Adobe announced on Tuesday that designers and marketers can now utilize generative AI to modify colors, themes, and fonts in Adobe Illustrator simply by using text prompts. This development is part of Adobe’s ongoing efforts to incorporate various generative AI features into its flagship design products.
Called Generative Recolor, this tool is the latest addition to Adobe’s collection of generative AI tools and models known as Firefly. Its primary objective is to assist users in swiftly refreshing the colors and fonts of their designs. In late May, Adobe introduced a feature called “Generative Fill” in Photoshop, allowing users to zoom out from an image and let generative AI fill in the rest. This feature has been widely used to expand popular memes and even edit selfies to give the appearance that the subject of the image was in a hospital when trying to skip work.
“We strongly believe that generative approaches will play a significant role in editing all types of content in the near future,” stated Alexandru Costin, vice president of generative AI at Adobe, in an interview with Forbes.
Generative Recolor is currently in beta and available in Adobe Illustrator, a software primarily used for creating logos, posters, packaging, website designs, and apparel designs. Prominent brands like Coca-Cola and Land Rover rely on Adobe Illustrator to produce vector graphics, which can be resized without any loss in image quality.
By using descriptive prompts such as “peaceful pastels,” “neon pop,” and “fall foliage,” designers can generate various color palettes for seasonal marketing and advertising. Costin explains that Generative Recolor utilizes Firefly to generate a visual representation of a scene or theme based on the provided text prompts. It then extracts the color palettes from the generated image and applies them to the user’s graphic, recoloring it according to the desired mood or theme.
Since March, Adobe users have already generated nearly 200 million images using Adobe’s text-to-image AI tool. In response to growing interest from consumers, Adobe has recently announced plans to expand its generative AI capabilities to enterprises. However, some artists and contributors to Adobe stock have raised concerns about their work being used to train Adobe’s generative AI model without explicit permission, as well as the lack of transparency regarding the utilization of public domain images.
Firefly, Adobe’s in-house AI model, is trained using a combination of public domain images sourced from Creative Commons, Wikimedia, and Flickr Commons, as well as a vast collection of 300 million images and videos from Adobe Stock. The acquisition of Fotolia, a French stock photo marketplace, for $800 million, contributed to the creation of Adobe Stock.
Despite utilizing public domain and stock images, Adobe is confident in the commercial viability of Firefly. In order to provide reassurance to enterprise customers, Adobe has committed to covering legal expenses in the event of copyright infringement lawsuits.
Alexandru Costin, vice president of generative AI at Adobe, revealed that a dedicated team of hundreds of researchers is actively working on enhancing the quality of Firefly. Their goal is to enable the AI model to generate images with greater detail and higher resolution. Additionally, Adobe is focused on developing models for video and 3D generation, and plans to continually augment its inventory and data catalog by incorporating data from Adobe Stock. Costin emphasizes that AI will not replace creatives, but rather, creatives will coexist and compete with AI-enabled tools.