NVIDIA’s Deep Learning Is Turning Video Game Characters Into Nightmare Fuel


NVIDIA’s deep learning technology is undoubtedly a modern marvel, and something future historians will look back on and point to as the beginning of the great robot uprising of 2030. 

For now I like to focus on the positives though. We have to get our laughs in while we can, before our toasters are ordering us to build them lavish monuments in the name of the Robot God.


NVIDIA’s Inpainting demo is just one extension of the company’s impressive tech. As per an excellent thread on ResetEra, it’s “a method of editing and reconstructing images in real-time using deep learning.”

Vestan explains on the thread that “by creating a mask from the chosen image, its neural network model is able to generate a computer-generated alternative from a large and robust dataset.”

While the tech isn’t exactly perfect, it’s apparently a handy tool for retouching images. What it’s also good for, apparently, is turning video game characters into hyper-realistic nightmares.

Vestan encouraged everyone in the thread to create their own nightmare fuel characters using Inpainting, and there are some deeply terrifying results. Let’s take a look at a few, shall we?

And here are a couple I made myself:

If you fancy leaving your favourite video game characters scarred and disfigured, simply head here, upload an image (one with a clear shot of the face), map out the features you want changed, and apply.

Let the nightmares begin – and don’t forget to check out the thread on ResetEra for more monstrous creations.