Disney has made a movie-quality AI tool that automatically makes actors look younger (or older).

To prove the power of artificial intelligence when it comes to changing shots realistically, Disney researchers She revealed a new aging/de-aging tool that can Make the actor convincingly look older or youngerwithout the need for weeks of complex and expensive visual effects work.

When watching a blockbuster like 2018 Ant-Man and the WaspMost viewers can easily figure out the work of the many visual effects studios that contribute to these films, what happens with their flashy moments when Ant-Man shrinks or grows to gigantic proportions. But sometimes the more subtle VFX work is the hardest to pull off realistic results, like shots featuring younger versions of actors Michelle Pfeiffer and Michael Douglas. To get results like those seen in the movie, the talented artists either need to spend weeks erasing wrinkles and other signs of age from the actor’s face, or replace them entirely with a computer-generated double.

Visual effects are a powerful tool for filmmaking, but there are many reasons to find ways to make them easier to create; From Reducing the burden on underpaid and underpaid artists, to make the tools more accessible to filmmakers who don’t work on huge Hollywood-sized budgets. Of course, even for the big studios, there’s a profit driver in being able to automate this kind of work, too.

That’s why companies like Disney are investing in research to help advance visual effects art, but in recent years these researchers have also been exploring how artificial intelligence can simplify visual effects work. Two years ago, Disney Research Studios developed AI-powered tools It can generate face swap videos With enough quality and resolution to be used in professional filmmaking (instead of questionably low-resolution GIFs being shared online). This year, researchers are showing off a new tool that takes advantage of artificial intelligence tricks to make actors look older or younger, barring the weeks of work usually needed to perfect those kinds of shots.

Using neural networks and machine learning to reduce or eliminate a person’s lifespan has already been experimented with, and while the results are convincing enough when applied to still images, they don’t produce realistic results when moving video, with temporal artifacts appearing and disappearing from frame to frame. , and the person’s appearance sometimes becomes unrecognizable while playing the changed video.

To make an age-adjusted AI tool that was ready for Hollywood’s demands and flexible enough to work on moving shots or shots where the actor isn’t always looking directly into the camera, according to Disney researchers, Detailed in a recently published paper, first created a database of thousands of randomly generated synthetic faces. Existing machine learning tools were then used to age and de-age these thousands of non-existent test subjects, and these results were then used to train a new neural network called FRAN (Face Re-Aging Network).

When FRAN is fed an input header, instead of generating a changing image of the header, it predicts which parts of the face will be altered by age, such as adding or removing wrinkles, and then superimposes those results on top of the original face as a conduit for added visual information. This approach meticulously preserves the artist’s appearance and identity, even as his head moves, when his face looks around, or when the lighting conditions in a shot change over time. It also allows AI-generated changes to be tweaked and tweaked by an artist, which is an important part of VFX work: making the adjustments blend so perfectly back into the shot that the changes are invisible to the audience.