"This technology has the potential to be very useful for real-time image enhancement on mobile platforms," says Jon Barron, a research scientist at Google who worked on the paper describing the system. "Using machine learning for computational photography is an exciting prospect but is limited by the severe computational and power constraints of mobile phones. This paper may provide us with a way to sidestep these issues and produce new, compelling, real-time photographic experiences without draining your battery or giving you a laggy viewfinder experience."

The algorithm looked at a data set of 5,000 photos taken with SLR cameras by various photographers. These photos included faces, buildings, locales, musical instruments, a wide variety. The researchers then hired five photography students in art school to adjust the photos for balancing color, tuning contrast, and other photographic metrics. After being comprehensively trained in Adobe Lightroom, the students created, as per the researchers request, "visually pleasing renditions" of the images.

After studying the images on pixel-level, the algorithm has gained a sense of what makes good photos pop, at least when it comes to things you could tweak in editing. The system could easily fit on a phone, Barron and his colleagues say it would take up the memory equivalent of a single photograph.

As far as machine learning photography systems go, getting nicer photos is a fairly timid goal. Many seem concerned with, inadvertently or on purpose, with creating nightmare fuel.