Make-A-Video, Meta AI, generate videos through words

One of the most recent innovations in the sphere of technology are artificial intelligences (AI) that generate images from words and/or phrases, however, the Meta intends to go further with this concept. The new AI of Zuckerberg call Make A Video manages to generate not only images, but videos, through words. Although it is not available to the public (for now), it is possible to activate notifications for when the tool is freely accessible. Come meet her!

from image to video

Mark Zuckerberg says you just need"describe how you want your art for a video to be created", and that's exactly what the Make A Video does. They are short videos, something that approaches a gif, for example, but which has already impressed many people with its technological capacity. The video quality is also not the best, which may not be a negative point, as this can be considered the pioneering tool in the video category.

Just like any other AI art program, the Make A Video was trained on “text image data,” or existing images that were labeled with a description explaining the object, person, or scenario depicted. But beyond that, the technology was also programmed to study video footage to understand “how the world moves”, according to the Meta.

The research of the Make A Video is based on recent progress made in text-to-image generation technology, thus enabling the text to video generation. The system uses images with descriptions to learn what the world is like and how it is often described. He also uses unlabeled videos to learn how the world moves — like people, certain objects, animals, etc. With these data, the Make A Video lets you bring your imagination to life by generating creative and unique videos with just a few words or a short text.

It is much more difficult to generate video than photos because in addition to correctly generating each pixel, the system also needs to predict how they will change over time. Make-A-Video solves this by adding an unsupervised learning layer that allows the system to understand motion in the physical world and apply it to traditional text-to-image generation.

Explains Mark Zuckerberg in a Facebook post about the new tool

Examples of Make-A-Video

Again, as this is a newer tool, obviously everything won't be as perfect as expected. In some videos, artificial intelligence is not always able to replicate movements or logic that actually exists in real life — or the realistic intent of what the video would like to represent. Check out some examples of possible imperfections generated by the Make A Video:

Still, the AI ​​program is impressive and potentially opens the door to improvements in assertiveness, as well as giving anyone the chance to create whatever videos they want. Unfortunately, technologies can also be an evil that we must always stay attentive. It is not difficult to imagine someone abusing the Make A Video all with create deepfakes and spread misinformation. The system also comes from Meta, a company accused of prioritizing profits over preventing misinformation.

Goal about the novelty

On the other hand, the Meta say you are committed to responsibly develop AI-based technologies.

As a way to reduce the risk of harmful content being generated, we examine, apply and iterate on filters to reduce the potential for harmful content to be displayed in videos.

Meta's positioning, in relation to possible bad faith uses with Make-A-Video

All videos generated in Make A Video will also include a watermark to indicate that it was created by artificial intelligence. Those who wish to test the technology at a later time can sign up for this application form. Then just wait until the Meta contact us to invite you to test the new tool.

Check out some more examples of those who had the opportunity to use the tool:

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

Adds Zuckerberg's company, talking about the future availability of the program freely to the public

There is also a specific page with a catalog full of Make-A-Video creations, there you can find videos created with the most different types of categories. Check it out and let us know what you think in the comments!

See also other features

This new AI-powered image creation — and now videos too — is making the rounds on and off the internet! recently a man won an art competition using a piece that he did not “create by hand”, but generated through a program. Just take a look.

Pig iron: PCMag, Mashable, TechCrunch e Meta.