A method of converting cyan, magenta, yellow, and black (CMYK) digital values into cyan, magenta, yellow, black, light cyan, and light magenta (CMYKLcLm) in six-color printers is proposed for providing a better print quality with less dot visibility, high color fidelity, and smooth color transition. Based on the attribute of the dependent variable Lm and Lc to M and C, the input M and C digital values are separated into MYLm and CLcLm digital values, respectively, according to three types of region: bright, middle, and dark. In bright regions, the input M digital values are separated into LmY digital values minimizing the colorimetric errors to compensate for the hue difference between M and Lm, while C is separated into LcLm in the same manner. In middle regions, to guarantee a gradual lightness transition between bright and dark regions, M and C are newly included along with LmY and LcLm used in bright regions, respectively, which raises a redundancy problem, i.e., many to one mapping between M and MYLm or C and CLcLm. To solve this problem, a digital-patch-based dot visibility metric, which avoids the need to scan a large number of printed patches, is proposed based on the standard deviation of the lightness component for digitally bilevel patches through model-based error diffusion, thereby making it possible to choose the unique MYLm and CLcLm with the lowest dot visibility score. Finally, in dark regions, the input M and C digital values are modified by the M and C with the closest color to reduce the use of unnecessary amounts of colorant. Moreover, to eliminate the appearance of color tones on the gray ramp, the input gray CMY digital values are converted into CMYK digital values using a simple gray component replacement, and the input CMY digital values in the predefined gray boundary regions are then transformed by a weighted sum of the CMYLcLm obtained from above-mentioned color separation and the separated CMYK on the gray axis. Experimental results show that the proposed method is able to produce printed images with both a lower color difference and a lower dot visibility when compared to conventional methods.