Meta announced that they have made a new AI system that can let people turn text prompts into short video clips. It’s similar to the viral AI art generator Dall E, except that the images can sort of come to life.
“Make-A-Video builds on Meta AI’s recent progress in generative technology research and has the potential to open new opportunities for creators and artists. The system learns what the world looks like from paired text-image data and how the world moves from video footage with no associated text,” wrote Meta in their blog.
The system would be familiar if you’ve used or have heard about Dall E mini—which we’ve covered before back in June. The app boasted that it is able to create images “from text captions for a wide range of concepts expressible in natural language”. This resulted in a huge curiousity amongst the online community to make the most ridiculous AI-generated images.
However, unlike Dall E mini, we aren’t able to have access to Meta’s Make-A-Video yet. Meta AI has asked on Twitter to send them prompts via the platform if you want to request anything for now, and you can also sign up on their site if you want to try Make-A-Video yourself.
According to Meta, the system is able to make three types of videos—surreal (like a teddy bear painting a portrait), realistic (like an artists brush painting on a canvas close up), and stylised ( like a hyper-realistic spaceship landing on Mars). You can see the videos in action on the Make-A-Video site.
The results look pretty cool, but honestly? I get kind of creeped out watching some of these AI videos. Like, for the teddy bear painting the portrait, the AI system used a large human hand as the teddy bear’s hand. It’s a little disturbing, but most of the other results like the brush painting one are incredibly impressive for something generated by AI.
“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 system can also create videos from images or take existing videos and create new ones that are similar,” wrote Meta.
Additionally, Meta also noted that they “want to be thoughtful” about how they build new generative AI systems like Make-A-Video. They said that they have taken steps to reduce “the creation of harmful, biased, or misleading content”. This is as previous AI generators that have several racial and gender bias, like the ‘face depixeliser’.
“We are openly sharing this generative AI research and results with the community for their feedback, and will continue to use our responsible AI framework to refine and evolve our approach to this emerging technology,” continued Meta.
