Generative Recolor: Enhance Color Fills With Natural Language and Generative AI

Cloudinary continues to lead the way in cutting-edge AI tools for image manipulation, and its latest addition is Generative Recolor. This powerful tool combines the capabilities of natural language processing (NLP) and generative AI to revolutionize color enhancement and customization. Joining Cloudinary’s growing suite of generative AI solutions, including Generative Fill, Generative Remove, and Generative Replace, Generative Recolor is a seamless and efficient way to manipulate colors in images at scale.

Try Generative Recolor and other cutting-edge features at our new Generative AI Playground.

Manually retouching photos for color adjustments often requires expensive software and skills. Additionally, creating color variants for e-commerce products can be costly and time-consuming, especially if it involves reshooting them. Generative Recolor addresses these problems by allowing users to swap colors at scale through API calls, eliminating the need for laborious image updates and making color variant creation more accessible. Easily recolor items within an image using natural language prompts to identify objects, and specify the desired color through a hex code or simple descriptions like “pink” or “light orange.”

Generative Recolor provides fine-tuned targeting of items based on your words, simplifying hyper-specific color changes. The feature is available in beta today, and you can find detailed information in the documentation or try the demo.

Generative Recolor opens up various use cases in different industries:

1. E-commerce variants. Easily create color variants for fashion items, such as clothing, bags, or automotive vehicles, after the photo shoot — without costly reshoots.

2. Product customization. Personalize and customize product variants at scale, ensuring customers can be confident in their purchase.

3. Accessibility. Utilize color replacement to make your content accessible to color-blind readers.

Using Generative Recolor is intuitive. You start by calling the e_gen_recolor effect and using natural language to specify the item you want to recolor, from a general description like “shirt” to a more specific reference like “the shirt worn by the person on the left.” 

Next, provide the to-color setting in a natural language like “blue” or a hex code for precision. You can also specify multiple objects by enclosing them in parentheses and using multiple_true if you want to select more than one object. Below are some examples to help you see how it all works:

For more advanced usage, refer to the detailed documentation.

Generative Recolor employs a two-step process to achieve seamless color changes:

1. Open Set Object Detection. NLP is used to identify specific objects in the image, enabling precise and nuanced object recognition, such as “the jacket on the left” or “the car in the foreground.”

2. Generative AI. Once the objects are detected, generative AI takes over to smoothly recolor the specified items, altering only the relevant pixels.

This dynamic combination of NLP and generative AI delivers a flexible and user-friendly approach to recoloring that will continue to evolve and improve over time.

Generative Recolor represents a significant advancement in image manipulation, enabling users to effortlessly customize and enhance colors in their images using natural language prompts. Whether for e-commerce, product customization, or accessibility considerations, this new feature empowers users to achieve impactful color transformations at scale. As Cloudinary’s suite of generative AI solutions grows, users can look forward to more automated and creative possibilities for their visual content. 

Want to discuss the topic of this article or have questions? Join us in the Cloudinary Community or Cloudinary Community Discord.