Last month, Google launched its latest experiment in generative AI — Whisk. This tool breaks away from the traditional text-to-image generation model, allowing users to create directly through images. Simply drag in an image to get started. Whisk extracts key features from your image, so the generated results may differ from expectations, such as changes in height, body type, hairstyle, or skin tone of the subject. We understand the importance of these details for projects, so Whisk allows you to view and edit underlying prompts at any time to better suit your needs.
The core function of Whisk lies in:
Providing three types of image input methods: subject, scene, and style; Creating personalized digital works by remixing, such as digital dolls, enamel badges, or stickers.
Usage examples
1. Generated image of a fantasy walrus
On a bright yellow background, this image showcases a fun application of Whisk: a fantastical walrus wearing a strawberry-patterned swimsuit and a floral crown. The inputs for generating this unique image included: a picture of a walrus (subject), a field of flowers (scene), and a cartoon-style cloud pattern (style).
2. Enamel badge generation of a frosted donut
This image demonstrates Whisk's ability to generate enamel badges from images: a colorful, sugar-dusted donut. The input on the left includes a photo of a real donut and a metal-cut silhouette, combining elements into the final enamel badge design.
3. Generated image of a fantasy unicorn cat
This image presents a purple-shining unicorn cat with a striking green eye. The cat rests on a large lily pad floating in water, with more lilies scattered in the background. The inputs for generating this fantastical scene included: a shining unicorn cat (subject), a natural scene with lilies (scene), and a landscape painting with trees and clouds (style).
4. Generated image of a fantasy fish
With a bright yellow background, this image displays a dreamlike fish with a city built on its back. The inputs for generating this fantastical scene included: a submarine (subject), a floating island (scene), and a landscape picture (style).
Whisk's technology
. Through this method, Whisk captures the essence of the image rather than simply copying it. Users can easily remix subjects, scenes, and styles based on this foundation, creating entirely new visual experiences.
Try it out
https://labs.google/whisk
