Adobe has begun the wide deployment of a new feature based on generative artificial intelligence in its Photoshop graphics editor. It is about creating images based on text descriptions using the Firefly model. This feature was announced in April of this year.
Previously, Photoshop’s Generative Fill feature only allowed you to add, expand, or delete certain parts of an image. Now, you can use this feature to create an image from scratch by entering a description into Photoshop’s updated AI generative tool, just like large language models like Dall-E and Mid Journey. You can then customize the result later.
«It really speeds up the creation time,» said Erin Boyce of Adobe Engadget in April. «The idea of getting something from your mind to the canvas has never been easier».
The feature is based on the Firefly Image 3 model, which recently sparked a backlash from artists against Adobe. Creators were furious with language in Adobe’s recent terms of service, interpreting it to mean that the company was free to use their work to train its generative AI models. However, Adobe later stated that it has a «commitment to creating creator-friendly AI», which means «never training models based on customer content».
In addition to image creation, Adobe introduced the Enhance Detail feature in Generative Fill. For Illustrator, the Generative Shape Fill feature for adding detailed vectors in a unique designer style, Enhanced Text to Pattern (creating customized vector patterns in the style of the artist), and Style Reference are now available. The company has also added a Mockup tool to create «high-quality visual prototypes of art on objects such as product packaging», advanced selection options, and more.
Source: Engadget