Undersampled images, such as those produced by the HST WFPC-2, misrepresent fine-scale structure intrinsic to the astronomical sources being imaged. Analyzing such images is difficult on scales close to their resolution limits and may produce erroneous results. A set of ``dithered'' images of an astronomical source generally contains more information about its structure than any single undersampled image, however, and may permit reconstruction of a ``superimage'' with Nyquist sampling. I present a tutorial on a method of image reconstruction that builds a Nyquist superimage from a complex linear combination of the Fourier transforms of a set of undersampled dithered images. This method works by algebraically eliminating the high-order satellites in the periodic transforms of the aliased images. The reconstructed image is an exact representation of the data set with no loss of resolution at the Nyquist scale. The algorithm is directly derived from the theoretical properties of aliased images and involves no arbitrary parameters, requiring only that the dithers are purely translational and constant in pixel space over the domain of the object of interest. I show examples of its application to WFC and PC images. I argue for its use when the best recovery of point sources or morphological information at the HST diffraction limit is of interest.