Conversion of light detector signals to intensity values is one of the most important factors influencing precision of spectroscopic observations. Most of the classical light detectors used in astronomical practice are more or less nonlinear. As the photoemulsion has long been the most widespread nonlinear light detector, many improvements in the calibration methods concerned its nonlinearity. In addition to it, there are other substantial sources of inaccuracy in the calibration process of real astronomical images and spectrograms. They are mostly related to real light conditions in telescopes and spectrographs, as well as to the wavelength dependent sensitivity of light detectors. Some of these factors can be taken into account and involved in the calibration process. Similar effects are considered when a flat-field is evaluated for CCD detectors or when image structure varies over a photographic plate.