Just a few months ago, the generation of fake photorealistic faces through generative adversarial networks (GANs) was one example of the use of artificial intelligence (AI) that was increasingly holding back the attention of the general public. Today it was all just an aperitif of what was to come. Now faces are not enough anymore, because artificial intelligence is now able to generate, out of nothing, photorealistic images of people with the whole body, of course, of people who do not exist, but who have their own. faces, hairstyles and hair. clothes, in a nutshell False Humans.
Create fake humans who never existed in the real world
Therefore, to show that, here we have a video in which we can see how the characters generated by the GAN are transformed as they move:
Like many other advances in artificial intelligence, this approach has both a positive and a negative side, as it is one more step towards eroding the credibility of multimedia content, which is the source of increasingly unreliable information thanks to progressively more realistic in-depth analysis. Plus, apart from all that, it offers a fast and inexpensive (mostly) system for producing clothing catalogs.
The algorithm that made this possible was developed by DataGrid, a tech startup based on the Kyoto University campus in Japan. Its engineers decided to overcome the problems faced by generating GAN faces, avoiding including much of the unnecessary information that could confuse artificial intelligence.
For that, they chose to place on a homogeneous white background the models posed on the photos used to form the system. Therefore, now we can say goodbye to any asymmetrical elements or weird color spots that might highlight the nature of the generated images.
In DataGrid, they had already succeeded in developing their own face generation algorithm, but they were looking for ways to give their digital models greater ‘expressiveness’, for which they bet not only to generate whole bodies, but also to give them an idea of what they are doing. movement as well.
So what do you think about this? Just share all of your views and thoughts in the comment section below. And if you liked this article, don’t forget to share it with your friends and family.