This Website Lets You Turn Gaming Characters Into Nightmare Fuel


This Website Lets You Turn Gaming Characters Into Nightmare Fuel

Have you ever finished a doodle of Crash Bandicoot or Sonic the Hedgehog and thought to yourself “oh boy, I really wish I could see this doodle rendered with realistic textures.”

If you have, then I’ve got a bit of a bad news, good news situation for you. The bad news is that there’s something seriously wrong with you. The good news is that thanks to an image-to-image translation demo over on this website here, your dreams can finally come true – but be aware that the results might not be what you wished for, as you can see from our Crash sketch above, or our Toad sketch below.

We have Pushbullet co-founder Christopher Hesse to thank for this tool, which, in Hesse’s own words “works by training on pairs of images such as building facade labels to building facades, and then attempts to generate the corresponding output image from any input image you give it.”

There are a few different examples to play with on the site; handbag, cat, and buildings. Obviously, attempting to draw something that isn’t a cat (like a video game character) using the cat tool, which was trained on around 2,000 stock images of felines, will produce a freakish abomination, since the AI is trying to work out how to make Crash Bandicoot or Toad look like a cat.

Obviously, drawing a cat, shoe, or handbag in the corresponding box will produce the best results, but if you’re just looking for a bit of anarchic fun, use it to draw your favourite gaming characters and marvel at the hideous results.

Below you’ll find some of the creations UNILAD Gaming wasted a good chunk of the afternoon on. Enjoy! Oh and before you say anything, yes we are professional artists – obviously.

Hesse noted that even when using the tool properly instead of trying to break it like we did, the results aren’t always perfect.

He wrote:

Some of the pictures look especially creepy, I think because it’s easier to notice when an animal looks wrong, especially around the eyes. The auto-detected edges are not very good and in many cases didn’t detect the cat’s eyes, making it a bit worse for training the image translation model.

Still, it’s as good a way as any to waste an hour. I recommend checking it out and drawing as many cats/Marios/Sonics as you want when you get a spare minute.

If you have a story you want to tell, send it to UNILAD via [email protected]