A new blog post lays out some of how the Lytro camera works
By far the trickiest thing to understand about the upcoming Lytro camera is the science behind its innovative ability to capture a light field rather than just an image. We delved into it briefly when the camera was first announced, but what the new technology boils down to is that instead of capturing each pixel on the sensor as a dot of color, it captures the direction and intensity of a ray of light. Think of it like vector instead of raster art. Illustrator vs. Photoshop, if you will.
In a new blog post, Lytro has outlined some of the technical and scientific details people have been asking about, specifically with regards to image formats. Since your everyday JPG can't handle all that information, nor changing focus or any of the other features of light field photography, they've had to come up with their own format. The Lytro sensor captures 11 million rays, each of which is "the intensity of light along a path through the sensor." This is four dimensions of information, which is then scaled back to two dimensions for viewing on a screen — so the output changes depending on the parameters you set, giving the images their unique qualities.
This scaling down also has the advantage of significantly decreasing the size of the image, allowing it to be shared more easily, while you retain the full data on your home computer. According to the official announcement, The 8GB Lytro will hold 350 pictures, which means individual files should run 23MB a pop. That's too large for easy sharing, but hardly unweildy. Since these files contain so much data, it's going to be extremely interesting to see what people can do with them as part of the Lytro developer program. That's a lot of information to parse, and hopefuly some very interesting applications will be put together to play with it.