Smartphone cameras started the selfie photo rage, and it shows no signs of slowing down. However, taking a picture of yourself with your phone can still result in a bad portrait, no matter how many megapixels your camera might have inside. Today, Adobe announced that its research team is exploring new ways to use artificial intelligence and machine learning, powered by Adobe Sensei, to help improve selfie photos.
The company demonstrated what it is working on in a YouTube video. It showed an app that can take a normal selfie photo and then change the image so that the photo looks like it was taken from a greater distance. The app’s tools can turn a selfie image that might seen too tight or crooked into one that looks more like a normal head shot taken by someone else. It can even turn a 2D image into one with limited 3D depth. The app can allow you to move the head in the selfie image a bit to get a better angle.
It showed an app that can take a normal selfie photo and then change the image so that the photo looks like it was taken from a greater distance.
Another special effect shown in the video is called automatic portrait masking. With the tap of a finger, the app can remove the background from the selfie, and then a person can return it to the photo, but with a more pleasing depth of field image effect. It makes it look more like a photo taken with a traditional camera. Finally, the demo video showed how users can search for a photo that you want to mimic for your own selfie portrait in the app, and then automatically apply the filters and other effects from that old photo on your image.