enter text prompts and get full-fledged illustrations in response. The other generates stylized text similar to an AI-powered WordArt.
Firefly is currently available in a browser-based beta version, but over time, Adobe plans to closely integrate generative artificial intelligence tools with its suite of creative apps, such as Photoshop, Illustrator, and Premiere.
Adobe is trying to make its tools easier to use than those of most of its competitors. Instead of entering a string of descriptors for styling, Adobe offers built-in options for art styles, lighting, and aspect ratio. You'll also be able to apply selected effects to an already-generated image, instead of having to create a new one every time you want to update an effect.
Adobe has also become one of the few companies willing to discuss what data its models are trained on. A key difference between Firefly and other similar tools is that Adobe has not violated artists' copyrights for its training.
Adobe emphasizes that all the data coming into its model is either non-copyrighted, licensed for training AI, or in the library of Adobe Stock.
Moreover, the company plans to pay artists who provide their images for AI training. That won't happen right away, but Adobe plans to develop some "compensation strategy" before the system comes out of beta.