Imagine you have a long table filled with objects. You stand at one end of the table and focus on the nearest group of objects and take a photo. With a normal digital camera, you would end up with a photo in which the closest objects are in focus. The farther away the other objects are, the more fuzzy and indistinct they'll appear in the picture.
Initially, a Lytro photo can look that way too. But if you indicate that you would like a different area of the image to be in focus -- you really want to see those cool toys you put on the other end of the table perhaps -- the focal point switches.
The Lytro's sensor can generate different images based on the light field captured within the camera. While you're limited by a physical camera's focus settings when you take a picture, the Lytro gives you choices. That's because you can change the focal depth of a Lytro image after you've already taken the photo.
As you focus a physical camera, you move the lens and aperture closer to or further from the sensor. A Lytro image -- the proprietary image file is called a Light Field Picture or LFP file -- contains information about the light field for that picture. Essentially, the software creates a virtual lens and sensor, simulating what would happen if you were to have physically changed the focus on a traditional camera. The virtual lens moves closer to or further away from the virtual sensor.
A complex collection of algorithms do the work. The light you capture with the camera serves as information. The algorithms are a set of directions for the Lytro software to follow. The software simulates what a physical camera would have done and generates the image with the appropriate focal point. Without the information from the light field, this wouldn't be possible -- the Lytro wouldn't have the information needed to switch focus.
The company offers a software product called Lytro Desktop that lets you store and manipulate Lytro photos on your computer. Eventually, the camera may also allow you to take photos with an extended depth of field. That means you may be able to take a photo with objects in the foreground and background and keep everything in focus at the same time. Extended depth-of-field algorithms essentially piece together a photo by taking the sharpest parts of every potential picture you would get from a Lytro camera shot. It's almost like taking a series of photos with different focal points, cutting out all the bits that are actually in focus, and then pasting it all together.