AI-Powered ‘Nightmare Machine’ Generates Horrifying Images

MIT researchers developed an algorithm trained to generate horrifying images in an attempt to find the scariest faces and locations possible, and then rely on humans to see which approach makes the freakiest images.

Using TITAN X GPUs and cuDNN to train their deep learning models, the researchers used the infamous style transfer technique and generative adversarial networks to curate the ghoulish images.

“We use state-of-the-art deep learning algorithms to learn what haunted houses, ghost towns or toxic cities look like,” said MIT Media Lab researcher Pinar Yanardag Delul. The algorithm extracts elements — such as a bruised-black palette — from scary templates and implants them in the landmarks.

colosseum
Watch the algorithm transforming the image

The researchers also used Google’s DeepDream method to develop ghastly portraits and voters can rate the images on the Nightmare Machine website to teach the system to make images even scarier.

Read more >

About Brad Nemire

Brad Nemire
Brad Nemire is on the Developer Marketing team and loves reading about all of the fascinating research being done by developers using NVIDIA GPUs. Reach out to Brad on Twitter @BradNemire and let him know how you’re using GPUs to accelerate your research. Brad graduated from San Diego State University and currently resides in San Jose, CA. Follow @BradNemire on Twitter