It's really quite simple. The "super CCD" has its pixels (photosites) organized in a diagonal matrix instead of a rectangular matrix. Since virtually all existing image processing software works ONLY with a rectangular pixel matrix, the images from a superCCD have to be converted to rectangular format (eg. .tiff, .jpg, .psd, ...). Since the real camera pixels are not in that format, some interpolation is required to fill in the gaps, which results in about double the number of pixels in the recorded file. One then has the option of having the camera collapse the file back down to a 6mp rectangular matrix, or leaving it as the 12mp rectangular expanded matrix. But there's still only 6mp worth of genuine image detail in there regardless.

-ml