Meta Unveils Movie Gen: 30B-Parameter AI Model for 16-Second Videos with Synchronized Audio

Oct 4, 2024, 02:02 PM

Meta has unveiled Movie Gen, a new suite of AI models that can generate high-definition videos with synchronized audio from text prompts. Movie Gen Video, a 30-billion-parameter transformer model, and Movie Gen Audio, a 13-billion-parameter model, enable capabilities such as text-to-video generation, precise video editing, personalized video creation using user images, and video-to-audio generation with optional text prompts. The system can produce realistic videos up to 16 seconds long at 1080p resolution and 16 frames per second, complete with ambient sounds and background music. Developed by Meta's AI research teams, Movie Gen is positioned as the most advanced media foundation model to date, aiming to set a new standard for immersive AI content creation. Meta plans to integrate Movie Gen into its apps, including Instagram and Facebook, by 2025, challenging competitors like OpenAI and Google's generative AI offerings.