“Ma! They’re doing weird things on the internet again!”
You know when someone said that excess of everything is bad, this is specifically what they were referring to. AI has become so advanced in the past few years that it’s becoming increasingly difficult to differentiate photoshopped images from reality. An Israeli team of researchers has come up with a new and improved method of adding photorealistic changes to real photos using AI-based generative adversarial networks.
The working of the software is simple; the user is prompted to input a description of what they want added in the picture and the machine does the rest. The result is a weird and hilarious combination of the original picture along with the extra details you added. For example, you have a picture of a cat, you add the description which can be as small as a one-word feature such as “cute”. The result will be a cat with enlarged eyes which is a characteristic trait often associated with that specific word.
Up until now the cat example was a very “cute” one but researchers decided to add a bit of nightmare fuel into it and used a picture of Tesla and SpaceX’s CEO, Elon Musk. By adding all kinds of descriptions from a Karen-bob cut to putting makeup on the poor guy’s face, we got a never-seen-before look of Elon Musk (and we can’t stop staring simply because it’s just plain disturbing). Apparently, Scott Manley, a fellow space enthusiast, found the results a little too interesting….
There’s no argument that neural networks have come a long way and this face manipulation is one of the greatest examples of it. While it has a lighter side to it, this technology can be used in a lot of different applications and provide a beneficial outcome as well.