Google has announced that it is granting trial access to its AI video generator, Veo 2, for Gemini Advanced subscribers. This text-to-video AI model is said to be capable of producing high-resolution video clips with "cinematic realism."

Starting today, subscribers can select Veo 2 from the dropdown menu of the Gemini model on both web and mobile platforms, and generate videos by entering prompts. The resulting videos are 8 seconds long and rendered in 720p resolution.

Google has set a monthly cap on the number of videos subscribers can generate, notifying users as they approach the limit. Videos created by Veo 2 are output in MP4 format, and mobile users can directly upload their creations to TikTok and YouTube using the "share" button.

According to Google, the upgraded AI model demonstrates a deeper understanding of real-world physics and human movements, delivering "smoother character animations, more realistic environments, and finer visual details across a variety of themes and styles." All videos generated by Veo 2 come embedded with SynthID digital watermarking to indicate their AI-generated origin.

In addition, Google has made its image-to-video tool, Whisk Animate, available to Google One AI Premium subscribers. Built upon the existing Whisk image blending tool, this feature allows users to convert images into 8-second Veo 2 videos. Whisk Animate is now globally accessible to subscribers through Google Labs.

It is reported that Google initially launched Veo 2 in an early access format last December, when the tool was priced at $0.50 per second of video on the Vertex AI platform. Since then, YouTube has integrated Veo 2 into its experimental Dream Screen feature, enabling users to create AI-generated video clips for Shorts.