Adobe Integrates Generative AI Editing Features into Photoshop

As generative AI transforms the tech landscape, Adobe is bringing this innovation to Photoshop with its new Generative Fill feature, set to launch later this year. This Firefly-powered addition offers users “a magical new way to work,” allowing them to add, remove, and extend visual content using natural-language text prompts.

Ashley Still, Adobe’s Senior VP of Digital Media, emphasizes that Generative Fill merges the speed of generative AI with Photoshop's precision, enabling users to manifest their creative visions swiftly. The feature parallels DALL-E 2's inpainting (filling specific areas of an image) and outpainting (extending images beyond their borders). For instance, to create a surreal sky, simply select the area and type a prompt like “surreal sky with strange colors.” If you want to expand a photo's aspect ratio, you can select the surrounding space and prompt it to extend the view.

Adobe’s technology ensures that the generated content aligns with the original scene's perspective, lighting, and style, enabling significant alterations with minimal effort. Users will have access to multiple AI-generated results for each prompt, enhancing creative exploration.

To address ethical concerns surrounding AI, Adobe's model draws exclusively from Adobe Stock images and other copyright-free public domain content. Additionally, through Adobe's Content Credentials initiative, AI-generated images will carry an invisible digital signature, indicating their origin as either human-made or AI-generated. This transparent approach aims to ease artists’ apprehensions about potential plagiarism as generative AI becomes more prevalent.

Generative Fill is now available in the Photoshop desktop beta and will be generally released in the second half of 2023. It is also offered as a module in the Firefly beta on the web.