Mark Zuckerberg has given users a glimpse of Meta‘s new artificial intelligence called Make-A-Video, which allows videos to be generated from text descriptions. Creating images with AI is already extremely impressive, allowing for beautiful portraits, landscapes, and other art to be created with just descriptive text. Meta is pushing the boundaries with AI and bringing something even more impressive – text-to-video AI generation.
AI image generation from text descriptions has been around for a while now and with tools such as DALL-E, Midjourney, and others, there are multiple options to create images from text. Creating videos from text descriptions was the next logical step in AI generation, and Meta may have delivered with its Make-A-Video AI.
So what exactly is this new Make-A-Video AI system from Meta, and how does it compare in terms of quality and ease of use? Well, from the videos shared thus far, it seems to be pretty far along and appears very similar to what DALL-E and other AI art generators would offer if they created videos. It’s not perfect by any means and some videos look better than others, but overall, the results are quite impressive. Make-A-Video can take previous still images and groups of images to create videos, and can even use videos to create variations of those videos. Mark Zuckerberg’s Facebook post about Make-A-Video gives us a glimpse of what to expect from the new AI.
Text-To-Video AI Is Tricky
As Zuckerberg explains in his post, “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.” This makes sense as there are more factors involved when creating motion compared to creating a still image. Meta uses a layer of unsupervised learning to help tackle this problem, which makes it easier for the system to understand motion. However, unsupervised learning can result in misuse of the software, such as porn, violence, or racism. Once this tool is available to the public, Meta is likely to ban specific phrases and words to help prevent misuse of its AI.
AI image generation is already extremely powerful and can create amazing works of art or just funny memes, but with AI video generation, more creative possibilities open up for users. Perhaps one day we will be able to create an entire movie with only text descriptions and advanced artificial intelligence. Until then. Meta‘s Make-A-Video technology is an important step toward that future.
