Google Introduces New Veo 2 and Imagen 3 AI Models for Video and Image

Google has just unveiled Veo 2 and Imagen 3, its latest advanced AI models. Veo 2 handles video generation, while Imagen 3 is for image generation.

These are available via VideoFX, ImageFX, and the new Google Labs experiment, Whisk.

Google – Veo 2 and Imagen 3

Veo 2

Veo 2 creates high-quality videos in a wide range of subjects and styles. It is said to understand real-world physics and human movement to help generate lifelike videos.

Moreover, the model is said to be capable of producing effects like a low-angle tracking shot or a close-up. It offers videos of up to 4K and up to a couple of minutes.

Imagen 3

The latest version of Google’s image generation model, Imagen 3, on the other hand, now produces brighter and more vibrant images with better color balance. It can also generate a variety of art styles that goes from photorealism to abstract art.

Moreover, it has better understanding of natural language prompts, increased detail and precision, and better text rendering.

To help reduce the risk of misinformation, videos made by Veo 2 and photos from Imagen 3 feature an invisible SynthID watermark to mark them as AI-generation.

Veo 2 is available via Google Labs’ VideoFX but there are plans to expand it to YouTube Shorts and other products in the coming year. Imagen 3, on the other hand, is globally available via ImageFX.

Source