Skip to content
LivePortrait
Turns static portraits into video/audio-driven 3D models in real time
Features
System Requirements
Minimum 16GB RAM. 13GB+ storage recommended.
macOS 15+: M-series chips required.
Windows 10/11: NVIDIA GPU with 6GB+ VRAM required.
Note: For NVIDIA GPUs, install a newer driver.Introduction
Developed by KwaiVGI (Kwai Visual Graphics Intelligence Team), LivePortrait is an open-source real-time portrait animation technology solution. Based on deep learning and computer graphics, this technology transforms static portrait images into dynamic 3D models with natural expressions, movements, and poses, supporting real-time driving via video, audio, or sensor data.
Technical Features and Functions:
- High-precision 3D Modeling：Using deep learning algorithms, LivePortrait constructs high-fidelity 3D facial models that capture fine details like wrinkles and muscle movements, achieving near-photorealistic portrait restoration.
- Real-time Expression and Motion Transfer：It extracts facial expressions, head poses, and other features from input videos or driving sources, then maps them to the target portrait model in real time, ensuring smooth and natural animations.
- Multi-modal Driving Capability：Supports animation driving through various input sources, including camera videos, audio, and motion sensors, adapting to different application scenarios.
- Cross-platform Deployment Optimization：Optimized for mobile devices, PCs, and other terminals, it achieves low-latency rendering while maintaining image quality, supporting platforms like iOS, Android, and Web.
- Open-source and Scalable：As an open-source project (hosted on GitHub), it provides a complete codebase and development documentation, enabling developers to customize features or integrate them into existing projects.
Core Advantages:
- High Realism and Real-time Performance：Combining deep learning with 3D rendering technology, it delivers photo-realistic portrait animations with millisecond-level response.
- Low-threshold Integration：Offers concise APIs and toolchains to reduce technical access costs, allowing users to use it without extensive 3D modeling experience.
- Diverse Application Scenarios：Suitable for virtual idols, digital human live streaming, interactive entertainment, video special effects, virtual avatars for remote meetings, etc., providing users with personalized dynamic visual experiences.
LicenseMIT