Meta has released the MovieGen video generation model, featuring the following capabilities:
- MovieGen Video: A 30-billion parameter Transformer model that can generate high-quality, high-definition images and videos from a single text prompt.
- MovieGen Audio: A 13-billion parameter Transformer model that processes video inputs to produce synchronized high-fidelity audio, as well as generate background music and ambient sound effects.
- Precise Video Editing: Using either generated or existing footage along with text prompts as inputs, the model can perform edits such as adding, deleting, or replacing elements, or making global changes like altering backgrounds or styles.
- Personalized Videos: By using character images and text prompts, the model can produce videos with state-of-the-art results in maintaining character presence and natural movement.