Google Whisk

Google Whisk is a Google Labs experiment that flips how AI image generation usually works. You don't write a paragraph describing what you want. You hand it pictures. Drop in one image for the subject, another for the scene, and a third for the style, and Whisk blends them into something new. Think of it as an AI image remix tool rather than a standard text-to-image generator.

Under the hood, the Gemini model writes a detailed caption for each image you upload, then passes those descriptions to Google's image generator, Imagen. That two-step setup matters because it explains the results. Whisk captures the gist of your subject, not a pixel-exact copy. Your subject might come back with a different hairstyle, height or skin tone.

That's the point, according to Google. Whisk is built for fast visual exploration, not precise editing. The company calls it a tool for exploring ideas and working through dozens of options, then downloading the ones you like. It behaves more like an image prompt generator than an editor. If you want to control exactly what's in the frame, you can open the underlying prompts and edit them yourself.

The biggest limit is availability. Whisk started as a US-only experiment, and Google has kept the door open to adjust or retire Labs projects at any time. So who can actually use it? Mostly people in supported regions. If you're outside the US or you need pixel-perfect output, Whisk isn't the tool for you.

Getting Started

1. Go to the Whisk page on Google Labs and sign in with your Google account.
2. Drop an image into the subject slot for the thing you want to feature.
3. Add a scene image for the setting and a style image for the look you're after.
4. Review the generated results and remix the subject, scene or style as needed.
5. Edit the underlying text prompts when you want tighter control, then download the images you like.