Aging apps have been all the craze this week, with FaceApp’s aging filter being at the forefront of the trend. In the past couple of days, every celebrity from Ludacris to Jonas Brothers has given us a sneak peek into their biological future. We now know what they might look like when they’re in their golden years, all with the power of artificial intelligence. But with that in mind, just how has FaceApp managed to transform celebrities (and everyone else, for that matter) into their elderly versions? The answer may somewhat surprise you. Although the app is definitely powered by AI, it is also informed by our understanding of the biology related to aging.
Back in 2017, the first version of FaceApp was released and designed by Wireless Lab, a Russian company. Though the first version was fairly rudimentary and mostly related to facial hair filters, the newest updates have expanded its aging algorithm by quite a lot. That is visible from geriatric versions of particularly youthful celebrities, such as Tom Holland.
With that in mind, it’s interesting to consider what new technology FaceApp has used in its newer iterations in order to achieve such a dramatically realistic effect. The explanation that the company gives on its website is expectedly vague. The design team simply states that the app will “add a couple of wrinkles to someone’s face.” But even a cursory glance at the results that the app achieves reveals that it does more than that.
While FaceApp has remained notoriously tight-lipped regarding the details of their software solutions, we do know that the basis of it is formed by a neural network, which is a sort of artificial intelligence.
Nonetheless, different scientists have been devoted to the study of facial aging for many decades. And their findings can give us a rough idea of the markers that the neural network powering FaceApp takes into account, as it transports the app’s users into their biological future.
Before delving deeper into the technology behind FaceApp, it’s worth taking a look at its historical precursors. And ages before FaceApp came into existence, there was Rembrandt. This is a Dutch artist from the 17th century, today famous for his series of extremely unflattering self-portraits, of which some 40 have been preserved to this day.
In 2012, Israeli scientists made a rough facial analysis of these portraits, whose original purpose was to discern forgeries from real paintings. However, the paper they published revealed something else. Their findings in the Israel Medical Association Journal also included “objective and subjective measures” of facial aging that they observed while measuring what sort of impact time had on the face of the artist. Believe it or not — these measures found their way into the logic used by the contemporary FaceApp imagery.
More specifically, the formula’s focus was on wrinkles that appeared as Rembrandt grew older. These included his glabellar and forehead wrinkles — the former being those that appear between someone’s eyes when they furrow their brow. In someone’s youth, these disappear along with that expression, but in later years, it seems that they stick around. Apart from that, the scientists also calculated how loose skin accumulated around people’s eyelids — the process known as dermatochalasis. Also, they observed the appearance of nasolabial folds, or in other words, the “smile lines” between the mouth and the nose.
Rembrandt’s commitment to self-portraits was, neatly for later scientists, coupled with a great love for realism. That made the process of studying aging facial features even easier. Thus, scientists were able to quantify the artist’s “jowl formation” as well as the progressive growth of fat on his upper neck. On top of all that, however, the most important metric they developed was the “brow index,” which documented how the artist’s brow line descended over time. Returning to the contemporary topic at hand, we can actually notice some of these markers within FaceApp’s technology.
What Makes Someone’s Face Appear Older?
Apart from wrinkles, there are other ways via which FaceApp gives its magical geriatric touch. There is also scientific evidence suggesting that we perceive age based on facial color contrasts as well.
A team of scientists from Pennsylvania and France gave a demonstration of the impact of color contrasts on perceived age on caucasian female faces. Those who had a high contrast of colors between different facial features and the surrounding skin usually appeared younger than those with low contrast.
That experiment was conducted in 2012. And five years later, the same group of scientists completed another study that observed facial contrasts across ethnic groups. Therein, they realized that aging definitely brought a lowering of facial contrasts, but most notably in South Asian and Caucasian women. Such a decrease in contrast was noticed in Latin American and Chinese women too but not to such a strong degree.
More importantly, they noted that an artificial increase in facial contrast made faces look far younger, further enhancing the impression that aging and facial contrasts were strongly related. Furthermore, they concluded that this was true across all studied groups, regardless of the cultural and ethnic origins of the observer or the observed.
Knowing all of this, we can more clearly examine the changes that FaceApp makes to people’s faces. There is definitely some sort of color manipulation happening, atop the extremely obvious wrinkling of the skin on the photos. It seems that these findings, along with Rembrandt’s facial markers, are some of the techniques used by FaceApp in their magical aging filter.