Menu Close

Faceswap has released the windows installer here.

“This installer will install missing prerequisites (Git, MiniConda), set up the environment, install the correct Dlib, Cuda, cuDNN and Tensorflow versions and create a desktop shortcut for launching straight into the FaceSwap GUI. If running this installer, you do not need to manually install any programs yourself.

For GPU version, a CUDA 9.0+ capable graphics card is required.”

 

If you would like to install faceswap the manual way, proceed below.

For DeepFaceLab, go here.

 

Python Deepfake Faceswap Tutorial 한국판

Updated 2019/05/2

*Huge updates to the programs with additions of different models and configurations for this update. Changes are included in the folder structure,training and converting sections.

Welcome to our deepfake tutorial for the faceswap script based on Python.

In this tutorial, we will teach you how to make deepfakes of your own. Faceswap uses AI machine learning to process images from face sets. Face sets are gathered through videos and photos of the idol you want to create a deepfake of (dataB), and the actress you to replace/faceswap with (dataA). We will be using these machine learning images to create your first deepfake.

1. Prerequisites 

2. Python: Folder Structure

3. Python: Extraction – (DataB) Idol Data Collection/Faceset Creation

4. Python: Extraction – (DataA) Model Data Collection

5. Python: Training

6. Python: Converting


1. Prerequisites

System Requirements:

Graphics Card (GPU)

• Minimum: NVIDIA CUDA enabled + 6GB+ VRAM (980 Ti) + compute capability 5.0+ (Maxwell)

• Recommended: NVIDIA CUDA enabled + 8GB+ of VRAM (GTX 1070 (Ti), 1080 (Ti), Titan V, etc.) + compute capability 6.1+ (Pascal)

– To check your GPU capability, refer to this wiki

Computer Processor (CPU)

• Minimum: Dual Core, 2.5 Ghz+ (please do not get anything with this crap spec)

• Recommended: Quad Core, 3.5Ghz+ (e.g. i5 6600k)

– Really anything goes as long as you’re not bottle-necking your GPU

RAM

• Minimum: 8 GB RAM

• Recommended: 16 GB RAM

– If you’re processing a lot of 1080p vids, you would probably need 32GB, but make sure your CPU does not bottle-neck your ram as well

The better your hardware is, the more time you’ll save during the process.

Software Requirements:

CUDA 9.0

cudNN 7

ffmpeg

Python 3.6.7

CMake and Microsoft Visual Studio 2015

Python faceswap installation

Optional: Shift+Right click to open cmd in directory


2. Python Folder Structure


3. Python: Extraction – (DataB) Idol Data Collection/Faceset Creation


4. Python: Extraction – (DataA) Actress Data Collection


5. Python: Training


6. Python: Converting

Congratulations!

You’ve just created your first Deepfake.
If you want to start becoming a creator on our site, please send your finished videos to an admin on our Discord server for assessment.
Please include our unique Intro at the beginning and the Logo inside of the video as well!

Disclaimer: You are in full responsibility of the content you create. We at DFKP offer this tutorial for educational purposes, do this on your own risk.We’re not responsible for any kind of damages to your system or malicious content that you create.

Questions? Comments?

Go over to our forums if you have any questions or comments

Disclaimer:

We create deepfakes of celebrities mainly inclusive of K-pop Korean idols in South Korea. These videos are fake and are produced with software using machine AI learning. All models and faces of celebrities or idols used in our videos are at least 18+ years of age.