Disney’s new aging tool is ready for the movie world: meet FRAN!

Disney's new AI powered aging tool, FRAN, is ready to take part in future movies and TV shows.
 Disney’s new aging tool is ready for the movie world: meet FRAN!
READING NOW Disney’s new aging tool is ready for the movie world: meet FRAN!

Visual effects experts at Disney say they’ve developed artificial intelligence technology that can visually age and age players faster, cheaper, and more accurately than any available method. In a published research paper, the team described the Face Re-aging Network, or FRAN, as “the first practical, fully automated, production-ready method for re-aging faces.” The researchers add that the industry standard (using 2D painting) to artificially age artists requires “manual work by frame-by-frame, which can take days to achieve.”

They also note that existing methods for automating the process “typically suffer from loss of face ID, low resolution, and unstable results in subsequent video frames.”

And the effort hasn’t been easy or always successful: The Irishman, which starred Robert De Niro, Al Pacino and Joe Pesci young in 2019, was outrageously expensive, Engadget reported. And for 2010’s Tron: Legacy, Jeff Bridges’ attempt to shrink his age by more than a quarter of a century has been described as “funny.”

The main problem, the researchers write, is that collecting enough images of aging humans to successfully train an AI is an “almost impossible” endeavor.

So they used StyleGAN2, a program that creates artificial faces, to generate thousands of samples of people ages 18 to 85. With this data, they were able to develop algorithms for how appearances typically change over time, such as where wrinkles appear and how chins begin to sag.

The team said the simple and intuitive technology “give artists localized control and creative freedom to navigate and fine-tune the aging effect.” They believe that FRAN is good enough to be used immediately; We look forward to the first movie that will use this technology.

Comments
Leave a Comment

Details
203 read
okunma21528