Adobe recently introduced Structure Reference capabilities in its text-to-image generator, Firefly, at the Adobe Summit event. This new feature allows users to input an image as a template for the artificial intelligence (AI) model to follow when creating a new image. Users can adjust the “strength” of the reference, controlling how closely the new image adheres to the original structure.
During a live demo, a user used an image of a person’s profile as a structure reference. Firefly then generated several new images that maintained the layout and composition of the original profile photo. This feature works alongside Style Reference, allowing users to generate images in a specific style by using an image reference.
The combination of Style Reference, Structure Reference, and detailed prompts empowers users to customize the generated images to better suit their needs without the need for repeated tweaks. Adobe’s VP of product marketing, Deepa Subramaniam, described this advancement as a transition from 2D to 3D, enabling a more comprehensive, multimodal approach inclusive of sketches and images.
The ability to use images as guidelines streamlines the creative process, particularly when verbal descriptions may be insufficient to convey the desired outcome accurately. This approach saves time and enhances the user experience by enabling quick visual representations without extensive manual adjustments.
This enhancement is available on both the web and mobile versions of Firefly, allowing users to experiment with Structure Reference for free on the Adobe Firefly website. This innovative tool offers creators a new level of control over the output of AI-generated images, revolutionizing the creative process.
HEALTH
WORLD AFFAIRS
TECHNOLOGY
CREATIVE
