LivePortrait
Efficient portrait animation generation tool
- Generate portrait animations of different styles (reality, oil painting, sculpture, 3D rendering) from static images
- Editing portrait videos using Kling generated source videos
- Control the opening and closing degree of eyes and lips based on a given scalar
- By fine-tuning animal data, precise driving of cats, dogs, and pandas can be achieved
- Implementing a 12.8ms generation speed with PyTorch on RTX 4090 GPU
Product Details
LivePortrait is a portrait animation generation model based on an implicit keypoint framework, which synthesizes realistic videos by using a single source image as a visual reference and extracting actions (such as facial expressions and head poses) from driving videos, audio, text, or generative sources. This model not only achieves an effective balance between computational efficiency and controllability, but also significantly improves generation quality and generalization ability by expanding training data, adopting a hybrid image video training strategy, upgrading network architecture, and designing better motion transformation and optimization objectives.
