There’s a long list of things wrong with the Justice League movie, not the least of which being the hasty and poorly-executed digital removal of Henry Cavill’s moustache that he couldn’t shave during the film’s reshoots. But not to worry, some random dude on the internet with a $500 used PC and a world-changing AI just fixed at least one part of that film.
If you haven’t been checking the news, over the past few weeks it’s come to light that artificial intelligence researchers have made incredible leaps and bounds when it comes to applying deep learning technology to manipulating images and video. It’s reached the point where easily accessible programs running on an average desktop PC can swap the face of one performer with another, with staggeringly lifelike results.
In order to pull off these face-swaps you need to collect and train an AI using thousands of images of the person you want to swap in, but online scraping tools are able to easily access a near unlimited supply of celebrity headshots, such as those of Henry Cavill, thanks to search engines such as Google. The results aren’t instantaneous, but a few days of intense CPU processing later, some guy working on a run-of-the-mill Windows PC is now able to pull off Hollywood-calibre visual effects tricks, such as removing Cavill’s moustache without making him look like a monster in the process.
The less-than-scrupulous parts of the internet have already embraced these AI tools to create fake porn videos of their favourite celebrities, which is understandably causing a lot of concern and scandal. It will soon be impossible to believe anything you see online. But at the same time, these tools are going to completely revolutionise the visual effects industry, allowing filmmakers without a Hollywood-sized budget to produce FX-laden movies that can easily compete with those coming out of the big studios.
If you run a visual effects studio and aren’t at least paying attention to these recent developments in deep learning and AI image manipulation tools, let alone working to adopt them into your pipeline, you’re missing out on a revolution that could soon put you out of work.