Adobe launches new Image 3 AI model in Firefly
An image generated from the text prompt ‘cat with pearl earrings’ by Adobe’s third-generation Image 3 AI model, available in beta in the Text to Image module of its Firefly web app.
Adobe has launched its Image 3 AI model in Firefly, its suite of generative AI tools.
Adobe describes it as providing “massive advancements” in the quality of the images generated, and more user control, including increasing use of reference images to guide output.
The model is available in beta in the Text to Image module of the Firefly web app and in Photoshop 25.9, the latest public beta of the image-editing software.
The third-generation model behind Adobe’s Firefly generative AI tools
Released last year, Firefly is a suite of generative AI art tools available both as a web application and integrated into Adobe Creative Cloud tools like Photoshop.
Since then, Adobe has steadily updated the underlying AI model, rolling out the Image 2 model last October, and now the third-generation Image 3 model.
Better visual quality, fidelity to text prompts, and new Structure Reference system
Adobe describes Image 3 as providing “massive advancements” in the quality and detail of images generated, and in the fidelity with which they adhere to users’ text prompts.
In particular, the company says the new AI model is better at generating realistic images of people, including crowds, and at generating plausible text in illustrations and icons.
There are also more options for customizing AI output, including Structure Reference, available in the Text to Image module of the web app, which makes it possible to generate images that follow the composition of a reference image as well as its visual style, as shown above.
The Imgae 3 model also features a new style engine, making it possible to personalize the images generated by a base text prompt in more subtle ways.
Pricing and system requirements
The Image 3 AI model is available in the web version of Firefly, which supports supports the Chrome, Edge, Firefox and Safari browsers on Windows 10+, macOS 12.0+ and ChromeOS.
Pricing is credit-based, with ‘fast’ generative credits available as part of subscriptions to Adobe’s Creative Cloud tools, and via separate subscriptions. Find more details here.
Read more about the Image 3 AI model in Adobe’s news release
Have your say on this story by following CG Channel on Facebook, Instagram and X (formerly Twitter). As well as being able to comment on stories, followers of our social media accounts can see videos we don’t post on the site itself, including making-ofs for the latest VFX movies, animations, games cinematics and motion graphics projects.