Google first announced Veo at its I/O a couple of months back, Google has now unveiled its successor- Veo 2. It is an advanced video generation model. In addition to this, the tech giant also announced Imagen 3 and a new Whisk experiment. All of these announcements clearly hint at Google’s ongoing efforts towards AI-driven content creation and integration with its Gemini model.
Veo 2 features an enhanced understanding of real-world physics, human movement, and expressions. This helps it get improved realism and detail in AI-generated videos. With this AI tool, users can fine-tune outputs and get cinematic precision by giving some specific prompts. The videos will look as if they were shot in reality.
You can type in prompts like a “low-angle tracking shot,” a “close-up shot of a scientist at work,” or specify a lens like “18mm” for a wide-angle perspective, and Veo 2 will know how exactly the scene will look in those situations and it will recreate it. Additionally, with prompts such as “shallow depth of field,” creators can recreate professional camera effects.
With Veo 2, Google has also significantly reduced the hallucination rate which is a recurring problem with generative AI models. To identify that the content generated by Veo 2 is AI-generated, an invisible SynthID watermark is used for transparency and traceability of AI-generated content.
Currently, Google is rolling out Veo 2 in VideoFX which is a Google Labs platform. In the future, it will further expand access to more users via a waitlist. The company also revealed that Veo 2 will integrate with YouTube Shorts and other products in 2025, signaling a push toward AI-powered video creation for content creators.
In addition to Veo 2, Google also showcased updates to Imagen 3, its image generation model, and also showcased a Whisk experiment, which explores further creative applications of its AI technology.