Similar to Midjourney and DALL-E, Adobe Firefly can create full-fledged illustrations or stylized text from text descriptions. In the future, the authors of the neural network want to add fast image editing, 3D model creation, and more.

What is known about the firefly

The main difference between Firefly and other neural networks is, Adobe did not infringe artists’ copyrights for its education. The developers claim that the system only learns licensed or non-copyrighted content from the internet, not the works of artists. Moreover, Adobe is willing to pay authors to provide their images for AI training.

Over time, Adobe plans to tightly integrate its manufacturer AI tools with its software suite for designers, photographers, and artists such as Photoshop, Illustrator, and Premiere.

The company is also trying to make AI tools easier to use than most of its competitors:

• Instead of entering sequences of descriptors to style an image, Adobe includes built-in options for art styles, lighting, and aspect ratios.
• You can also apply these effects to an already created image, instead of creating a new one every time you want to update the effect.
• The Text Effects tool works similarly, with built-in options for effect size and background color.

Also planned for Photoshop is a potentially controversial feature: Adobe wants to allow authors train the system on their own work so it can help them create content in the artist’s personal style. However, this is subject to abuse: someone can train the system to another artist’s style to clone another artist’s work. That’s why the company is now considering how to deal with it.

The exact launch date of Firefly is still unknown.