AI text-to-image generators have been in full swing lately. Even text-to-video has become a thing, now Meta has introduced its own. Just like the rest of these generators, Meta’s Make-A-Video AI model lets you turn textual prompts into short video clips.
Make-A-Video offers three different styles of videos: surreal, realistic, and stylized. And while I still find it impressive that this technology even exists – even the “realistic” videos give me the creeps.
“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,” Meta CEO Mark Zuckerberg writes 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.”
All of these videos were generated by an AI system that our team at Meta built. We call it Make-A-Video. You give it a text description and it creates a video for you.We gave it descriptions like: "a teddy bear painting a self-portrait", "a baby sloth with a knitted hat trying to figure out a laptop", "a spaceship landing on mars", and "a robot surfing a wave in the ocean".This is pretty amazing progress. 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. 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.We plan to share this as a demo in the future. In the meantime, enjoy the videos…
Posted by Mark Zuckerberg on Thursday, September 29, 2022
It’s worth noting that Meta isn’t the first to use this kind of technology in a social media app. TikTok introduced it in August, but it’s text-to-image rather than text-to-video. So it looks like Meta will finally do something first rather than ripping off TikTok’s ideas.
As I said, Meta’s text-to-video generator creates surreal, realistic, and stylized types of videos. The surreal ones are… Well, surreal, and all of the examples are going to give me nightmares for sure.
Some realistic video examples are pretty awesome, like this artist brush painting on a canvas:
But this couple walking in the rain is creepy as hell, they look like conjoined twins:
Zuckerberg writes that Meta plans to share this as a demo in the future, although it’s not specified when. In the meantime, check out the announcement page and see the videos generated from a few lines of text.
[via Digital Trends]