Lytro’s new perspective: Stanford dissertation leads to a photographic revolution

As graduate student at Stanford engineering, Ren Ng loved to climb rocks and he loved photography. He and his friends would often take pictures as they climbed. Time and again, though, it troubled Ng that his images always lacked that certain something, that ideal blend of focus, depth of field and exposure present in the most satisfying pictures. Great photography always seemed elusive; more a happy accident, than a certainty. And so, Ren Ng set out to solve the problem.

In 2006, Ng (pronounced “ung”) earned a doctorate in computer science with a dissertation titled “Digital Light Field Photography.” In late February, the startup company Ng founded upon graduation, Lytro, shipped its cameras to the world. The Lytro is the first camera that frees people from the need to focus before taking a picture, allowing them to later select any point in the photograph as the center of focus, and to refocus at will ever after.

Becoming Lytro

The Lytro looks unassuming enough. It is reminiscent of a kaleidoscope more than what one would traditionally consider a camera. It is a sleek square tube of colorful anodized aluminum and rubber. It is about four-and-a-half inches long and an inch-and-a-half square. At one end is a lens; at the other, an LCD screen. The Lytro boasts just a power button, a shutter button, a zoom slider and one well-camouflaged USB port. Missing are all the knobs and buttons of the typical high-end camera.

While the design does Steve Jobs and Jony Ive proud, the simplicity of its form belies the transcendent machine inside.

The Lytro is made possible by two advances in the industry. The first is the broad explosion in digital sensor resolution. This is the megapixel number touted in sales and marketing literature.

As it turns out, beyond a certain threshold, those millions upon millions of pixels amount to overkill for all but the most extreme photographic needs. For most people, the magic number is about 12 megapixels, a point at which differences in resolution are virtually indistinguishable to the human eye, even in poster-sized enlargements.

Lens of lenses

The second shift descends from Ng’s work at Stanford. It is all in the sensor and in the mathematics.

“Optically speaking, you can think of Lytro as placing an insect eye — a single lens made up of thousands of tiny lenses — between the main lens and the digital sensor. It’s so close to the sensor that it becomes part of the sensor,” said Ng in a recent interview in the Lytro headquarters in Mountain View, California.

This intermediate lens is actually an array of thousands of microlenses — a lens of lenses. In essence, each microlens captures its own picture of the light coming through the main lens of the camera and projects it on a small section of a digital sensor. Thus, the Lytro doesn’t actually make one image, it makes thousands of images, each very small in size.

From there, mathematics takes over. Each microimage is unique, varying ever-so-subtly in its perspective. From these tiny differences, the Lytro can calculate the angle of all the available light rays in the camera and, thus, the ideal focal length to any point within the camera’s purview.

If you remember those early digital images where up-close the page just looked like a bunch of colored squares, and as you step away the subject became apparent as Abraham Lincoln, that’s what’s happening. Lytro creates thousands upon thousands of those images, each slightly different from the next. Using algorithms, Lytro has an infinite ability to compute and re-compute images, each one with its own point of focus.

Reviewers are already saying that photography will never be the same and that we may be on the verge of a new photographic era to rival the ages of Kodak and Polaroid. If that is true, someday perhaps among the likes of those who gave photography to the world and then democratized it — Louis DaGuerre, George Eastman, Thomas Edison, and Edwin Land — we may add the name Ren Ng.

Stanford Days

Ren Ng was born in Australia to an astrophysicist father of Malaysian descent who spent time in the United States. Ren attended two years of high school here and then entered Stanford as an undergrad. After graduation, Ng entered the School of Engineering to study computer graphics thinking he might move into medical imaging.

His thesis advisors included Professor Pat Hanrahan, an expert in computer graphics and algorithms; Professor Marc Levoy, the inventor of numerous innovative photographic devices, including an open-source digital camera that became known as “Frankencamera”; and, Mark Horowitz, chair of electrical engineering.

In 1996, Hanrahan and Levoy had collaborated to create the world’s first light-field camera and the process they describe as “synthetic focus” — using algorithms rather than lenses to focus images. Their version, however, required an array of full-sized, real camera lenses adapted into a grid and tethered to computer chips to capture the light field. It was the size of a refrigerator.

Ng began his PhD in 2002 and was well on his way to earning his doctorate, having published papers in the industry’s leading journal, SIGGRAPH, as a student. Yet, the more he thought about the challenges of creating a new camera, the more it drew him in. He decided to suspend his dissertation to see if he could build a human scale light-field camera.

He told Hanrahan, “I’m going to change my topic.” It was an unusual move for any doctoral candidate, much less one so close to earning his degree.

“I put everything else on hold. I read all the literature, whatever I could get ahold of,” said Ng, spreading his hands about three or four feet wide to demonstrate a bookshelf’s worth of titles he consumed. “I wanted to reinvent the camera.”

Stanford difference

Ng created a working prototype from a re-engineered medium-format camera. He wrote a doctoral thesis that earned best dissertation recognition from the Association of Computer Machinery. Things began to flow.

“Birdseed” funding followed from the Stanford Office of Technology Licensing (OTL) and from Manu Kumar, founder of K9 Ventures, who, like Ng, is a Stanford engineering alumnus.

“I met Manu at an all-day celebration of the fortieth anniversary of Stanford’s computer science department,” recalls Ng. “I showed him the camera and the software and he slid a check across the table.”

Today, investment in his company has surpassed $50 million, including capital from such tech notables as Andreessen Horowitz, NEA, Greylock Partners and K9 Ventures.

Asked to place a finger on the crucial difference in getting his company off the ground, Ng laid credit precisely at the doorstep of the Stanford School of Engineering.

“Stanford has been key to getting Lytro off the ground,” said Ng. “Stanford has professors with tremendous technical insight. Stanford has the kinds of people floating around who are eager to help new entrepreneurs start companies. Stanford provided role models. I don’t think I could have done this anywhere else.”