A recent student of mine lost out on a bid to a machine. Tech companies in Japan, Germany, etc have made AI capable of generating fake images of fashion models. The technology used to make them is called a Generative Adversarial Network (GAN), which pits two algorithms against each other. GANs have been used to make the technology behind “deep fake” videos. Below are some examples of how this technology is being used to replace real life models. Basically – photograph your clothing on a mannequin and upload. Voilà! -Erika
ORIGINAL ARTICLE on NEXT WEB.com
You’ve already seen how AI can be used to generate eerily realistic images of human faces, food, and even Airbnb homestays. Kyoto-based firm DataGrid is taking things a step further to creep you out with complete human bodies conjured afresh from images it’s gleaned.
That’s right – according to the company, the models you’re looking at in the image above and the video clip below aren’t real people. Each image has been dreamed up by DataGrid’s neural network from scratch.
DataGrid presented this demo without many details, but noted that its tech could find use in creating virtual models for advertising and displaying apparel. If it’s widely adopted, it could reduce the need for brands and online stores to invest in photo shoots to showcase their products.
The company previously developed an AI to spit out images resembling Japanese celebrities using what are known as generative adversarial networks (GANs). When you feed them a large amount of data, they can learn from it to create content similar to what it saw.
The technique was developed back in 2014, and it’s clearly come a long way since then. Tristan Greene has a simple explanation of how GANs work here, and you can delve into more details on this page.
Different outfits used to generate various model images, eg. the jacket from outfit #1 is added to outfit #2 to customize the visualization.
ERIKA GENTRY: Art & Education 