In a study titled “StyleCLIP: Text-Driven Manipulation of StyleGAN Imagery,” a team of researchers from Israel reviled that they have found a way to make photorealistic changes to photos by using an AI-based generative adversarial network (GAN). And as reported by Futurism, the only thing users have to do is to put the description of the edit they want to be done.
For example, users can turn an image of a “cat” into a “cute cat” by simply adding “cute cat” in the description. Coming on makeup by simply typing “with makeup” or “without makeup” we can edit a particular image. To see few examples click here and if you want to try it do visit the official GitHub page here.
Well, researchers didn’t stop here and taking some interesting test subjects gave Elon Musk and Mark Zuckerberg a series of makeovers by using the same technology.
Photorealistic Changes to Elon Musk
Researchers edited Elon Musk’s pic several times by applying different hairstyles of fringe hair, bob cut, curly hair, straight hair. Other pictures showed him with tanned skin, pale skin, and even makeup in one photo.
Here have a look at these photorealistic changes to one of Elon Musk’s photos by yourself:
Photorealistic Changes to Mark Zuckerberg
With Tesla CEO, researchers also applied their makeovers to Facebook CEO Mark Zuckerberg by giving him a beard, putting glasses on his face, made him bald, etc.
These edits are not only for people’s photographs but also work on other photographs such as a photo of buildings that can be edited by simply typing the change needed in the description.
By the way, these edits are remarkable. Neural networks are becoming more democratized in recent years by allowing everyone with PCs to have a go for themselves.