Google’s Veo is an advanced AI video generation model developed by Google DeepMind. Unveiled at Google I/O 2024, Veo is capable of producing high-quality 1080p videos from text prompts, supporting a range of cinematic effects, and offering creative control over the video production process.

In December 2024, Google introduced Veo 2, an enhanced version of the model that generates videos with resolutions up to 4K. Veo 2 exhibits a better understanding of real-world physics and the nuances of human movement and expression, resulting in more realistic and coherent video outputs.

Veo 2 is available through Google Labs’ VideoFX platform on a waitlisted basis. Users can sign up for access via a Google Form, with access granted provisionally by Google.

The model also understands the language of cinematography, allowing users to specify genres, lenses, and cinematic effects to achieve desired results. Despite these advancements, Google acknowledges that Veo 2 may still produce occasional inaccuracies, such as extra fingers, but notes that these occurrences are less frequent compared to earlier versions.

Veo 2 incorporates DeepMind’s SynthID technology, embedding invisible digital watermarks into generated content to help identify AI-generated media and reduce misinformation.

For more detailed information, visit the DeepMind Veo page.