UNC AI Lab Revolutionizes Facial Aging for Film and Forensics
At the University of North Carolina's Spatial & Physical Intelligence (SPIN) Lab, researchers are delving into the future of digital identity with generative facial editing AI models. This innovative work, led by a team of students, faculty, and alumni, is making it possible to visualize how individuals might look years or even decades from now.
Pioneering Facial Aging with Generative AI
Overseen by computer science assistant professor Roni Sengupta, the SPIN Lab's primary focus is on realistically aging human faces in photos and videos. The goal is to ensure the resulting features are as accurate as possible. The lab’s most recent project can generate what a person might look like at 70 years old, factoring in variables such as whether they followed a good skincare routine or experienced significant weight changes over their lifetime.
Transforming Visual Effects in Filmmaking
Much of the research at the SPIN Lab is driven by the need for advanced visual effects tools in the film industry. The technology is designed to modify specific aspects of a person’s face, like aging them, while meticulously preserving other key features. Sengupta emphasizes that a major objective is to maintain the original style of the image or video. For a creative director to have complete control, the AI model must be able to preserve elements of a character's appearance, including their clothing and jewelry, ensuring stylistic consistency.
A New Tool for Aiding the Missing
Beyond Hollywood, this generative AI technology has profound real-world applications. It can be used by the general public to edit photos of missing or deceased individuals, offering a glimpse of what they might look like today. Sengupta noted that families have used similar technology, bringing in photos of children who went missing at a young age to generate age-progressed images that could aid in their search.
Future Ambitions and Current Limitations
Looking ahead, Sengupta hopes to develop the technology for forensic research. However, she acknowledges that the algorithms are not yet robust enough to be used by law enforcement in active searches for missing children. "That definitely requires more experiment, more validation," Sengupta said. She explained that the goal is not to produce one single outcome but several possibilities. "That's what The [Aging] Multiverse project comes into," she added, referring to the project's ability to generate multiple potential aged appearances.
UNC students and UNC Assistant Professor Roni Sengupta at the SPIN lab on Sept. 16, 2025.
Hands-On Research for the Next Generation
The SPIN Lab also serves as a critical training ground for students. UNC senior Amisha Wadhwa, who worked in the lab over the summer, focused on a project for detecting deepfakes and understanding their generation. She said the experience provided her with valuable technical skills that are directly applicable to her future career in the AI industry.
