Meta Unveils Make-A-Video AI That Turns Text Into Short Clips
Meta introduces Make-A-Video, a new artificial intelligence system that generates five-second silent video clips from simple text prompts. The tool expands on recent text-to-image generators by predicting how pixels change over time.
Meta introduces a new synthetic media engine called Make-A-Video that generates short video clips from text prompts. Similar to text-to-image tools like Midjourney, this new platform creates silent videos up to five seconds long based on simple descriptions like a dog wearing a superhero cape or a spaceship landing on Mars.
The AI system learns what the world looks like from paired text-image data and understands motion by analyzing unlabeled video footage. Users can also generate new videos from still images or transform existing videos into similar variations, expanding creative possibilities for artists and creators.
Despite the impressive demonstrations, Meta acknowledges technical limitations and the potential for problematic outputs stemming from its massive web-scraped training data. The company currently restricts public access but plans to share the research with developers soon to test the technology's true capabilities beyond these early curated examples.