Adobe, which recently integrated its Adobe Firefly and Adobe Express tools with Google’s Bard conversational AI service, has announced the introduction of Generative Fill, a new generative AI tool in Photoshop that enables users to easily extend images and add or remove objects using text prompts. This feature, powered by Adobe’s AI image generator Firefly, is now available in beta and is expected to have a full release later this year.
Generative Fill operates within individual layers of a Photoshop image file, providing users with three options to choose from when expanding image borders or generating new objects. While users can leave the prompt blank for the system to autonomously extend the image, providing some direction yields better results. Similar to Photoshop’s Content-Aware Fill feature, Generative Fill offers users more control over the process.
Generative Fill in Adobe Photoshop exhibited impressive capabilities, although it has certain limitations. The tool excelled at filling in backgrounds and empty spaces, but there were instances where generated objects like cars and puddles seemed out of place within the image. Nonetheless, Generative Fill demonstrated its proficiency in replicating features from the original photograph, including the ability to mimic light sources and accurately reflect existing elements in generated water.
The integration of AI image generation tools like Generative Fill into mainstream applications such as Photoshop makes them accessible to a wider audience. Creators can now leverage these tools within familiar software to bring their visions to life more efficiently.
One noteworthy aspect of Firefly, the underlying AI model, is its training data. Adobe ensures that the model is trained only on content for which the company has the right to use, including Adobe Stock images, openly licensed content, and copyright-free material. This transparent approach to training data offers reassurance to creatives and agencies concerned about potential legal complications when utilizing AI tools.
Generative Fill supports Content Credentials, a system that attaches attribution data to images edited or created using AI. By inspecting an image on verify.contentauthenticity.org, viewers can access information regarding the content’s origin and authenticity.
Ashley Still, Senior Vice President of Digital Media at Adobe, highlighted the integration of Firefly into creative workflows, emphasizing how it accelerates ideation, exploration, and production for all users. Generative Fill combines the speed and ease of generative AI with the precision and power of Photoshop, enabling users to bring their imaginations to life more quickly.
While Generative Fill is currently only available in the beta version of Photoshop, users can experiment with the tool by downloading the desktop beta app or accessing it as a module within the Firefly beta app. Adobe, which made this known in March, plans to release Generative Fill on the public Photoshop app in the second half of 2023.
Adobe has been incorporating AI-powered tools into its products, including higher-quality object selections in Photoshop and Firefly integration with Adobe Illustrator for vector-based image recoloring. Additionally, Adobe aims to integrate Firefly with Adobe Express, a cloud-based design platform, though the release date for this integration is yet to be confirmed.
Photoshop’s new generative AI capabilities, powered by Firefly, offer users the ability to extend images beyond their borders, add objects to images, and remove objects with enhanced precision. While currently available in beta (which you can download here), these features demonstrate Adobe’s commitment to empowering users with AI-driven tools for creative expression.
