clock menu more-arrow no yes mobile

Filed under:

If you’ve ever wanted to imagine yourself younger or older — or a different gender — this new selfie app can help

FaceApp can even turn that frown upside down.

Here’s what FaceApp founder Yaroslav Goncharov looks like as transformed by his own app.
Yaroslav Goncharov

Moving from one gender role to another is a long, time-consuming process. Trust me, I know.

But realistically changing genders in a photo is now a snap, thanks to a new, free iPhone App. And FaceApp is also good for seeing what you might look like older or younger, too.

As someone who tends to be in the middle gender-wise, I was curious what the app would do when I clicked on the “male” and “female” buttons. As I suspected, it showed me that there was plenty of room in both directions of the gender spectrum, if I had any interest in that sort of thing.

Ina Fried

Here is me as the photo was taken, more masculine, younger, and more feminine (starting from the top left and moving clockwise).

FaceApp, which launched this week, is the brainchild of Yaroslav Goncharov, a former Microsoft and Yandex engineer.

Goncharov said he started working on the app to see if artificial intelligence could add a realistic smile to a photo where someone’s mouth was closed.

“A friend of mine had a photo she loved, but wished she smiled when the photo was taken,” Goncharov said. “I was too lazy to spend hours in Photoshop, and since I had background in deep learning, I decided to employ a generative deep neural network to do the job.”

While there were academic papers on how to change various face attributes, most reported producing a rather low-resolution, non-photo-realistic image. But, with machine learning getting better all the time, Goncharov said he decided to give it another whirl.

“When I showed the first results to my friends, they were blown away — it looked like pure magic,” he said.

Although such techniques help apps like Prisma and Pikazo transform photos into works of art, Goncharov said most of the filter apps that try similar adjustments don’t produce convincing results.

“We use deep generative convolutional neural networks,” he said. “This technology is quite mature for some tasks, such as artistic-style transfer or super-resolution, but has extreme challenges for photo-realistic tasks. I don’t think that there are currently any commercial products or research papers that can claim similar quality on such photo modifications.”

This article originally appeared on

Sign up for the newsletter Today, Explained

Understand the world with a daily explainer plus the most compelling stories of the day.