On Monday, Google unveiled a new experimental artificial intelligence (AI) tool named Whisk, designed to merge images and produce distinctive outputs. This latest offering from the Mountain View tech company joins a series of playful AI applications they have recently launched, such as GenChess, which utilizes the Imagen 3 AI model to create unique chess pieces. With Whisk, Google aims to demonstrate how AI can transform a selection of images into original artwork.
Google’s Whisk Can ‘Remix’ Input Images
In a recent blog entry, Google described its new tool. Currently, Whisk is accessible only in the United States through Google Labs, a platform dedicated to the release of experimental AI-driven tools. As with other offerings from the tech giant, Whisk is in a testing phase, and the company cautions users that its performance may not always align with expectations.
While AI image generators have become commonplace, most usually require either text or a combination of text and images as input. Typically, such models depend on natural language prompts to ascertain what they should create. Whisk, however, differentiates itself by allowing users to input solely images, triggering the model to generate outputs based on those visuals.
To use Whisk, individuals need to submit three images: one representing the subject, one depicting the scene, and another illustrating the style. The AI then processes these visual cues to create a novel image that integrates elements from all three inputs. Users also have the option to upload just two images, focusing on the subject and scene, to generate an image.
Google has clarified that Whisk operates on the Gemini model, which analyzes the uploaded images and formulates a detailed natural language prompt that is subsequently provided to the Imagen 3 model. This prompt is curated to encapsulate the core essence of the images rather than creating a straightforward combination of them.
Given its experimental nature, Whisk may produce outputs that diverge from user expectations. To enhance user control, the tool allows individuals to refine and adjust the generated images. They can examine the underlying prompt created by Gemini and modify it or incorporate additional details to achieve their desired output.
“We built it for rapid visual exploration, not pixel-perfect edits. It’s about exploring ideas in new and creative ways, allowing you to work through dozens of options and download the ones you love,” Google stated.
