Adobe has unveiled a new feature update to its generative AI tool, Adobe Firefly, called Structure Reference. This new capability, available in the Text to Image module, allows users to apply the structure of an existing image to newly generated images, providing an unprecedented level of creative control.
The Structure Reference feature reduces the trial and error process of crafting the perfect prompt to match the user's imagination. By using an existing image as a structural reference template, users can generate multiple image variations with the same layout. When combined with Adobe's Style Reference, which applies the style of a reference image to the user's prompt, creators can quickly bring their ideas to life with both structure and style references.
Structure Reference is not revolutionary—other generative AI tools like Magnific, Stable Diffusion and Midjourney have similar functionality. However it does offer interesting new possibilities for Firefly users. For example, users can redesign an entire room by uploading a picture or sketch of an existing room and clicking 'generate'.
They can also transform children's drawings into photorealistic images, adding depth, texture, lighting, and color. Additionally, users can colorize black and white images and line drawings, or create entirely new subjects while maintaining the same structure as the reference image.
All Firefly-generated outputs automatically include Content Credentials, promoting transparency in the digital ecosystem. Additionally, Adobe differentiates it's offering by training the Firefly models only on licensed content, such as Adobe Stock, and public domain content where copyright has expired. This ensures they are safe for commercial use.
Structure Reference is now available in the Firefly web application for both free and paid subscribers.