Google has launched a new generative AI tool that it says will allow you to “remix your subjects, scenes and styles in fun, new ways”.
Initially available in the US only, Whisk combines the smarts of Google’s Gemini AI and Imagen models to put a slightly new spin on AI-assisted image generation.
In the ongoing battle with the likes of OpenAI, Microsoft and Apple for AI domination, the search giant also announced updated versions of its existing tools: Veo 2 for video creation and Imagen 3 for images.
How to Use Whisk
Announced in a blog from Google Labs – the company’s so-called “home for AI experiments” – Whisk will work a little differently to other AI image generators, breaking, the prompt line down to three basic choices: subject, scene and style.
“Instead of generating images with long, detailed text prompts,” it says, “Whisk lets you prompt with images. Simply drag in images, and start creating.”
This just in! View
the top business tech deals for 2025 👨💻
An introductory video for the new tool (see below) shows a simplistic interface for creating images in three steps:
- Subject: Use text to describe the subject of the image, or drag in an already existing photo or image;
- Scene: Choose the scenario where you want your subject to be placed. You also have the option to ‘roll the dice’ and let Whisk pick for you at random;
- Style: Tell Whisk how you want your image to look. Suggestions provided by Google Labs are ‘fantastical’, ‘enamel pin badge’ and ’90s vintage anime’.
“Not a Traditional Image Editor”
Google says that the Gemini model then works “behind the scenes” to write a more detailed prompt based on the three elements you have fed in. Imagen 3 creates the final image based on that longer prompt.
If the image is not as you expected or you want to nuance it further, you’re then given the option to see and edit the longer prompt that Gemini has created from your basic instructions.
“Artists and creatives… have been describing Whisk as a new type of creative tool — not a traditional image editor. 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 Labs
Veo and Imagen Updates
Alongside the whisk ‘experiment’, Google also announced iterative updates to its Veo and Imagen tools.
With the quality of AI video generation improving at speed – as demonstrated by OpenAI’s recently released Sora – Veo 2 seeks to improve detail and realism from its predecessor. Google gives examples of how users can specify the use of cinematic effects like low-angle tracking shots, depth of fields and close ups to achieve better results.
“While video models often “hallucinate” unwanted details — extra fingers or unexpected objects, for example — Veo 2 produces these less frequently, making outputs more realistic.” – Google Labs
While with Imagen 3, the promise is of brighter, better composed images. Google says the tools will now follow prompts more faithfully and “render more diverse art styles with greater accuracy — from photorealism to impressionism, from abstract to anime”.
