Google Launches Whisk: AI-Powered Tool for Fun and Easy Image Remixing

Google Labs is testing a new tool called Whisk, an experimental image generator designed to make photo remixing fun and accessible. Instead of relying on text prompts, Whisk allows users to prompt with images, giving them the freedom to alter subjects, scenes, and styles.

How Whisk Works

The tool leverages Google’s Imagen 3 image-generation model. Users select three images: one for the subject, one for the scene, and a third for the style. For example, you might combine a photo of yourself (subject) with a futuristic cityscape (scene) and anime art (style). Whisk then generates a detailed caption that guides Imagen 3 in creating your customized image. If you’re feeling chatty, you can also add text prompts like “Subject is riding a flying bike” for more detail.

Manage Expectations

Google notes that Whisk sometimes misses the mark. Since it zeroes in on a few key characteristics, the generated subject might look different—alterations in height, weight, hairstyle, or skin tone are possible. However, you can always go back and adjust the prompts.

While there are plenty of trainable and customizable image generator platforms like Civit.AI and Ideogram that offer advanced features, they often become too complex for casual users. Whisk is a breath of fresh air by keeping creativity simple and accessible for anyone who wants to play around with AI image generation, even if you’re not a tech wizard.

The Whisk experiment is currently open to U.S.-based users at labs.google/whisk. So, if you’re in the States and curious, give it a spin and see how far your imagination can take you.