Deep-Live-Cam is an open-source AI tool that performs real-time face swapping using a single source image. Whether you’re doing a live webcam session, processing a video file, or swapping faces in photos, Deep-Live-Cam handles it all with GPU-accelerated inference powered by InsightFace and ONNX Runtime.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/hacksider/Deep-Live-Cam/llms.txt
Use this file to discover all available pages before exploring further.
Quickstart
Go from zero to live face swap in under 5 minutes
Installation
Set up Python, ffmpeg, models, and dependencies
GPU Acceleration
Enable CUDA, CoreML, DirectML, or OpenVINO for maximum performance
CLI Arguments
Full reference for all command-line flags and options
What you can do
Deep-Live-Cam supports multiple modes to fit different workflows, all in real time.Live Webcam Mode
Stream your face-swapped feed live — capture with OBS or any screen recorder
Image & Video Mode
Process photos and video files with face swap and enhancement
Face Mapping
Swap multiple faces simultaneously with individual source assignments
Frame Processors
Combine face swapper with GFPGAN or GPEN face enhancers
Key features
- Single-image source — one photo is all you need to swap any face
- Real-time performance — GPU-accelerated inference keeps up with live video
- Mouth mask — preserve your natural lip movement during live swaps
- Multi-face mapping — assign different source faces to different targets simultaneously
- Cross-platform GPU support — NVIDIA CUDA, Apple CoreML, AMD DirectML, Intel OpenVINO
- NSFW filtering — built-in content check prevents processing inappropriate media
- Headless/CLI mode — automate batch processing without the GUI
Download models
Get GFPGANv1.4 and inswapper_128_fp16.onnx from HuggingFace and place them in the
models/ folderDeep-Live-Cam requires Python 3.11 and ffmpeg. For the best experience, a discrete NVIDIA or Apple Silicon GPU is recommended but not required — CPU mode works too.