Posts
NVIDIA Neural Network Generates Photorealistic Faces With Disturbingly Natural Results
Imagine playing a game like Skyrim or a sports title where the characters you encounter look like real people or creatures, and not rendered graphics. If implemented correctly, it could add a new level of immersion to gaming. It may not be far off, either. NVIDIA released a paper over the weekend detailing a new training methodology for generating unique and realistic looking faces using a generative adversarial network (GAN).
The result is the ability to render photorealistic faces of "unprecedented quality." How NVIDIA achieves this is by using an algorithm that pairs two neural networks—a generator and a discriminator—that compete against each other. The generator starts from a low resolution image and builds upon it, while the discriminator assesses the results, sort of like a constant critic pointing out where things have gone wrong or off track.
GAN in and of itself is not a new technology, but where NVIDIA differentiates itself is through a progressive training method it developed. NVIDIA took a database of photographs of famous people and used that to train its system. By working together, the neural networks were able to produce fake images that are nearly indistinguishable from real human photographs. Here is a look at the process:
"We describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively, starting from low-resolution images, and add new layers that deal with higher resolution details as the training progresses. This greatly stabilizes the training and allows us to produce images of unprecedented quality, e.g., CelebA images at 1024² resolution. We also propose a simple way to increase the variation in generated images, and achieve a record inception score of 8.80 in unsupervised CIFAR10," NVIDIA explains.
There are issues with NVIDIA's method, one of them being the relatively small size of the images. Warping and other abnormalities tend to occur as well. But it is still promising, with plenty of real-world applications ranging from content creation to video games. There is also the potential for abuse, such as upping the fake news ante, but that is a topic for another day.
image: https://hothardware.com/ContentImages/NewsItem/42578/content/NVIDIA_Photorealistic_Faces.jpg
Everyone of these faces are machine-rendered...
Everyone of these faces are machine-rendered...
The result is the ability to render photorealistic faces of "unprecedented quality." How NVIDIA achieves this is by using an algorithm that pairs two neural networks—a generator and a discriminator—that compete against each other. The generator starts from a low resolution image and builds upon it, while the discriminator assesses the results, sort of like a constant critic pointing out where things have gone wrong or off track.
GAN in and of itself is not a new technology, but where NVIDIA differentiates itself is through a progressive training method it developed. NVIDIA took a database of photographs of famous people and used that to train its system. By working together, the neural networks were able to produce fake images that are nearly indistinguishable from real human photographs. Here is a look at the process:
"We describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively, starting from low-resolution images, and add new layers that deal with higher resolution details as the training progresses. This greatly stabilizes the training and allows us to produce images of unprecedented quality, e.g., CelebA images at 1024² resolution. We also propose a simple way to increase the variation in generated images, and achieve a record inception score of 8.80 in unsupervised CIFAR10," NVIDIA explains.
There are issues with NVIDIA's method, one of them being the relatively small size of the images. Warping and other abnormalities tend to occur as well. But it is still promising, with plenty of real-world applications ranging from content creation to video games. There is also the potential for abuse, such as upping the fake news ante, but that is a topic for another day.
Read more at https://hothardware.com/news/nvidia-neural-network-generates-photorealistic-faces-disturbing-results#d0JjghQ8mcBFbCKK.99
