← Back to Discover

Deep-Live-Cam

An open-source app that swaps a face in live webcam video, or in saved images and videos, using a single source photo, run from a desktop interface or the command line.

Video editingImage editing
Visit website ↗

What it does

Deep-Live-Cam is an open-source project for real-time face swapping and video deepfakes from one source image. In webcam mode you pick a source face, press Live, and wait for the preview to appear, which the README puts at 10 to 30 seconds. To stream the result it suggests a screen capture tool such as OBS. In image and video mode you choose a source face and a target image or video, and the output is saved in a folder named after the target. Passing a source argument runs it in command-line mode. Among its options are a mouth mask that keeps your original mouth for more accurate movement, face mapping to apply different faces to several subjects at once, and a many-faces setting that processes every face. The README lists live shows, meme making and watching films with a different face among its uses. Manual installation takes technical skill: Python 3.11 to 3.14, pip, git, ffmpeg and two model files, with roughly 300MB of models downloaded on first run. It can run on CPU, and GPU acceleration is documented for NVIDIA (CUDA), Apple Silicon (CoreML), Windows (DirectML) and Intel (OpenVINO). The page also advertises pre-built Ultimate builds for Windows, Mac Silicon, CPU, NVIDIA and AMD, with extra features and priority support, and names deeplivecam.net as the only official website. The code is under the AGPL-3.0 license. The project says it blocks inappropriate media, expects users to get consent before using a real person's face and to label shared output as a deepfake, and may add watermarks or shut down if the law requires. Its credits note that the insightface models it relies on are for non-commercial research use only.

Capabilities

  • Real-time face swap on a webcam feed from a single source image
  • Image and video mode that writes the swapped result to a folder
  • Mouth mask, face mapping and many-faces options
  • GPU acceleration for CUDA, CoreML, DirectML and OpenVINO, or CPU only
  • Built-in checks that stop processing of inappropriate media

When to use it

  • You want to experiment with live face swapping on your own machine
  • You make video content and have the consent of the people whose faces you use
  • You are comfortable setting up Python tools, or want a pre-built version