Why does backlighting make people black?

Backlight photography involves the problem of tolerance, which refers to the camera's ability to record the brightness range of the scene. The difference between the brightest part and the darkest part of the scene surface can be expressed by the proportion of light and dark. For example, the brightest part of a scene is 50 times brighter than the darkest part, so their light-dark ratio is 1: 50.

The ability of the lens of a mobile phone to correctly show the difference between light and dark of the shooting scene is the tolerance of this mobile phone, which is also called exposure latitude. It is said that the brightest light that human eyes can feel is 50000 times that of the darkest light, so even in the case of backlight, our naked eyes can see the scenery clearly. But for digital cameras, including mobile phones, they can't correctly show the scene with too big difference between light and dark. The sensor tolerance of digital cameras such as mobile phones is lower than that of human eyes, so in environments with large light and dark differences such as backlight,

The digital photos taken are either the sky is correctly exposed and the face is black, or the face is correctly exposed and the sky is overexposed. With the progress of mobile phone photography and the improvement of AI algorithm, many mobile phones have added algorithms optimized for backlight shooting. For example, VIVO does more advertisements in this respect. First of all, it is a backlit portrait. It is obvious from the official pictures that the light in the photo environment is very strong.

According to common sense, in this scene, the sun in the sky is covered by clouds, but the light shining on us is still very strong. Generally, photos taken by mobile phones are black. However, with the AI backlight photo blessing, the task in the photo is not "black" because of backlight, and the facial contour and skin color are clear and natural. Obviously, scores that are too dark are optimized by reasonable highlighting.