Google’s new AI video model is less absorbed in physics

Google may have only recently started rolling out its app View Generative artificial intelligence for Enterprise clientsbut the company is wasting no time rolling out a new version of the video tool to early testers. Google announced on Monday Preview View 2. According to the company, Veo 2 “understands the language of cinematography.” In practice, this means that you can indicate a specific type of film, cinematic effect, or lens when claiming the form.

Additionally, Google says the new model has a better understanding of real-world physics and human movement. Correctly modeling humans in motion is something all generative models struggle to do. So the company’s claim that the Veo 2 is better when it comes to both problem points is noteworthy. Of course, the samples provided by the company are not enough to know for sure; The real test of the Veo 2’s capabilities will come when someone orders it Create a video of a gymnast’s routine. Speaking of things that video models suffer from, Google says Veo will produce artifacts like extra fingers “less frequently.”

Separately, Google is rolling out improvements to Picture 3. As for the text-to-image model, the company says the latest version generates brighter, better-composed images. In addition, it can render more diverse art styles with greater accuracy. At the same time, it is also better to follow directions more faithfully. Immediate commitment was an issue it highlighted when the company made Imagen 3 available to Google Cloud customers earlier this month, so if nothing else, Google is aware of the areas where its AI models need to work.

Veo 2 will be rolled out gradually to… Google Labs Users in the United States. For now, Google will limit testers’ ability to produce up to eight seconds of footage at 720p resolution. For context, Sora It can create up to 20 seconds of footage at 1080p, although doing so requires $200 per month. ChatGPT Pro subscription. As for the latest improvements to Imagen 3, they are available to Google Labs users in more than 100 countries through imagefx.