Player not loading? Watch on YouTube
This tutorial demonstrates FaceFusion 3.8 through Pinokio on a Windows computer. The speaker walks through downloading Pinokio, finding FaceFusion in its app browser, installing it, and activating the local server. Pinokio downloads are available for Windows, macOS, and Linux, though the demonstration uses Windows.
After launching FaceFusion, the speaker opens the server URL in a browser. A source image supplies the replacement face, while a target video supplies the footage. The interface shows a preview before the speaker starts processing. This is a local AI workflow that uses the computer's own resources; the tutorial describes it as free, with no subscription or generation credits required.
The demonstration uses CUDA acceleration because the speaker has a compatible device. Reference mode targets a selected face. Switching to many mode applies the source face to multiple detected human faces. In the example, the model leaves a dog's face unchanged. The speaker then previews the processed video and shows where to download it.
The opening narration recommends 8–12 GB of VRAM. The closing narration instead mentions 8 GB of RAM as a guess, so the video does not establish a consistent minimum hardware requirement.
