Google launches Whisk for creative AI image remixing

Google has introduced Whisk, a new generative AI tool designed to make image creation more interactive and creative. Unlike traditional AI tools that rely heavily on lengthy text prompts, Whisk allows users to start with images. Simply drag and drop images into the interface, and Whisk lets you generate unique creations by combining these visuals in novel ways.

Whisk enables users to input up to three images: one for the subject, one for the scene, and another for the style. This approach gives users the flexibility to create anything from digital plushies to enamel pins or stickers. Some early examples from Whisk include a fantastical fish, a whimsical walrus, and even a glazed doughnut reimagined as an enamel pin.

At the core of Whisk is Google’s Gemini model, which analyzes uploaded images and generates detailed captions. These captions are processed through Google’s Imagen 3 model, ensuring the essence of your images is captured without directly replicating them. This means you can experiment with endless variations of your chosen subject, style, and scene. However, the results may sometimes differ from expectations. For instance, Whisk might alter characteristics like height, weight, or hairstyle. To address this, Google lets users edit the prompts generated by the AI, providing greater control over the outcome.

Whisk is designed as a tool for rapid visual exploration rather than a traditional image editor. It’s aimed at artists, designers, and other creatives looking to quickly iterate on ideas and concepts. In early testing, users have praised Whisk for its ability to help them explore dozens of creative options efficiently.

Whisk is now available for users in the US. To try it out, visit labs.google/whisk.