This showreel from Google DeepMind's new AI video generator Veo 2 was released a few weeks ago. Like other AI video apps, all the shots were generated from... Read more
This showreel from Google DeepMind's new AI video generator Veo 2 was released a few weeks ago. Like other AI video apps, all the shots were generated from simple text prompts - similar to those found in shooting scripts or briefs for graphics companies. The results are the most impressive demonstration of AI video yet.
As someone who works in documentaries, I can't help feeling that as mind-blowing as the fantasy shots are, it's the remarkably convincing depictions of the 'real world' that are the real shocker. (Who is the beekeeper? Or the researcher with the microscope? From what corner of the AI's imagination do they come from?)
Like OpenAI and others, Google has been coy about the material they've used to train the AI. Yet, from the quality of these shots, it would seem the AI has been 'watching' a lot of high-end material.
AI-generated videos are beginning to populate YouTube, but the revolution is just getting started. We haven’t seen anything yet. In the meantime, the story of how AI learned to see and understand our visual world will make for a great documentary.
As someone who works in documentaries, I can't help feeling that as mind-blowing as the fantasy shots are, it's the remarkably convincing depictions of the 'real world' that are the real shocker. (Who is the beekeeper? Or the researcher with the microscope? From what corner of the AI's imagination do they come from?)
Like OpenAI and others, Google has been coy about the material they've used to train the AI. Yet, from the quality of these shots, it would seem the AI has been 'watching' a lot of high-end material.
AI-generated videos are beginning to populate YouTube, but the revolution is just getting started. We haven’t seen anything yet. In the meantime, the story of how AI learned to see and understand our visual world will make for a great documentary.
