Google Launches AI to Create 1080p Videos Longer Than One

This article summarizes Google Launches AI to Create 1080p Videos Longer Than One and explains the most important context behind the development. It also highlights the potential impact on users and the broader market.

Overview

At the Google I/O event on the morning of May 15 (Hanoi time), Google for the first time introduced a command-based video creation AI called Veo with the ability to create 1080p videos longer than one minute with many visual and electronic styles. Different images, competing with OpenAI's Sora.

Under Google's introduction, Veo has the ability to understand natural language and cinematic language, can "accurately grasp the tone of the prompt", to create videos that closely express the user's creative vision.

The model can also create consistent and coherent footage, with human subjects, animals, and objects moving realistically throughout the shot.

Google said that you can make requests to extend the video length to 1 minute and 10 seconds, and use words to prompt additions and refine the results. The duration previously announced by OpenAI Sora is a maximum of 1 minute.

Veo is built on 5 video creation models including Generative Query Network (GQN), DVD-GAN, Imagen-Video, Phenaki, WALT, VideoPoet and Lumiere, combined with many other techniques. This helps improve the quality and resolution of the output video.

At the event, Google also introduced an image-generating AI, Imagen 3, with the ability to create images with "incredible levels of detail", realistic, lifelike images.

Imagen 3 also better understands natural language and predicts user intent and can create images with a variety of styles.

Both Veo and Imagen 3 are currently available for trial to some content creators, but have not yet been released to the general public.