Google’s new AI video model sucks less at physics

Google may have recently started rolling out its Veo generative AI to enterprise customers, but the company is wasting no time in offering a new version of the video tool to early testers. On Monday, Google announced a preview of Veo 2. According to the company, Veo 2 “understands the language of cinema.” In practice, that means you can reference a specific genre of film, cinematic effect, or lens when ordering the model.

Additionally, Google says the new model has a better understanding of real-world physics and human movement. Correctly modeling humans in motion is something that all generative models strive to achieve. So the company’s claim that Veo 2 is better when it comes to both pain points is notable. Of course, the samples the company provided are not enough to know for sure; The true test of Veo 2’s capabilities will come when someone tells it to generate a video of a gymnast’s routine. Oh, and speaking of things video models struggle with, Google says Veo will produce artifacts like extra fingers “less frequently.”

Separately, Google is rolling out improvements to Image 3. Of its text-to-image model, the company says the latest version produces brighter, better-composed images. Additionally, it can represent more diverse art styles more accurately. At the same time, it is also better at following directions more faithfully. Rapid stickiness was an issue I highlighted when the company made Image 3 available to Google Cloud customers earlier this month, so at least Google is aware of the areas where its AI models need to work.

I see 2 will be implemented gradually google labs users in the U.S. For now, Google will limit testers to generating up to eight seconds of footage at 720p. For context, Sora can output up to 20 seconds of 1080p footage, though doing so requires a $200-per-month ChatGPT Pro subscription. As for the latest improvements to Image 3, they are available to Google Labs users in more than 100 countries through ImageFX.