Gadgets

Neural networks "revived" people in photographs (3 videos)


Looking at static photos, you cannot geta comprehensive view of the person filmed. To improve the situation, the company MyHeritage tried to use artificial intelligence to automatically improve the quality of images. As a result, the application Deep Nostalgia (deep nostalgia) was created, which analyzes the facial expressions of a photographed person and creates a short video.

The Deep Nostalgia feature is similar to Live Photos iniOS and iPadOS, when a video is recorded a few seconds before and a few seconds after the capture button is pressed. This function was used to capture the perfect shot, which may not coincide with the moment the shutter was clicked. Meanwhile, Deep Nostalgia, on the other hand, "animates" a static photo, and this action is available even for pictures taken not on a modern smartphone.


The video creation process is fully automated.Users just need to upload the photo through the MyHeritage website, where it will be first analyzed and improved to not only improve the quality of the final animation, but also to simplify the execution of the deep learning algorithm. The analysis studies the direction of gaze, the turn of the head of the captured person. Deep Nostalgia also has limitations.


First, you cannot "revive" a group of people, only one person can move. Also, the movement extends only to the face and neck.
To test the capabilities of a new applicationyou must register on the MyHeritage website, after which you will be able to create five free animations. For further processing of photos, the user will have to pay for a subscription.

Source: gizmodo
Photo-to-video service: myheritage