Adobe has unveiled an ingenious artificial intelligence (AI) feature, known as Generative Recolor, for its flagship design product, Adobe Illustrator. The ambitious feature incorporates generative AI to facilitate seamless alterations of colors, themes, and fonts of graphics via text prompts.
The announcement comes as part of Adobe's ongoing mission to enhance its design products by integrating an array of generative AI features under its Firefly suite. Generative Recolor is the newest addition to this collection, following the release of the highly popular Generative Fill feature in Photoshop just a few months ago.
According to Alexandru Costin, Vice President of Generative AI at Adobe, the company envisions that generative editing will soon become a standard approach across all content. Generative Recolor is currently in its beta phase, available in Adobe Illustrator – the respected software application for creating vector graphics, including logos, posters, packaging, websites, and apparel designs. Well-known brands, such as Coca Cola and Land Rover, are already using Adobe Illustrator to create images that can be resized without any loss of quality.
Generative Recolor relies on Firefly, Adobe's generative AI system, to create variations of color palettes and revitalize designs. Users simply provide text prompts with descriptive phrases like peaceful pastels, neon pop, or fall foliage, and the AI analyzes the mood or theme to generate a rendering of a scene or theme. Then it extracts the color palettes from that image and applies it to the user's graphic.
Since its launch in March, Adobe's AI-powered text-to-image tool has been used to create almost 280 million images. In response to the growing interest from consumers, the company recently announced its plans to extend generative AI capabilities to enterprises. However, concerns have been raised by artists and Adobe stock contributors about their work being used to train Adobe's generative AI model without explicit permission and a perceived lack of transparency in public domain image usage.
Firefly, Adobe's AI model, has been trained on a combination of public domain images sourced from Creative Commons, Wikimedia, Flickr Commons, and an extensive library of 300 million images and videos from Adobe Stock. Despite the issues raised, Adobe remains confident in Firefly's suitability for commercial applications and has even offered to compensate enterprise customers for any legal expenses in the event of copyright infringement lawsuits.
Hundreds of Adobe researchers are currently focused on enhancing Firefly's quality, generating images with greater detail and resolution. The development of models for video and 3D generation is also underway, alongside the continuous addition of data from Adobe Stock.
Ultimately, Costin believes that creatives will not be replaced by AI but will instead find themselves in competition with others utilizing AI. And with the growth of no-code platforms like AppMaster, creatives have powerful tools at their disposal that not only simplify tasks but also enhance collaboration and productivity.