Whisk by Google (Now Google Flow)

Whisk is an experimental AI image generation tool from Google Labs that takes a unique approach to creative image creation. Rather than generating images from text prompts alone, Whisk lets users drag and drop images as inputs — a subject image, a scene image, and a style image — and uses Google's Imagen 3 model to remix them into entirely new compositions.

The workflow is intuitive and visual. Drop in a photo of your dog as the subject, a mountain landscape as the scene, and a watercolor painting as the style, and Whisk generates your dog painted in watercolor on a mountain. This image-to-image approach is more accessible than prompt engineering because users can express their creative intent through examples rather than words.

Behind the scenes, Whisk uses Gemini to understand the essence of each input image — extracting the key characteristics of the subject, the spatial structure of the scene, and the artistic qualities of the style. It then combines these understandings to guide Imagen 3 in generating the final output. Users can also add text prompts to refine the generation further.

Whisk generates multiple variations for each combination, letting users explore different interpretations. The results can be iterated upon by adjusting inputs, adding text guidance, or selecting favorite variations as starting points for further generation.

The tool is particularly compelling for creative professionals exploring ideas, social media creators making engaging content, designers generating mood boards, and anyone who thinks visually rather than verbally. Whisk is available as a free Google Labs experiment, with output quality powered by Google's latest Imagen 3 model.

Whisk represents Google's broader push into creative AI tools that complement rather than compete with traditional design software, making generative AI accessible to visual thinkers who may struggle with prompt-based workflows.