DeepFaceLab
Open-source face swapping pipeline for high-fidelity deepfake video.
Retired
DeepFaceLab appears to be discontinued or shut down — repo archived Nov 2024; community forks continue This page is kept as an archive record and is excluded from search engines.
About
DeepFaceLab is an open-source deepfake system created by iperov for face swapping in video. It provides a full local pipeline from face extraction through model training to final merge, and it earned more than 19,000 GitHub stars as the most widely used free face-swap toolkit.
The repository was archived read-only in November 2024, but the code and Windows builds remain available under GPL-3.0. It supports face replacement, de-aging, and full head swaps, and it is commonly used by VFX artists and researchers who want complete control over the pipeline.
What it can do
- Face extraction with S3FD detector
- Model training with SAEHD architecture
- Face replacement, de-aging, and head swap
- XSeg masking for precise edits
Setup
Download a Windows build, extract faces from the source and destination videos, train a model on your GPU, then merge the swapped face back into the footage.
Prerequisites
NVIDIA GPU with 8GB or more VRAM, Windows 10 or 11, and 16GB of RAM.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 131; Trending score 78/100 from researcher assessment.
★ 19,300 GitHub stars
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.