Spectral X-ray imaging with single photon processing detectors gains substantial interest for many applications. In this paper we discuss fundamental parameters as contrast to noise ratio (CNR) and spectral response as a function of the material in the object. Image properties have been simulated for different photon energies using MCNP5, assuming an ideal detector with 32 x 32 pixels. Simulations are supported by experimental results obtained with detectors from the MEDIPIX family. The CNR is strongly dependent on the number of incident photons and the number of photons absorbed in the object. The requirement for substantial absorption in the object limits the range of useful photon energies. In most cases the CNR is improved when high energy photons are removed from the spectrum. Materials can be uniquely identified or layers of different materials can be separated provided that there is a substantial difference in their spectral X-ray absorption. In most cases an absorption edge in the spectrum is needed to obtain good results. Several examples of material identification and material separation are discussed.