Google Launches Whisk, an AI Tool for Remixing Images Using Visual Prompts

The experimental tool allows users to combine subject, scene, and style images to generate unique creations, showcasing rapid visual exploration.

Whisk is an experimental AI image generator from Google Labs that uses images, rather than text, as prompts to create new visuals.
Users can upload images for subject, scene, and style, which are analyzed by Google's Gemini AI to generate detailed captions fed into the Imagen 3 model for image creation.
The tool emphasizes creative exploration rather than precise edits, allowing users to refine and edit generated results through text prompts or additional inputs.
Whisk is currently available in the United States through Google Labs and is designed for quick inspiration, generating outputs like stickers, enamel pins, and plushie-style images.
Google positions Whisk as a novel creative tool, with user feedback helping shape its development, but acknowledges that outputs may differ from expectations due to its experimental nature.