Customary Method - Desaturation

There are two ways to desaturate an image.  The first is Image > Adjustments > Desaturate.  The second is to create a Hue/Saturation adjustment layer on top of the image with Saturation set to -100.  Since only the second method is non-destructive, it is the only method that will be discussed here.

The advantages of the desaturate method using an adjustment layer is that it is non-destructive, intuitive and easy to do.  It is non-destructive because the RGB image retains all of its color information.

When used as is generally recommended, the desaturate method does not accurately replicate either perceived or measured luminance.  The algorithm used to calculate the resulting tone is different than the algorithm used by either grayscale or the Curves control to calculate tone.  The %K and Curves value we can measure in the color version of the RGB image will not match the %K or the Curves value created by desaturation.

It appears that the desaturate method derives tone by averaging the highest and lowest channels and ignoring the in between channel.  For example, given an area of a color image whose RGB values are 175, 81, 7, the Curves control calculates the tone to be 101, where 101 = (175*.30) + (81*.59) + (7*.11).  However, when desaturated, the resulting tone is 91, which is the simple average of the highest and lowest channel values, where 91 = (175 + 7)/2.

I have observed that images converted by desaturation have less contrast (appear flatter) than the original.  This is due to how desaturation derives tone.  Figure 1 shows the two sample images converted by the customary desaturation method.  Figure 2 shows the two sample images converted in order to retain measured luminance while still using Hue/Saturation desaturation.

Hue/Saturation usually gives unsatisfactory results.  However, that is because most people use it incorrectly when converting their color images to black and white.  When used properly, it is one of the most accurate ways to convert a RGB image to black and white when retaining measured luminance.  Figure 1 shows what we will get when Hue/Saturation is used improperly.  Figure 2 was created using the Tone Management System desaturate method.