Google's 'Whisk' can generate multiple AI-generated images from image prompts

New Delhi: Google is launching its latest venture in the realm of generative AI with an experimental tool named "Whisk." 

Google's AI research unit, DeepMind, has also introduced Veo 2, a new video generation model that promises to challenge OpenAI's Sora in the field of AI-generated content. 

This new initiative allows users to generate images by using other images as prompts rather than relying on lengthy text descriptions.

"Whisk" is part of Google's Labs program, where the tech giant experiments with cutting-edge AI technologies. Unlike traditional image generators that require users to describe what they want in detail, Whisk innovates by letting users drag and drop images to specify the subject, scene, and style for new images. This intuitive method is aimed at those who prefer a more visual or less verbose interaction with AI tools.

The tool leverages Google's latest image generation model, Imagen 3, in conjunction with the Gemini language model to create images that capture the "essence" of the input rather than exact replicas. Google emphasises that Whisk is designed for "rapid visual exploration" rather than for precise, pixel-perfect edits.

Initial availability is limited to users in the United States who have signed up for the Google Labs program. This launch marks another step in Google's ongoing commitment to expanding the capabilities and accessibility of generative AI, following recent updates like Veo 2 for video generation.

Veo 2 aims to elevate the standards of video production with capabilities that include creating high-definition videos at resolutions up to 4K and extending video lengths beyond one minute, showcasing a significant leap over its predecessor and competitors.

Veo 2's introduction comes with claims of superior realism, with an enhanced understanding of real-world physics, human expressions, and cinematographic techniques. This model is said to excel in generating videos with complex scenes and motion sequences, addressing some of the longstanding challenges in AI video generation. 

Currently, Veo 2 is available for testing through Google's experimental video platform, VideoFX, though with a waitlist due to high demand. Google plans to expand access in the coming weeks and integrate Veo 2's technology into other products, including YouTube Shorts, to empower creators with advanced video generation capabilities.

Google has also improved its Imagen 3 image-generation model, which now generates brighter, better-composed images. It can render more diverse art styles with greater accuracy—from photorealism to impressionism, from abstract to anime.