Disney De-ages Actors in Seconds With Their New AI Tool
Note when it was such a big deal to make actors look older or younger in films? Back in the day, the amount of postproduction work required to achieve realistic results was enormous, but now Disney researchers have unveiled FRAN, a brand-new artificial intelligence tool that can compellingly age or de-age an actor in a fraction of time.
Disney Research Studios explains in an academic paper that FRAN (which stands for face re-ageing network) is a neural network that has been trained using a large database containing pairs of randomly generated synthetic faces at varying ages, avoiding the need to search thousands of images of real individuals of various (documented) ages that depict the same facial expression, pose, lighting, and
background.
FRAN uses this data to predict which areas of a real person’s face will age and how, and then overlays the new details, such as adding or removing wrinkles and jowls, onto video footage. As a result, Disney Research Studios claims to have developed the first practical, fully automated, and production-ready method for re-ageing faces in video images.
There are a few limitations, however, and this sort of research isn’t unique. Disney noted in its research that FRAN may be unsuitable for significant alterations such as However, there are some limitations, and this type of research is not unique. According to Disney’s research, FRAN could be unsuitable for major changes such as re-ageing to and from very young ages, and greying of scalp hair isn’t reflected when ageing up an actor because it wasn’t present in the dataset that was used to train the tool. Given that manual VFX work and even practical prosthetic makeup application are not subject to these constraints, FRAN is unlikely to replace many industry jobs for a long time.
Despite its potential benefits in filmmaking, it’s unclear whether Disney intends to make this technology available to the general public, and there’s still plenty of room for improvement, so it may be a while before we see this level of elaborate visual effects work essentially automated within the industry.