The creative software company Adobe Inc. today announced some major updates to its Firefly image generation tool, which is powered by advanced generative artificial intelligence models.

At the Adobe MAX conference in Los Angeles today, Adobe announced a new, more powerful Firefly Image 2 mode that’s said to be able to generate higher-quality images than the original Firefly model. In addition, there’s a new Firefly Vector model that can generate the vector graphics used by designers, and a Firefly Design model that can produce templates and other features within Adobe Creative Cloud.

Adobe announced Firefly in March, saying it will eventually become a “family of generative AI models for creative expression.” It enables users to generate art and other images simply by describing them with natural language text.

Since that announcement, Firefly has been incorporated into multiple Creative Cloud apps. For instance, Photoshop has its Generative Fill and Expand tool that allows users to use text prompts to expand or change images. Firefly’s generative AI capabilities are also available for video in Adobe’s video creation app Premiere Pro, which is used by broadcasters and video producers to turn raw footage into production-ready video.

With Firefly Image 2, the tool can now generate more photorealistic images than before, rendering high-frequency details such as facial features, foliage, hair, hands, skin and texture when creating images of humans, for example. The images Firefly Image 2 generates also feature more vivid colors and contrast, and are higher-resolution, Adobe said.

Users can also access some new AI-powered editing tools to customize the images Firefly generates. For instance, users can adjust the depth of field and add motion blur.

There’s also a new Prompt Guidance feature that helps users to improve their text-based prompts and describe what kind of image they want to create more accurately. Meanwhile, the Generative Match feature makes it possible to select from a preselected list of images, or an uploaded image, to replicate the style of those images when creating a new one.

Adobe said Firefly Image 2 is available to access now in the web-based Firefly beta and will soon be integrated with its Creative Cloud applications. So Photoshop and Premiere Pro cannot access it just yet.

Alongside Firefly Image 2, there’s a new Firefly Vector model for Adobe Illustrator that’s available now in Adobe Illustrator beta. Adobe says this is the world’s first generative AI model for vector graphics, and provides a way for designers to generate editable vector images using text prompts. Adobe says the graphic generated will be split automatically into “logical groups and layers.” Vector graphics are preferred by designers, because unlike traditional image formats such as JPG and PNG, they can be scaled to any size without reducing the quality of the image.

As with the original Firefly model, Firefly Vector was trained on licensed content from Adobe Stock and public domain materials that aren’t copyrighted, meaning that the images it creates are safe for commercial use.

Finally, the new Firefly Design model is available now, powering the new text-to-template beta feature within Adobe Express. It’s designed to generate customizable templates for print, online advertising, social media posts, videos and more. For example, a user could input text saying they need a flyer for a real-estate open house. The model then generates different design template options that fit the theme.

Adobe said Firefly users have now created more than 3 billion images with the tool since it launched about six months ago. Its popularity has been increasing rapidly too, with 1 billion of those images generated in the last month alone. The tool has attracted a lot of new users, with more than 90% of Firefly users new signups, Adobe said.

Images: Adobe