That price is from 3 years ago.Meta’s new AI-powered video generator produces high-resolution footage with sound, the company announced today. The announcement comes a few months after rival OpenAI unveiled its text-to-video model, Sora, but Movie Gen has not yet been made publicly available.
Movie Gen uses text input to automatically generate new videos and edit existing footage and still images. The audio added to the video is also AI-generated, harmonizing the image with ambient noise, sound effects and background music, according to a New York Times report. Videos can be generated in a variety of aspect ratios.
In addition to generating new clips, Meta said Movie Gen can also create custom videos from images or take existing videos and modify different elements to them. One example shared by the company shows a still portrait photo of a woman. The added video shows her sitting in a pumpkin patch and having a drink.
Movie Gen can also be used to edit existing footage, changing styles and transitions, or adding things that weren’t there before. In an example shared by Meta, a relatively innocuous video of what appears to be an illustrated runner has been edited in several ways using AI. In one frame, he is holding pom-poms. In another photo, the background has been edited to depict a desert. In the third, the runner is wearing a dinosaur costume. Changes can be made using text prompts.
AI businesses have advanced the technology even further since powerful AI picture and video generators became widely available almost two years ago. In the last six months, both smaller startups and major tech giants like Google and OpenAI have been developing tools of a similar nature. Though it was first shown in February, OpenAI’s Sora has yet to go live; last week, a co-lead on the video generating project departed to join Google.
Chris Cox, the chief product officer of Meta, states on Threads that the product “[isn’t] ready to release this as a product anytime soon” since the generation time is too long and it is still too expensive.
AI image generators have raised concerns about ownership and potentially harmful use cases, and AI video generators exacerbate them. AI startup Runaway reportedly trained its video generator with thousands of scraped YouTube videos, but YouTube CEO Neal Mohan said this already violates the platform’s terms of service. Meta said in a blog post that it trained Movie Gen using “a combination of licensed and publicly available datasets,” but did not disclose which datasets.
Filmmakers, photographers, artists, writers, actors and other creative people are also concerned about how AI generators will affect their livelihoods, and have been at the center of several strikes, including the historic joint Hollywood strike by the Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA) and the Writers Guild of America (WGA) last year.