The spatiochromatic characteristics of a dataset of 29 natural scenes (representative of the natural terrain) were explored. The image capture was done using a video camera capable of recording images through a set of 31 narrowband interference filters spanning the visible spectrum. The system allows the measurement of spectral radiance and reflectance for every point within a given scene. The Fourier amplitude spectrum characteristics of the dataset were explored across the visible spectrum. Our results show that (a) the mean distribution of spectral reflectance and radiance in our dataset is consistent with the shape of the h.v.s. spectral sensitivity curves and (b) the characteristics of the Fourier amplitude spectrum show no significant variation with wavelength. We explored the consequences of having visual receptors different from ours. Our results show that (c) there is no significant variation of the Fourier amplitude spectrum with the bandwidth of the receptor and (d) luminance images produced from these hypothetical receptors show a Fourier amplitude spectrum whose characteristics depend on those of the receptors. Luminance and chrominance images, based on assumptions about human cone response and signal coding were derived from the dataset. Our results show that (e) the different spatial transfer functions of colour and luminance in human vision are not reflected in the spatial-frequency characteristics of our scenes which appear to be rich in high-spatial-frequency chrominance information (f) the Fourier amplitude spectrum of our luminance and chrominance images, when plotted in log-log co-ordinates is a straight line with slope similar to that found for achromatic images in other studies.