Make-A-Video AI: ChatGPT, But For Videos

AI chatbots can do your homework, write essays, and explain concepts from a simple user prompt. Soon, it could create videos for you!

Facebook company Meta is developing an AI tool called Make-A-Video that creates videos from text descriptions.

In other words, it is like the popular ChatGPT AI tool but creates videos instead of text. Meta describes it as a work in progress, but the initial features may shock you!

How does Make-A-Video work?

We’re pleased to introduce Make-A-Video, our latest in #GenerativeAI research! With just a few words, this state-of-the-art AI system generates high-quality videos from text prompts.

Have an idea you want to see? Reply w/ your prompt using #MetaAI and we’ll share more results. pic.twitter.com/q8zjiwLBjb

— AI at Meta (@AIatMeta) September 29, 2022

Meta AI discussed the Make-A-Video tool in Cornell University’s research-sharing platform, arXiv.

Based on the paper, it takes samples from a database, similar to how other AI platforms work. However, Meta’s proprietary technology uses the training data in short clips.

You may head to makeavideo.studio to see the samples divided into three categories: surreal, realistic, and stylized. Here are some examples: 

• A teddy bear painting a portrait
• Cat watching TV with a remote in hand
• Horse drinking water
• Clownfish swimming through the coral reef
• Hyper-realistic spaceship landing on Mars

Also, Make-A-Video could turn your photos into clips. For example, its website shows a picture of a vintage ship animated so that it sails through waves.

Provide more pictures, and it could turn them into a single clip. For instance, the Meta AI tool animated two asteroid photos so that the first would realistically turn into the second. 

Moreover, Make-A-Video lets you control how the AI-generated video moves. For example, its website shows that it can make a 3D bear dance in four ways.

Photo Credit: techcrunch.com

The homepage promotes the tool as “The New State of the Art in Video Generation.” Furthermore, Make-A-Video’s results represent text prompts more accurately than other tools.

Its AI-generated videos also deliver three times higher quality. Meta AI allegedly reduces “the risk of harmful content being generated.”

Moreover, the company says it adds watermarks to all its videos. Yet, it has not shared specific details about its training data and policies for potential AI copyright issues.

Conclusion

The Facebook company Meta is developing an AI tool that creates videos from text prompts. However, it says the program is a “work-in-progress.”

At the time of writing, it has not specified a release date for this tool. Meta AI opened a waitlist for people who want to try its new project.

Unfortunately, it closed the Google Form for registration. Stay tuned to future updates from Meta and other digital trends at Inquirer Tech.