Veo was trained on a vast amount of footage, some of which may have been sourced from Google's own YouTube. Google has already made Veo available to select creators, including Donald Glover and his creative agency Gilga. However, the model is not perfect and has limitations, such as objects disappearing and reappearing without explanation, and incorrect physics. Therefore, Veo will remain behind a waitlist on Google Labs for the foreseeable future. As it improves, Google plans to incorporate some of Veo's capabilities into YouTube Shorts and other products.
Key takeaways:
- Google has unveiled Veo, an AI model that can create 1080p video clips around a minute long from a text prompt, at its I/O 2024 developer conference.
- Veo is built on Google's preliminary commercial work in video generation and is competitive with leading video generation models like OpenAI's Sora.
- The AI model was trained on a large amount of footage, some of which may have been sourced from YouTube, according to Douglas Eck, who leads research efforts at DeepMind in generative media.
- Despite its capabilities, Veo has limitations such as objects disappearing and reappearing without explanation, and incorrect physics, which is why it will remain behind a waitlist on Google Labs for the foreseeable future.