A new method for calculating the optimal colors of any linear input device is proposed. The algorithm searches systematically the two types of optimal colors in the R+G+B planes of the device. In this way, the loci of optimal colors associated to the input device can be plotted in its own chromaticity diagram, but also in CIE (human) color spaces. Thus, we can make a comparison in CIE color spaces of the color gamut of the linear input device relative to that of the human visual system (MacAdam limits). This algorithm has been applied to some linear input devices, fulfilling or not the Luther condition, associated or not to real digital cameras, even to animal visual systems. We have found that the color gamut of a linear input device is the same as that of the human visual system, regardless of the fulfillment or not of the Luther condition. This is a corollary derived from the definition of optimal color, because it has not got any possible physically metamers except itself, independently of the capture color space. This also implies that the shape and volume of color solid of the positive spectral reflectances depends on the color space used, and, if the input device does not fulfill the Luther condition, the metamerism of the device will work in a different way redistributing the color stimuli. But the number of the distinguishable colors can be different depending on the chromatic discrimination properties of each input device. So, this work may open new ways to compare the color gamuts, actually the number of distinguishable colors, between the human visual system and non-human input devices, above all those associated to animal visual systems, because their color metric may differ from that of the human visual system.