Meta Movie Gen: Revolutionizing AI-Generated Videos with State-of-the-Art Technology
Discover Meta Movie Gen, a groundbreaking AI model that sets new standards in video generation, personalization, and editing. Experience high-quality, customizable videos and explore the future of media creation with advanced generative AI capabilities.
Frequently Asked Questions
Have questions about Movie Gen? We've got answers. Browse through our FAQs to learn more about Movie Gen
Movie Gen is a suite of foundation models developed by Meta that can generate high-quality, 1080p HD videos with synchronized audio, and also perform precise instruction-based video editing and generation of personalized videos based on a user’s image.
Movie Gen offers capabilities such as text-to-video synthesis, video personalization, video editing, video-to-audio generation, and text-to-audio generation.
The largest model in the Movie Gen suite is a 30B parameter transformer designed for text-to-video generation, capable of generating 16-second videos at 16 frames-per-second.
Movie Gen achieves video personalization by conditioning the model on an image of a person, allowing it to generate videos featuring the chosen person while maintaining the identity and following the text prompt.
Movie Gen introduces a model that can perform precise video editing based on textual instructions, utilizing a multi-stage training process that includes single-frame editing, multi-frame editing, and video editing via backtranslation.
Movie Gen has a model called Movie Gen Audio that generates high-quality cinematic sound effects and music synchronized with the video input, aligning with the actions and following an input text prompt.
Movie Gen aims to release benchmarks such as Movie Gen Video Bench, Movie Gen Edit Bench, and Movie Gen Audio Bench to enable future benchmarking and accelerate research in media generation models.
Movie Gen scales its training data and model size by pre-training on internet-scale image, video, and audio data, and using a simple Transformer-based model trained with Flow Matching, which allows for efficient training of large-scale media generation models.
Movie Gen has broad applications in media creation, allowing for the generation of personalized and high-quality videos, as well as precise editing, which can be utilized in film production, advertising, content creation for social media, and more.
Movie Gen can generate audio for videos of arbitrary lengths through a single-shot generation for shorter videos and an audio extension technique for longer videos, which involves generating audio in segments and stitching them together seamlessly.