Adobe releases AI art generator Firefly to take on Midjourney and DALL-E

Firefly is Adobe's first attempt at using AI to make images. It is made to work better with drop-down menus and buttons.
Christopher McFadden
Adobe has officially released Firefly
Adobe has officially released Firefly

Adobe has released Firefly, a "family of creative AI models" that can be used to make images. The first two tools that Adobe has released are similar to DALL-E or Midjourney, which allow users to type in a prompt and have an image created in return.

The other tool generates stylized text, like an AI-powered WordArt. Adobe calls Firefly a beta at launch, which will only be available through a website. But Adobe plans to add generative AI tools to its creative apps like Photoshop, Illustrator, and Premiere at some point in the future.

“We’re not afraid of change, and we’re embracing this change,” says Alexandru Costin, VP of generative AI and Sensei at Adobe. “We’re bringing these capabilities right into [our] products, so [customers] don’t need to know if it’s generative or not.”

Unlike other generative AI tools that require users to enter specific keywords to get good results, Firefly is designed to give consistent results by using drop-down menus and buttons to determine the overall look and feel of the generated image.

This method makes Firefly easier to use because users can change the style of an image without having to recreate it. Firefly can also make text effects, combining generative AI with other art and design.

Adobe plans to pay artists who contribute training data, and it has designed Firefly to generate diverse images of people of different ages, genders, and ethnicities to avoid bias issues. To date, Adobe has trained Firefly on hundreds of millions of Adobe Stock images, openly licensed content, and public domain content to avoid copyright problems.

Firefly’s first two tools will be available in a public beta. Adobe plans to integrate generative tools into its various apps and services, such as AI-generated outpainting in Photoshop, vector variations in Illustrator, and image restyling in Premiere.

Adobe aims to be part of the transformational power of AI and wants a future where creators can train their AI models on their work and where generative AIs seamlessly integrate with its full range of products. However, Adobe intends to employ AI responsibly and address two significant issues with generative AI: copyright concerns and biases.

It also plans to develop a compensation strategy for artists who contribute to its training data and is working on a “Do Not Train” system that allows artists to block AI (i.e., opt-out) from training on their work.

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board