According to Meta, the Movie Gen AI model can create films up to 16 seconds long in 1080p resolution and can also provide background music for videos.

Movie Gen, a new AI model from Meta, can generate both audio and video given textual inputs. Meta's Movie Gen AI model can produce films and associated audio based on user descriptions, competing with OpenAI's Sora. According to the business, it can also create customized videos that feature real people in different settings, captured in their own images. Text inputs can be used to alter or improve the output videos. Reuters noted that Meta is unlikely to make Movie Gen available for developers to utilize publicly, in contrast to the Llama series of AI models.

Meta clarified that the Meta Movie Gen model has been developed for both text-to-image & text-to-video tasks in a research article introducing the new Meta AI model. It creates many colored pictures when asked, each of which acts as a frame for the movie.

According to Meta, Movie Gen is capable of producing 1080p high-definition videos that can be 16 second duration at 16 frames per second (FPS). The model can produce films with varying resolutions, lengths, and aspect ratios within its constraints. According to the business, the model may reason regarding object motion, camera motion, subject-object communication, and other topics after learning real-world sights by "watching" videos.

According to Meta, the Movie Gen model may generate related audio using text-to-audio and video-to-audio methods. According to the business, it is capable of producing 48 kHz audio with synchronized music and theatrical sound effects for the visual input. Although the model can only produce short AI-generated movies, it can produce "long-form coherent audio for videos up to several minutes long."