What is a Deepfake and How Are They Made?

You may have seen the video in which Nicolas Cage’s face is transferred to another character that has recently appeared on the Internet. No, we don’t mean the Face / Off movie clip. These are scenes from various films in which Nicolas Cage did not exactly play, but they look really very convincing. How did they do it?

Well, the answer is not “witchcraft,” but technology that has been dubbed “deepfake”, and it is causing quite a stir in various industries and communities.

What is deepfakes?

The word “deepfake” comes from the term “deep learning” and, of course, “fake”. Deep learning is a specialized branch of machine learning that is again part of the general field of artificial intelligence.

With the surge in computing power and the recently discovered ways of computers to process and analyze huge amounts of data from the real world, computers can now suddenly do things that most of us could not even imagine. Deepfakes uses this technology to synthesize human images, create photos or videos of things that these people have never done and never said.

How does deepfake technology work?

Deep learning, the underlying foundation of deepfake techniques, describes the current application of neural network modeling to massive datasets. Neural networks are not a new concept or technology, but they have been pretty rudimentary until now.

An artificial neural network mimics the learning process that occurs in a biological brain, at least to some extent. When you study or otherwise communicate with the outside world, the connections between your brain cells change.

They form patterns and logical structures, strengthening some connections and weakening others. When you master something, such as learning to drive a car or play tennis, these brain circuits become fast and efficient. After all, you’re doing something so well that you don’t even have to think about it.

Essentially, this is the same as what happens with a deep learning system. He looks at a bunch of examples of something, and then gradually it becomes better to “understand it.”

In the case of deepfakes, the program looks at examples of the face you want to transpose as well as the video you want to transpose it to. With enough training, it can eventually synthesize a face that matches the training data and then easily superimpose it on any other face.

What software are used to create deepfakes?

There are a number of deepfake apps available. FakeApp was the first application we know of, aimed at empowering normal people to create deepfake content. A website that doesn’t exist now and it’s not easy to find a copy.

Deepfake creators have now mostly moved to the DeepFaceLab app, which is hosted on GitHub and has spawned an endless stream of tutorials on places like Reddit.

Making deepfakes

This article is not intended to be used as a tutorial, so we are going to provide an overview of how deepfakes are made in practice, but we will not provide exact instructions on how to make them yourself.

There are several reasons for this, but the main one is that the legality of creating deepfakes is highly controversial. In the same way that we do not provide precise instructions to help you use pirated software or engage in other illegal activities, we do not provide step-by-step instructions for creating deepfakes.

In addition, the actual creator of DeepFaceLab has released a step-by-step video tutorial for the software that anyone can follow if they choose to take responsibility for such risk.

To get an idea of ??how good this material is, visit this website, where a page refresh creates a new photo of a non-existent person.

Our goal is to help you understand the technology as you will surely come across it more and more over time. With that said, these are the main steps involved in creating a deepfake with DeepFaceLab

After downloading and unpacking DeepFaceLab, you will come across a folder containing many other folders and a number of batch files.

There is a folder called “workspace” that contains training models, source videos and results. DeepFaceLab works with specific filenames and locations, so batch files can work. For example, the source file is always named “data_src” and the destination is “data_dst”.

There is no programming interface that most people know. Just a list of numbered batch files representing the steps in the process. First, frames of the source and destination video are extracted as image files.

Several stages of analysis are then performed, followed by GPU training, in which the neural network learns what it needs about the faces in two videos to build the model. Finally a new composite video is created.

Misuse of deepfakes technology

As we mentioned above, deepfakes are very controversial. This may violate existing laws in some countries, and new laws are being developed regarding the technology and its applications.

Deepfakes can be used to create hoaxes such as videos of the president saying crazy things. It can be used to insert people into pornographic films for the purpose or harassment or otherwise harm them.

As you probably guessed, you can do a lot of damage with a convincing deepfake, and the consequences if you get caught could become much more serious in the near future.

Ask about everything

Now that this technology exists and works as well as it did before, it means we need to look at media like video in a whole new light. If someone is spreading a video on social media of a famous person or politician saying or doing something controversial, you should first ask if the video is even real.

Most poorly made deepfakes are obviously fakes if you know what to look for and have seen a few. However, in some cases, even a trained eye may find it difficult to tell that there is some kind of manipulation of computer graphics, and as technology improves, this will eventually become impossible.

–