Google has announced that it is rolling out a new generative AI experiment, “Whisk.” This innovative tool allows users to create images using other photos as prompts, rather than relying on traditional text descriptions.
How Whisk works
According to a company blog, Whisk simplifies the image generation process by letting users drag and drop images to define the subject, scene and style of their desired creation.
“Whisk lets you input images for the subject, one for the scene and another image for the style. Then, you can remix them to create something uniquely your own, from a digital plushie to an enamel pin or sticker,” the company added.
The underlying Gemini AI model then analyses these images, extracts key characteristics and generates a detailed text description. This description is then fed into Google’s Imagen 3 image generation model, which then creates a final image.
This approach allows for unique and creative remixing of visual elements, enabling users to generate novel images that capture the essence of their input, rather than simply replicating them.
“This process captures your subject’s essence, not an exact replica. That way, you can easily remix your subjects, scenes and styles in novel ways,” the company added.
Refine the final image
While Whisk’s ability to extract key characteristics offers creative flexibility, it can also lead to unexpected results. The generated images may differ from user expectations in terms of details like height, weight, hairstyle, or skin tone.
To address this, Whisk allows users to view and edit the underlying text prompts, providing greater control over the final output.
“In our early testing with artists and creatives, people have been describing Whisk as a new type of creative tool — not a traditional image editor. We built it for rapid visual exploration, not pixel-perfect edits. It’s about exploring ideas in new and creative ways, allowing you to work through dozens of options and download the ones you love,” Google added.
Whisk is currently available for users in the US to experiment with at labs.google/whisk.
