Beauty is in the AI of the beholder

Micron Technology | February 2019

The Modern Marvel of the Smartphone Camera

We are a visual species; 80 percent of what we learn is through our eyes. The majority of the processing power in the human brain is eye-driven, to help us capture visual cues from the environment and store memories, make decisions, form opinions, establish relationships and store memories for future enjoyment.

Whenever possible, we want our devices to replicate this human capacity. We will trust an invention that approaches or matches what we do naturally. This is a huge driver for advancing image technologies, our mission and passion here at Micron.

Well, that day is here, or close to it, due to breakthroughs in smartphone camera processing and storage. Your smartphone’s memory and storage performance are not only driving rapid access and processing, but also the complex forms of artificial intelligence and machine learning required to support a new generation of camera capabilities.

Now that the smartphone camera has met and, on some levels, surpassed the performance of expensive high-end digital cameras, including using multiple images to shoot in extreme low light conditions, automatically creating shallow depth of field for portraits, or recognizing and setting the optimal scene capture mode, the next wave of innovation will be around the transformation of photographers. Ordinary consumers now have the tools to become professional photographers, and professional photographers can access advanced apps to create photographic quality on the level of an Ansel Adams or Annie Leibovitz.

It is said that the very best camera in the world, is the one you happen to have with you.

What we’re seeing now are multiple cameras integrated into a single smartphone, and insane amounts of processing power, like 16GB of DRAM and 1TB of storage on some versions of the Samsung Galaxy S10 plus. Why is this? The resolution of the camera sensors is continually increasing. Most high-end smartphones have still picture resolution that are greater than 20 megapixels and video resolutions that are 4K pixels. However, image processing, by its nature, is conducted in a “block by block” fashion, and there must be tremendous reserves of memory to capture, process and store repeatedly. Plus, we may not want to share a photograph as we shoot it. We may want to save its full uncompressed raw form for post processing later in programs like Adobe Photoshop and Lightroom.

The convergence of phone OEMs, memory and storage manufacturers like Micron, and photography API providers, is powering an entirely new platform for photographers at every level. Now, layer on top of that innovation, a layer of machine learning that has analyzed billions of photographs, which allows the smartphone to automatically establish the optimal settings, so you can capture your child blowing out candles on a birthday cake as a forever moment or shoot the Milky Way over the Mediterranean coastline and get it exactly right.

It’s safe to say, the human eye has met its match.