Phone cameras have come a long way in the past decade, and one of the biggest shifts has been in the way that people capture photos. Unlike traditional cameras that use various lenses and other physical tools to create an ideal image, smartphones use digital manipulation to transform pictures caught with a very basic lens. Modern smartphone photography is now all about the phone’s software instead of the hardware.
For example, Apple’s recent models allow users to employ “bokeh” on their phone. In traditional photography, specially shaped lenses are used to make out of focus lighting blur into a pleasantly soft background. For Apple users, this effect is achieved by the phone’s software adjusting pixels after the picture is taken.
The reason that smartphones are changing photography is mostly due to advances in machine learning. Photo adjustments that once took hours of uploading pictures to a computer and carefully manipulating pixels in photo editing programs can now be done almost instantly. Algorithms that find a way to fine tune and speed up the process can have truly impressive results.
One of the most interesting modern applications has been in Google’s newfound ability to take pictures at night. With the smartphone’s capability to capture and refine dim details in the dark, people can take nighttime images that look almost like day. This sort of shot is almost impossible with normal photography techniques that would require extensive exposures and infrared sensors.
In the future, we can expect even more stunning applications of machine learning and photography. The next breakthrough is likely to be using machine learning to shift perceptions of space. Smartphone cameras have such small lens barrels that most images suffer from a subtle distortion. Usually, the only solution for this distortion is a tilt-shift lens that adjusts the angle between the photographer and the picture they want to capture. However, new machine learning may use 3D sensors to discern the angle of a phone camera and reduce distortion.