Whisk is a new experimental tool from Google Labs designed for rapid visual exploration and creative ideation. Currently available only in the United States through labs.google/whisk, this AI-powered platform represents a departure from traditional image editors by focusing on quick creative exploration rather than pixel-perfect editing. As part of Google's latest AI initiatives alongside Veo 2 and Imagen 3, Whisk offers users a unique approach to generating images by combining visual elements from multiple source images.

Whisk is Google Labs' experimental AI image generation tool that uniquely allows users to generate images using other images as prompts instead of text. It combines Google's Gemini model for image understanding with Imagen 3 for generation, focusing on rapid visual exploration rather than pixel-perfect editing. The tool accepts multiple image inputs for subject, scene, and style, then creates new images that capture the essence of the inputs while allowing users to refine results through editable text prompts. Image-Based Prompting: Users can upload images instead of writing text prompts to generate new images, making the creative process more intuitive and visual Three-Part Input System: Allows separate image inputs for subject, scene, and style, enabling more controlled and diverse creative outputs Editable Text Prompts: Users can view and modify the underlying text prompts generated by Gemini to fine-tune the output images Quick Iteration: Designed for rapid visual exploration and experimentation, allowing users to generate multiple variations quickly

Intuitive visual-based input system Quick and easy creative exploration Flexible editing capabilities with text prompt modification

Currently only available in the US Not designed for pixel-perfect editing May miss specific details from original images

Access Whisk: Go to labs.google/whisk (Note: Currently only available in the US) Sign in: Sign in with your Google account to access the tool Input Images: Upload or select images for three key elements: Subject (what/who you want to create), Scene (the environment/background), and Style (the visual aesthetic you want) Optional: Add Text Details: You can add additional text prompts to further refine what you want to generate Generate Image: Let Whisk process your inputs - it uses Gemini to create captions of your reference images and feeds them to Imagen 3 to generate new images Review & Iterate: Review the generated image. If needed, you can view and edit the underlying prompts to adjust the output Download & Share: Click the download icon to save images you like. You can also share your creations through Google's Discord channel Remix & Explore: Use the remix feature to generate variations and explore different creative possibilities with your uploaded images