Meta (Facebook 2.0) has unveiled a brand new AI system that turns text into video. The AI tool is called Make-A-Video. Yes, you read that right, most people are not even aware that text-to-image AI models exist, and now we’re moving on to the next frontier: Text-to-video.
Make-A-Video is a state-of-the-art AI system that generates videos from text.
Look at this example 5-second clip that’s generated with this prompt:
A dog wearing a Superhero outfit with red cape flying through the sky
Although the result is pretty accurate, the aesthetic is like a trippy home video.
But despite the quirky initial results, the tech is a huge step forward in the AI image generation space.
“This is pretty amazing progress. It’s much harder to generate video than photos because beyond correctly generating each pixel, the system also has to predict how they’ll change over time,” — Mark Zuckerberg
Meta did not mention any release date when the Make-A-Video tool will be available for public use, but it will apparently prompt other AI companies like OpenAI and Stability.AI to make their own similar models.
In fact, Stability AI’s CEO, Emad Mostaque, said in a tweet that the team is working on a model that can output better results.
Make Still Image Move
The fun does not end with text-based video generation. The AI tool can also make videos from existing images.
Here’s a video clip generated from two pre-existing photos.