Leibson's Law

Although introduced yesterday, I’ve held off writing this blog entry until I think I understand what’s happening in Fujifilm’s new Super CCD EXR image sensor, which touts high dynamic range (HDR). Fujifilm has been producing HDR CCD sensors for a few years now. The first one that caught my eye was the company’s Super CCD SR, which placed a big photosensor and a smaller photosensor at each pixel site. The big “S” sensor had a higher charge capacity, so it could accommodate brighter incident light. The smaller “R” sensor was electrically shallower, so shadows in the image could fill it more deeply since it needed fewer incident photons to fill the small sensor with electrons.

As a result, the small sensors were excellent for shadow detail while the large sensors could handle the image highlights. I thought it was a neat trick and a nice design, but it never took the market by storm as far as I could tell. In the megapixel wars, more megapixels sells to the average consumer more easily than “high dynamic range,” which your average camera buyer won’t understand and therefore won’t buy.

Fast forward to the year 2008—today. There are many more camera enthusiasts interested in HDR photography, thanks to some pioneers who have published stunningly beautiful HDR images. (Take a look at Sean McHugh’s beautiful HDR images taken in and around Cambridge University at his site here.) However, today’s HDR images are composites of three, five, or more images shot with bracketed aperture and then overlaid and mixed in popular image-editing packages such as Photoshop, Paint Shop Pro, or Photomatix.

Now, Fujifilm is giving HDR another shot with the new Super CCD EXR sensor. This sensor has only one size pixel but two tricks eke HDR out of the same-sized pixel sensors. The first trick is a rearrangement of the conventional Bayer color-pixel distribution so that there are always at least two adjacent pixels of the same color, as shown below. This arrangement allows the camera to read the CCD sensor twice by interrogating half of the pixels at one point and the rest at a later time. The pixels interrogated early are consequently less sensitive to light and are used to pick up image highlights. The sensor sites read later have more time to accumulate charge for dimly lit areas. The result: an HDR image with “one” exposure.

Fujifilm has not yet introduced a camera using this technology but the company has said that the Super CCD EXR sensor will initially appear in a 12-Mpixel premium FinePix compact camera with a 1/1.6" sensor. The camera will launch in the spring of 2009. This camera wil presumably be able to operate its 12-Mpixel Super CCD EXR sensor in two modes: a high-resolution mode with normal dynamic range using all 12 Mpixels in a similar manner and a high-dynamic-range mode that devotes half of the sensor's pixels to dynamic-range expansion. Although I've been interested in HDR imaging for years, this latest innovation is particularly interesting to me this week because I'm currently writing a white paper on pre- and post-processing algorithms used in imaging. Fujifilm's Super CCD EXR sensor adds yet another new dimension to this fascinating topic and the introduction of this sensor further illustrates the synergistic fusion of silicon sensor technology and advanced image processing.