Disney Created New AI System Helps to Change the Age of Actors
17 January 2023
Gizmodo reports that Disney researchers have developed a novel neural network that can alter the visual age of actors in television and film. The technology will make it possible for producers of television shows or movies to use an automated process that will take less time and cost less to make actors look older or younger.
Disney refers to this process as “re-aging,” and special effects crews typically use either a 3D scanning and 3D modeling process or a 2D frame-by-frame digital retouching of the actor’s face with Photoshop-like tools when they need to make an actor look older or younger. Depending on the length of the work, this procedure may take weeks or even longer.
On the other hand, Disney’s brand-new AI method, Face Re-aging Network (FRAN), automates the procedure.”The first practical, fully automatic, and production-ready method for re-aging faces in video images,” according to Disney.
Disney researchers used StyleGAN2 to randomly generate thousands of examples of synthetically aged faces ranging in age from 18 to 85 in order to build FRAN.FRAN gained a general understanding of how aging affects an individual’s appearance from that training data. It can apply those aging principles, frame by frame, to a real actor in motion now that training is finished.
They avoided the “seemingly impossible task” of collecting images depicting “a variety of identities, ages, and ethnicities in different viewpoints” by generating the training data synthetically.
Disney refers to the outcome as a “production ready” solution, meaning that it produces output of sufficient quality to be used in a real movie or television show. It may be the very first AI solution of its kind to dynamically alter an actor’s age in video despite different expressions, lighting, and perspectives. Additionally, the researchers created a user-friendly interface for FRAN that will make it simple for artists to utilize the tool in a production setting.
The research was presented by Disney on Wednesday in a paper titled “Production-Ready Face Re-Aging for Visual Effects,” and it was submitted for inclusion at the December 2022 ACM SIGGRAPH Asia conference. Gaspard Zoss, Prashanth Chandran, Eftychios Sifakis, Markus Gross, Paulo Gotardo, and Derek Bradley are the authors of the paper, all of whom are employed by Disney Research Studios in Zurich, Switzerland.
We would not be surprised to see technology similar to FRAN widely utilized in future Disney productions, despite the fact that no plans have been announced. Disney has a history of inserting computer-generated actors into Star Wars films and television shows, including some that have been de-aged using CGI.
The research group led by Derek Bradley, Gaspard Zoss, Prashanth Chandran, Eftychios Sifakis, Markus Gross, Paulo Gotardo, and Eftychios Sifakis calls their method “the first practical, fully automatic, and production-ready way to re-age faces in video images. “The project is based on the ever-increasing popularity of photorealistic digital aging of faces in video. However, the ones that are currently available in the industry appear to be time-consuming and lower the overall quality of the final product.
The research team explains that facial identity loss, poor resolution, and unstable results could all occur during the editing process. The ongoing work process requires the editors and specialists to physically deal with the video pictures, outline by-outline. Just to retouch the face of the actor or actress, they have to play and pause almost constantly. With FRAN, age can be magically added or subtracted from the appearance of screen stars with just one click.
According to the findings of their study, FRAN is the first re-aging system capable of providing high-resolution re-aging results on videos despite being temporarily stable. According to the video that the research team shared as part of their study, FRAN appears to seamlessly adjust the participants’ ages as they move their heads or change their facial expressions. In addition, the conditions of free viewpoint, motion, and illumination must be taken into account prior to achieving these results, according to the research team.
Disney’s research team is aware that there are still parts of FRAN that need to be tweaked and improved, but they see the AI tool as a useful tool that can help the advertising and entertainment industries. Beside re-maturing the screen stars – which may be important for a film’s brief – FRAN comes through as a computerized partner to video editors and specialists who currently have the likelihood to chop down their re-maturing responsibility to only a couple of hours or even minutes.