This AI Generating Faces That Does Not Exist In Real World

AI generating fake faces

The above-featured image shows faces of different people who are not real and doesn’t exist. And the fact is they were created by an artificial intelligence engine. Yes, all these millions of faces are generated automatically in this new website that carries the successful name of This person does not‘.

Yes, you can visit this website to believe this magic. The website will show you a new face until we get tired. But how it happens? Its all about an AI algorithm is known as Generative Adversarial Networks (GAN).

True Faces Of False People

Nvidia already showed the capabilities of this GAN. In fact, they are the developers of this algorithm and today are also its main drivers. The man behind the web of ‘This person does not exist’ is Phillip Wang, who decided to set up the site to show the capabilities of the GAN. Wang developed a new code based on the Nvidia algorithm, which he named StyleGAN, and uploaded it to the web to give life to this site.

According to the details of StyleGAN, Wang explains that it is a new demonstration of how simple it can be to create “people” that are not real, something we have seen for example in the ‘virtual influencers ‘, and whose possibilities They are very, very spacious.

What are GANs and where they come from ?

The concept of ‘Generative Adversarial Networks’ was born in 2014 when Google scientist Ian Goodfellow develops the idea, which is later collected by Tero Karras, one of the leading scientists and researchers at Nvidia.

In the beginning, GANs were not able to generate images that exceeded 1024 pixels. As of 2017, and after improving the training methods, Nvidia achieves that the GANs increase their capacities by reducing the processing time and manage to create images of higher resolution and with greater realism. The GANs are composed of two types of neural networks: the generative and the discriminatory, which compete with each other millions of times to perfect the skills of generating images.

But Nvidia is not the only one that is working in the GAN, since today companies such as Google and Facebook are developing their own algorithms based on this Nvidia research, so in the coming years improvements in technology are expected, not only cover the face, but the whole body and even the location, creating scenarios that would seem familiar but not real.

You, think this is Fascinating? Then, let us know ur opinions……

Leave a Reply