AI Video Creator

Is Make-A-Video publicly available today?

Our goal is to eventually make this technology available to the public, but for now we will continue to analyze, test, and trial Make-A-Video to ensure that each step of release is safe and intentional.

​

How can I sign up to get updates or early access?

If you're interested, you can sign up to express interest in future releases of Make-A-Video research and receive notifications. Use the sign-up form to indicate you’d like to be contacted about upcoming releases.

​

How does Make-A-Video generate videos from text?

Make-A-Video leverages advanced AI that learns from images with descriptions to understand what the world looks like and how it is described, and it uses unlabeled videos to learn how the world moves. With this learned knowledge, it generates videos based on textual prompts.

​

What video styles can I choose from?

You can create videos in Surreal, Realistic, or Stylized styles.

​

How long can a generated video be?

Current support is for videos up to 10 seconds in length.

​

Can Make-A-Video animate a still image or interpolate between images?

Yes. You can add motion to a single image or interpolate motion between two images to create in-between frames.

​

Do generated videos include a watermark?

Yes. All videos generated by Make-A-Video include a watermark to indicate they are AI-generated.

​

Can I create variations of a video based on the original?

Yes. You can create variations of your video based on the original prompt or video you started from.

​

What data is used to train Make-A-Video and how is safety addressed?

The technology analyzes millions of pieces of data to learn about the world. It uses images with descriptions to learn visual concepts and unlabeled videos to learn motion. To reduce the risk of harmful content, filters are applied and iterated on as part of responsible development.

​

What are the limitations of Make-A-Video?

• Duration is limited to a maximum of 10 seconds.
• The quality and coherence of the output depend on the clarity and specificity of the text input.
• The system relies on existing training data, which may limit representation of novel concepts.
• All videos include a watermark to indicate AI generation.

​

Where can I read the Make-A-Video research paper?

You can read the Make-A-Video research paper for detailed technical background and methodology.

​