Google has introduced Videopoet, a new AI technology that expands the capabilities of image-generating AI such as Dall‑E or Stable Diffusion. This advanced AI is not only capable of creating videos on its own, but can also edit existing videos and transform simple images into moving scenes.
Videopoet supports different modes. From text to video and image to video to video stylization and the generation of audio for videos, everything we can do with this Google AI video generator. Google shows what Videopoet can do on the official website.
Rookie the Raccoon Goes on Wild Adventures
To demonstrate the capabilities of AI, Google researchers created a one-minute video called “Rookie the Raccoon.” In this one, a raccoon experiences exciting adventures that take him around the world and even into space. The AI shows how it uses various effects to tell a coherent story in the short clips.
However, the research results go even further. The AI research page shows how existing videos we can change using this tool. For example, the AI transforms a hamster that was originally seen with a gold nugget in its hand into a hamster on the beach, complete with a Hawaiian shirt and an inflatable ball.
What Makes Google’s AI Special?
Videopoet isn’t the only video AI on the market, but it stands out with some special features, as the researchers point out. A standout feature of Videopoet is the ability to bring a still image to life using a text prompt.
This is demonstrated in examples such as the “Mona Lisa” turning towards the camera or yawning. Another aspect is the precise control of camera movements in the clips, which can be controlled using targeted text prompts.
In the future, Google plans to expand the model to include the so-called any-to-any generation. This means that text, audio, image, and video can be generated in any combination. What’s more! They plan to include AI-powered subtitle creation, among other features.
Read more: How to Know if You Are Blocked on Instagram