Google Labs Tests Whisk Tool for Blending Three Images Into One
Google's experimental division introduces Whisk, an image generator that uses photos instead of text prompts to remix a subject, scene, and style. The tool currently runs exclusively for users in the United States.
Google Labs launches a new experimental image generator called Whisk that allows users to prompt with images instead of text. The tool uses Google's Imagen 3 model to blend three distinct images into a single creation by assigning each picture a specific role as the subject, the scene, or the style.
Users simply upload their chosen images, and Whisk automatically generates a detailed text caption to guide the final output. People can also add traditional text prompts to further refine the creation, adding specific actions or details like a subject riding a flying bike to ensure the final image matches their vision.
Because the tool only extracts a few key characteristics from each reference image, Google warns that the final results may vary in attributes like height, weight, or hairstyle. Users maintain control by viewing and editing the underlying text prompts at any time, though the experiment remains exclusively available to individuals located in the United States.