Google Veo 2 AI Video Model Now Available for Advanced Gemini Users
Google is rolling out a new way to create videos as well as animate existing images using artificial intelligence. Trick? These videos are very short and you have to pay to watch them.
On Tuesday, Google announced that users of Gemini Advanced and Google One AI Premium can now create AI-generated videos using the company’s Veo 2 video model. Google originally introduced this model back in December , touting an improved understanding of real-world physics, the nuances of human emotion, and the ability to create certain types of shots (e.g. low angle, close-up, shallow depth of field).
While the Veo 2 is capable of generating minutes-long 4K video, Gemini Advanced users won’t get that experience here. Instead, you’ll be able to shoot eight-second videos at 720p, which is quite a limitation. Speaking of limitations, there is a limit to the number of videos you can create with Veo 2 each month, although Google doesn’t disclose this publicly. Instead, the company says it will alert you when you’re close to your limit, which doesn’t seem like a particularly transparent system.
While Gemini Advanced users will be able to use Veo 2 directly in Gemini, Google One AI Premium subscribers also have access to the model in Whisk , the company’s AI-powered media generator. As part of the Veo 2 release, Google is announcing Whisk Animate, which uses Veo 2 to animate an image created with Whisk.
How to Create a Veo 2 Video Using Gemini Advanced
If you pay $20 a month for Gemini, Veo 2 is out right now. To get started, open the Google chatbot, then select “Veo 2” from the model drop-down list. From here, give the AI prompts as usual, only this time with a specific video in mind. While Google encourages users to write things like short stories for Veo 2 to work, remember that you can only create a maximum of eight seconds of video. This is a very short story.
If you don’t see Veo 2 as an option, sit tight. Google says the full rollout could take several weeks, so not all Gemini Advanced users will be able to access it today.
How to Animate a Whisk Image Using Veo 2
Google One AI Premium subscribers can try this feature at labs.google/whisk . You can start by creating an image using Whisk as usual – by suggesting Whisk whatever you want, or by letting the AI randomly select a generation for you.
Once your image is created, you will see a new “ANIMATE” option at the top. Click on it and then tell the AI how you want to animate the image.
What’s the point?
I don’t have a subscription, so I can’t try Veo 2 in any capacity. But according to Google demos, the model does a pretty good job of responding to suggestions. For example, Google demonstrated how four different versions of a person eating ice cream, including 3D art, pixel art, illustration, and realistic generation, can be animated with a single prompt.
However, they are far from perfect: each one exhibits the hallmarks of AI-generated video , including visual glitches and elements morphing into each other. Some of the results here are also half-baked: for example, the 3D animation shows a man bringing ice cream to his mouth but never eating it, while the realistic generation shows him eating with his fingers even though there is no ice cream there.
Of course, what’s new about this feature is that you can create a short animation of any image you want, whether it’s AI-generated or not. But I have a hard time imagining a practical or widespread use here, especially since the feature is locked behind Google’s $20/month paywall.