FaceFusion setup with Python 3.11 and CUDA 11.8 or 12.8

This FaceFusion tutorial walks through a Windows installation for local AI face swapping. It opens with a batch-file launcher and a brief image-swap preview using the face enhancer, then moves into dependency checks and installation. The presenter notes that the first preview downloads models before it can finish.

The demonstrated environment uses Python 3.11.9 and Visual Studio 2022. The prerequisite section also checks Git, FFmpeg and the installed CUDA build. Although the presenter mentions CUDA 12.4, the requirements-file instructions specifically cover CUDA 11.8 and 12.8. The description links separate files for those two versions.

The workflow covers cloning the repository, creating and activating a Python virtual environment, editing the matching requirements file, and installing dependencies with pip. For the GPU setup, the presenter changes ONNX Runtime to ONNX Runtime GPU and sets version 1.21.0. He tells viewers without NVIDIA graphics to use the unedited requirements instead; this walkthrough does not establish performance on those machines.

After installation, the presenter starts FaceFusion and copies its local address into a browser. Initial startup requires additional downloads, so this is not an offline setup procedure. Detailed face-swapping controls and WhatsApp live-call use are left to separate videos.