The near-infrared region is much more rarely used in imaging applications than the visible wavelength area. This is partly due to applications which often require evaluation for human vision point of view. Near-infrared can still offer useful information which may not be available at visible wavelengths. Many cameras are also sensitive to near-infrared region. Due to all this, it has been used in applications like face detection. In this study, the skin spectra of different complexions are acquired from three media: real faces, photos and prints. The photos and prints are reproductions of RGB images taken just before spectral imaging of faces. The skin is studied and analyzed from these spectral images. The goal is to examine if the media source can be recognized for database search and if different part of the face can be separated.