Adobe has unveiled the latest version of its Firefly AI model, bringing some huge updates to Generative Fill in Photoshop.
The Firefly Image model is the smarts behind Adobe Firefly and the generative AI tools that have rolled out in Photoshop over the last year.
Today, Adobe announced the Firefly Image 3 model and all the upgrades and new features arriving with the beta.
One of the key updates this time around is Reference Image, a feature that allows users to generate images in Photoshop without typing a word. Rather, users can now simply upload a photograph or illustration of an object – like a guitar – and Firefly will generate new images that match the characteristics of the reference, such as the shape, style, and colours.
Adobe has also announced a Generate Similar tool that makes it easier to finetune your AI-generated results when they’re close to what you’re looking for but not quite right. Meanwhile, the Enhance Detail tool improves the sharpness and clarity of images.
The company is also adding a Generate Background feature to Photoshop, allowing users to create backgrounds in one click, which sounds a leap ahead of the app’s already impressive Sky Replacement tool.
You can see some of the new features in action in the video below:
The Firefly Image 3 model is also smarter and better at producing detail this time around.
You can expect images generated with the Image 3 model to include richer detail and better structure, including straighter lines which can be a serious challenge for AI image generators.
The new model can produce a broader range of styles, from illustrations and photographs to icons and vector styles, as well as a more varied range of human emotions. The detail in human faces should be especially apparent in pictures of crowds where detail might previously have been overlooked.
Adobe Firefly Image 3 model enters its beta on April 23.
