Recently I saw some video like these :

Rick Astley singing Tokyo Ghoul - Unravel

Yagoo singing Hololive - Shiny Smily Story

I can’t help laughing and I want to explore the technology behind it.

Deepfake - First Order Model :

Then I found a video called “Everybody Can Make Deepfakes Now!”.

In the video, it stated the source code and even an example colab notebook is available, and it is one of the most accessible paper in this area.

https://github.com/AliaksandrSiarohin/first-order-model

I can’t wait to try running my own experiments.

Here are the Deepfake videos I made:

Gurenge (From 鬼滅之刃)

Unravel (From 東京喰種)

Inferno (From 炎炎消防隊)

Why spread this knowledge, especially now, with the source codes?
The Video Author think It is to inform the public that these DeepFakes can now be created quickly and inexpensively, and they don’t require a trained scientist anymore. If this can be done, it is of utmost importance that we all know about it.

Tutorials

Here are tutorials of how to make your own deepfake video:

Main Points :

The Demo Link with Code
Tutorial Code

In order to get best results :

your source image must be 256x256 size
If your using your own video than you must crop your video using face-alignment module mentioned above, the size of that video will be 256x256 by default
Align the your image with driving video first frame, you can using any video editor to do this, otherwise your resultant video would be massy and all over the place…