Meta, the parent company of Facebook, revealed on Friday the development of a new artificial intelligence model known as Movie Gen. This technology is designed to produce realistic video and audio clips in response to user commands, and the company claims it can stand alongside offerings from prominent media generation startups such as OpenAI and ElevenLabs.
The samples shared by Meta included a variety of videos showcasing animals swimming and surfing, alongside clips featuring real photos of individuals engaged in activities such as painting on a canvas.
Additionally, Movie Gen has the capability to generate synchronized background music and sound effects tailored to the content of the videos. In a demonstration, the AI modified a video by placing pom-poms into the hands of a man running alone in a desert, while another clip transformed a skateboarder’s environment from a dry parking lot into one splashed with water.
According to Meta, the videos generated by Movie Gen can last up to 16 seconds, with audio clips extending to 45 seconds. The company has shared findings from blind tests suggesting that the AI model’s performance is competitive with those from startups including Runway, OpenAI, ElevenLabs, and Kling.
This announcement comes at a time when Hollywood is grappling with the potential applications of generative AI technology in video production. Earlier this year, Microsoft-backed OpenAI showcased its product Sora, which can create feature film-quality videos from textual prompts.
Despite the interest in using these tools to streamline and enhance filmmaking, there are concerns about the implications of systems trained on copyrighted material without authorization.
Legislators have also voiced apprehensions regarding the misuse of AI-generated deepfakes, particularly in electoral contexts across various countries, including the United States, Pakistan, India, and Indonesia.
Representatives from Meta indicated that the company is not planning to make Movie Gen available for open use by developers, unlike its Llama series of large language models, citing the need for careful consideration of the risks associated with each model. No specific evaluations about Movie Gen were provided.
Instead, Meta plans to collaborate directly with the entertainment industry and other content creators to explore the use of Movie Gen. It is expected that this technology will be integrated into Meta’s products in the coming year.
The blog post and accompanying research paper released by Meta noted that a mix of licensed and publicly available datasets formed the foundation for the development of Movie Gen.
This year, OpenAI has engaged in discussions with Hollywood executives and agents to explore potential partnerships around Sora, but as of now, no agreements have been finalized. Concerns regarding OpenAI’s practices heightened in May when actress Scarlett Johansson accused the company of using her voice inappropriately without consent for its chatbot.
In September, Lions Gate Entertainment, the studio behind popular franchises like “The Hunger Games” and “Twilight,” announced it would be providing AI startup Runway with access to its film and television assets to support the training of an AI model. In exchange, the studio and its filmmakers would be able to utilize the model to enhance their creative work.
© Thomson Reuters 2024
(This story has not been edited by NDTV staff and is auto-generated from a syndicated feed.)
