AI · AI model
LivePortrait
Efficient portrait animation from still photos guided by driving videos.
LivePortrait is an open-source portrait animation framework developed by researchers at Kuaishou Technology. It accurately transfers facial expressions, gaze, and head movements from a driving video or motion template onto a static portrait photograph or existing video. It is designed for high computational efficiency, running in milliseconds per frame on modern GPUs.
What it's for
- Animate static portrait photos using facial motions from a driving video
- Edit head poses and expressions in existing portrait videos (video-to-video)
- Generate animated talking avatars for game design, social media, or virtual assistants
- Animate animal faces using the specialized animal model weights
Features
- Stitching and retargeting modules for accurate eye and lip movement control
- Image-to-video and video-to-video animation modes
- Gradio web UI and CLI inference scripts
- Driving video auto-cropping and motion template (.pkl) extraction
- Apple Silicon (MPS) and Windows standalone package support
- Community integrations with ComfyUI workflows
- Dedicated model variant for animating animal faces
Pricing
Open source. Free and open source under the MIT license.
Details
- Type
- AI model
- Category
- AI
- Topic
- Avatar Animation
- Open source
- Yes
- Platforms
- Linux, Windows, macOS, Command line, Code library, Self-hosted, Web
- For
- AI researchers, video creators, animators, and developers building talking-head avatar workflows.
- Alternative to
- SadTalker, HeyGen, D-ID, First Order Motion Model
- Tags
- liveportrait, portrait animation, talking head, computer vision, pytorch, comfyui
Why it was saved
- @artemiy.miller: Accurately transfers real-time facial expressions, gaze, and lip sync to a still photograph.
Compare
Related
Browser that installs and runs open-source AI apps with one click
AI platform to generate personalized portrait models and talking animated avatars
AI camera assistant interface built around the iOS Dynamic Island
AI framework generating first-person egocentric videos from third-person video footage
Animate still portraits using facial expressions from a driving video.
Open-source vision model for zero-shot object segmentation and tracking in video
Updated