Will This Crazy New Technology from Nvidia Replace the Portrait Photographers of Tomorrow? (VIDEO)

The headline above may be a bit of a stretch, at least for now. But in the quick video below, you’ll see how new technology from Nvidia can create photorealistic faces, like those you see here, of people who don’t really exist.

Nvidia is known for solving many of the most complex problems in computer science, and they’ve conjured up a variety of virtual worlds for the film industry, gaming, and artificial intelligence applications like in this project.

As you can see, Nvidia has developed a new machine learning methodology, relying upon advanced algorithms, that can create fake human faces with remarkable detail and quality. The process begins with a large database of celebrity images. The photos in the database are then used to “train” the sophisticated AI system to make “portraits” of its own.

Nvidia’s technology begins with low-res files, and continues adding layers, with more and more detail, as the system continues to learn. The process accelerates until it generate images that are barely distinguishable from photos of actual people—even adding modifications to lighting, hair, and other subtleties that contribute to a realistic appearance.

After watching the video, be sure and look at the tutorial we posted yesterday, demonstrating how to use lens flare to create portraits with an artistic look.