Meta introduced a base model capable of creating realistic-looking movies, rivaling OpenAI’s Sora and Google’s Veo in the emerging conceptual AI picture contest. On October 4, two innovative designs were unveiled:
- The 30B factor Movie Gen Video.
- The 13B factor Movie Gen Audio.
Both are based on Meta’s Llama 3 type. In 2025, the software large anticipates incorporating Movie Gen into Instagram.
What is the Model Family from Movie Gen?
The Movie Gen types are text-to-video or text-to-audio conceptual AI. According to Meta, Movie Gen may produce videos that are up to 16 seconds longer. In assessment, OpenAI’s Sora, now accessible to the community, you generate one-minute videos with various scenes. Veo, which is available to choose designers, can make films about a minute longer.
Movie Gen is controlled using normal speech. Users can then describe the scene they want to view, including each individual element and the general mood. They can even modify game factors based on natural language words prompts, such as removing or adding elements from a picture.
The personalisation aspect was enabled by “post-training processes”, Meta said. Following these steps, the AI was “maintaining the identity of the person while responding to the word prompt.” This allows people to spot themselves— or someone else — into a custom-made image.
In the first glance of the product, Meta’s product appears to be aimed mainly at content creators. In a blog post, Meta stated that the goal is to “help people show themselves in new methods and to give people options where they might not otherwise have them.”
SEE: There are strategies to help projects succeed, but modern transformation is occasionally seem like a blip.
Lights, actions, and audio
According to Meta’s research paper, Movie Gen Audio has the ability to make music or sound effects for videos that are “up to many days long.” The song is generated at 48kHz, and it can either be used as a song or to match the images on display.
Meta points out Llama 3’s ability to address safety and spoofing issues
For businesses, creating AI-created videos may drastically shorten the amount of time needed to create both internal and external information. On the other hand, using AI-generated content, particularly without identification, can create confusion among people and reduce confidence, evidenced by a new report by the the Journal of Hospitality Marketing and Management.
Perhaps in an effort to tackle the trust issues, Meta added a logo to Video Gen’s photos. In the lower left corner of the video is a open” beauty” creative frequently used to show AI.
Concerns are raised by surveillance and the use of relational AI to produce unsettling, harmful, or misinformed articles, especially in business use cases where the popularity of the business could be in jeopardized. Meta mentioned a statement from September that looked into safeguarding its AI types, including the Llama 3 home, in the news of Movie Gen. The report provides more information about the model’s ability to prevent unacceptable content and how images will have both visible and invisible tags.
