Disney's FRAN is a neural network that can age or de-age actors in five seconds

(Image Courtesy: Disney)

Disney's Face Re-Aging Network (FRAN) is a neural network that can make actors look older or younger in just five seconds. Artists can still manually adjust the images to make it more realistic but FRAN will help simplify the process.

To put things into context, the way de-aging or aging is achieved in film is time-consuming and laborious process. It requires artists to scrub footage frame-by-frame and manually adjust an actor's appearance in each.

The research team that built FRAN said that there have been various attempts to automate the process, but existing technology suffers, "from facial identity loss, poor resolution, and unstable results across subsequent video frames."

To combat this, Disney researchers first set out to solve the problem of big data sets. Using real-life faces to teach the neural network is nearly impossible, since that would require data to be collected over extended periods of time for just one face.

Instead the researchers leveraged existing re-aging technology to construct the data set. The researchers said that while the results failed at being, "real images," they do provide, "photoreal re-aging results on synthetic faces."

The researchers then leveraged this data set and fed it to a U-net Architecture to, "formulate facial re-aging as a practical image-to-image translation task."

The resulting network allowed them, "to advance the state of the art for re-aging real faces on video, with unprecedented temporal stability and preservation of facial identity across variable expressions, viewpoints, and lighting conditions."

Finally, the researchers stressed that FRAN also provides complete manual control to artists, allowing them, "to direct and fine-tune the re-aging effect."