Products

Whisk, Google's novelty that mixes images with artificial intelligence

Experts at Google Labs propose a more intuitive and faster alternative to text prompts for generating images with artificial intelligence. For now only in the United States

19 December 2024

2' min read

Google Labs experts propose a more intuitive and faster alternative to text prompts for generating images with artificial intelligence

Artificial intelligence is capable of performing numerous tasks, even very complex ones, returning results of ever better quality. Everything, however, is usually done by typing long text commands, resulting, in some cases, in a rather long and cumbersome operation: however, a novelty proposed by the experts of Google Labs could be the right alternative. The announcement was made by Tommaso Iljic, Director of Product Management at Google Labs, in a post published on the official Google blog: Whisk, in fact, is the latest experiment in generative artificial intelligence implemented by the team, which proposes an alternative way of creating images with generative artificial intelligence.

Launched in the United States a few days ago, this new feature does not require the typing of long and complex text commands, but allows you to launch prompts by providing images: simply drag and drop them into the dedicated boxes to begin creation. Whisk, in fact, allows you to insert an image to define the main subject, another for the scene or context and a last one to establish the style.

"Behind the scenes, the Gemini model automatically writes a detailed caption of your images. It then inserts those descriptions into Google's latest image generation model, Imagen 3. This process captures the essence of the subject, not an exact replica. In this way, you can easily remix your subjects, scenes and styles in new ways,' Iljic explains in the blogpost and continues, 'In our early tests with artists and creatives, many described Whisk as a new kind of creative tool, not a traditional image editor. We created it for quick visual exploration, not for pixel perfect editing. It's about exploring ideas in new and creative ways, allowing you to work through dozens of options and download the ones you love".

The tool is therefore able to analyse the images and extract certain fundamental characteristics, so it may propose images that differ from expectations: for example, the generated subject may have somatic features that differ from the initial idea, but these can be refined or modified with text commands.

Whisk is currently available, via this page, only to those in the US. There are no details about an Italian debut, however, you can join the waiting list by providing a valid Gmail address to be notified as soon as there are updates.