We’ve seen AI programs that can generate professional-looking still art from a mere line of typed text, but now Facebook parent company Meta is bringing the technology to the video realm.
The company today introduced Make-A-Video, which can create short videos from whatever you tell it to. For example, it can generate a clip of a robot surfing an ocean wave, a teddy bear painting a self-portrait, or a spaceship landing on Mars.
Although the video quality is a little low, the results are still stunning, as Meta CEO Mark Zuckerberg showed in a demo. The technology is able to create consistent, realistic-looking images frame by frame, without any major flaws.
“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,” Zuckerberg wrote in a Facebook post. "Make-A-Video solves this by adding a layer of unsupervised learning that enables the system to understand motion in the physical world and apply it to traditional text-to-image generation."
Like other AI art programs, Make-A-Video was trained on “text-image data” or existing images that have been labeled with a description explaining the object, person, or setting depicted. But on top of this, the technology was also programmed to study video footage to understand “how the world moves,” Meta says.
“With just a few words or lines of text, Make-A-Video can bring imagination to life and create one-of-a-kind videos full of vivid colors, characters, and landscapes,” the company adds. “The system can also create videos from images or take existing videos and create new ones that are similar.”
That said, the technology isn't perfect. As you can see in the videos below, the Make-A-Video can't quite replicate a sense of motion, especially through moving limbs.
Still, the AI program is impressive, and potentially opens the door for anyone to dream up whatever videos they'd like, without paying for the Hollywood special effects. But it's also scary. It’s not hard to imagine someone abusing Make-A-Video to create “deepfakes" and spread disinformation. The system also comes from Meta, a company that’s been accused of prioritizing profits over stopping misinformation.
However, Meta says the company is committed to developing AI-powered technology responsibly. “As a way to reduce the risk of harmful content being generated, we examine, applied, and iterated on filters to reduce the potential for harmful content to surface in videos,” Meta says on the Make-A-Video site.
All videos generated over Make-A-Video will also include a watermark to indicate it was created by AI. The company adds: “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.”
Those who want to test out the technology in the future can sign up on this page. You can find more samples of the AI-created videos here.