According to Wikipedia, ‘Deepfakes (a portmanteau of “deep learning” and “fake“) are synthetic media in which a person in an existing image or video is replaced with someone else’s likeness.’
Using deep learning, videos can be manipulated to change the faces of the people in a video. Here is an example of a famous deep fake video by Buzzfeed:
This article is a simple how-to guide that directs you to the right resources for creating a deep fake video from your home.
Create a DeepFake using DeepFaceLab
DeepFaceLab is a popular software for creating deepfakes. You can check out their GitHub repository here.
Once you head onto the repository, go to the releases section of the README.md file and open the link according to your current operating system.
Follow this detailed video tutorial from Cinecom on how to use DeepFaceLab
Once you have downloaded the executable file, we suggest following this tutorial from Cinecom to create your deep fake video:
The tutorial is 11 minutes long (minus the MSI advertisement) and it takes you to a step by step process on how to create your deepfake video.
In the video, Cinecom takes a home alone clip and uses his own video to swap out the faces in the original clip. There are a total of 10 batch files in DeepFaceLab and the video takes you through the entire process of how to execute them.
Although the model training may take some time, you can always continue to train the model from where you left off.
How did you find the deepfake tutorial from Cinecom? Please let us know in the comments if you get stuck somewhere so that we can help you out.
Also, a recent update by the DeepFaceLab developers shows a new executable file in the Google Drive link. We will try to cover it as soon as possible.