Last Updated: 16 December 2024
Google has announced new updates to its AI tools: Veo 2 and Imagen 3, along with a new experiment called Whisk. These tools aim to simplify video and image creation, offering more advanced and accessible features for users.
Veo 2, the next version of Google’s video model, takes video creation to a new level. It can generate high-quality videos in various styles, offering resolutions up to 4K and clips over two minutes long.
Veo 2 understands physics, human motion, and emotions better than before, creating more realistic and visually stunning outputs.
Users can add details like genres, camera angles, and cinematic effects to customize their videos. For example, you can ask for a close-up shot with a shallow depth of field or a sweeping wide-angle view.
The tool is currently available through Google Labs' VideoFX, with more users expected to access it soon. Google is also prioritizing safety with SynthID watermarking technology, which makes it clear when a video is AI-generated, helping prevent misuse.
DeepMind has worked closely with creators like Donald Glover and The Weeknd to improve Veo 2. Their insights helped make the tool more user-friendly and creatively powerful. Veo 2’s training involved video-description pairs from high-quality datasets, though Google hasn’t disclosed specific data sources.
This approach has sparked debates about copyright and fair use, especially since there’s no way for creators to opt out of these datasets.
Google is addressing these concerns by adding filters to block harmful or explicit content and engaging with the creative community to ensure the tool meets ethical standards.
Imagen 3, the latest image-generation model, offers brighter and more detailed images.
It supports a wide range of styles, like photorealism, anime, and impressionism, and follows user prompts more accurately than before.
Its improved textures and details make it one of the best models available today.
Imagen 3 is part of Google Labs’ ImageFX tool, now available in over 100 countries. A new user interface feature called “chiplets” helps users refine their prompts by suggesting related terms, making it easier to customize their images.
An examples of Imagen 3's rich detail and image quality composition
Google also introduced Whisk, a creative tool that combines Imagen 3 with the Gemini model’s visual capabilities.
With Whisk, you can create or upload images, remix them, and turn them into unique designs like stickers or pins.
Gemini adds detailed descriptions to your images, making the remixing process smoother and more enjoyable.
These tools showcase Google’s commitment to empowering both artists and businesses to explore new creative horizons.
