You may have seen a video of Nicholas Cage’s face transposed to another character playing the circles on the internet recently. No, we don’t mean a clip from the movie Face/Off. These are scenes from several movies that Nicholas Cage certainly didn’t star in, but they do look very convincing. How did they manage this?
Well, the answer isn’t “witchcraft”, but a…
technology called “deepfake” that is causing quite a stir in
different industries and communities.
What is a “Deepfake”?
The word “deepfake” comes from the term “deep”
learn” and of course “fake”. Deep learning is a specialized branch of
machine learning, which in turn is part of
the general field of artificial intelligence.
With the dramatic increase in computing power and
newly discovered ways in which computers use vast amounts of
real-world data, computers can now suddenly do things most of us could
never imagine. Deepfakes apply this technology to synthesize human images,
taking pictures or videos of things people never did and never said.
How does Deepfake technology work
Deep learning, which underlies deepfake
methods, describes the modern application of neural net simulation to massive
data sets. Neural nets are not a new concept or new technology, but up to modern
times they were pretty rudimentary.
An artificial neural net simulates the learning process that takes place in biological brains, at least to some extent. When you learn or otherwise interact with the outside world, the connections between your brain cells change.
They form circuits and logic structures, strengthening some connections and weakening others. As you master something, such as learning to drive a car or play tennis, those brain circuits become fast and efficient. After all, you’re so good at something that you don’t even have to think about doing it.
That is essentially the same as with a deep learning system. It looks at countless examples of something and then gets better and better at ‘understanding’.
In the case of deepfakes, the software looks at samples of the face you want to transpose and the video you want to transpose it to. With enough training, it can eventually synthesize a face that matches the training data and then seamlessly overlay it on another face.
What software is used to create?
There are a number of applications that make it possible:
people to make deepfakes. FakeApp was the first app we know is focused on giving
normal people a chance to create deepfake content. The website is now defunct and
finding a copy is not easy at all.
Deepfakes creators have now largely switched to an application called DeepFaceLabwhich is hosted on GitHub and has spawned an endless stream of tutorials on places like Reddit.
Create a deepfake
This article is not intended as a tutorial, so
we are going to give an overview of how deepfakes are made in practice, but
no exact steps to make your own.
There are a few reasons for this, but the main ones:
one is that the legality of creating deepfakes is highly controversial. just like us
would not provide exact steps to help you pirate software or do any other illegal
activities, we will not provide step-by-step instructions for creating deepfakes.
By the way, the actual creator of DeepFaceLab has a step-by-step video tutorial for the software that anyone can track should they want to take responsibility for such a risk.
To get an idea of how good this stuff has gotten, look at this website where a new photo of a person that doesn’t exist is generated when you refresh the page.
Our goal is to help you de
technology, as you will no doubt have to deal with it more and more over time.
That said, these are the broad stages of creating a deepfake with
Deep Face Lab.
After downloading and extracting DeepFaceLab
you are confronted with a folder with many other folders and a series of batch
There is a folder called “workspace” that
contains the training models, the source videos, and the output. DeepFaceLab
works with specific file names and locations so that the batch files can work.
For example, the source file is always named “data_src” and the destination is
There is no software interface as most people know it. Just a list of numbered batch files that represent the steps of the process. First, the source and destination videos extracted their frames as image files.
Then several steps of analysis are performed, followed by GPU-based training, where the neural network learns what it needs about the faces in the two videos to build a model. Finally, a new composite video is created.
Abuse of deepfakes
As we mentioned above, deepfakes are very
controversial. It may violate existing laws and new laws in some countries:
in the making to deal with the technology and its applications.
Deepfakes can be used to create hoaxes, such as a video of a president saying crazy things† It can be used to insert people into pornographic films for the purpose of harassing or otherwise harming them.
As you can probably imagine, a convincing deepfake can do a lot of damage and the consequences if you get caught could become much more serious in the near future.
Now that this technology exists and works as
well when it does it means we have to watch media like videos in a brand new
light. If someone circulates a video on social media of a famous person or
politician says or does something controversial, you should first ask yourself whether
the video is even real.
Most poorly executed deepfakes are clearly fake,
once you know what to look for and have seen a few. However, in some cases even
a trained eye may have trouble telling it is some kind of CG manipulation
happens and as technology improves, it eventually becomes impossible.