Skip to content

FasterLivePortrait

Real-time portrait animation with TensorRT and ONNX

Visit official site

What fasterliveportrait is

The project is described as a deep-learning tool for real-time portrait animation. It reaches 30+ FPS on an RTX 3090 GPU with TensorRT, counting pre- and post-processing.

What you can do with it

  • Show an animated portrait during a video call
  • Turn a still portrait photo into a moving video
  • Animate several faces at the same time

Getting started

  • Install Docker and pull the project image
  • Run the container with a Docker command
  • Place converted ONNX model files in the checkpoint folder
  • Configure onnxruntime-gpu or TensorRT

Pros & cons

✓ What we like

  • No PyTorch dependency after the code rewrite
  • Docker image plus Windows and MacOS packages

! What to watch out for

  • The listed speed is measured on an RTX 3090 GPU
  • Setup needs ONNX model files and onnxruntime-gpu or TensorRT

FAQ

How fast does it run?

The card lists 30+ FPS on an RTX 3090 GPU with TensorRT.

Does it need PyTorch?

No, the rewritten code uses ONNX or TensorRT for inference.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review