A long time ago in a classroom far, far away, I was taught that printed media should be understandable also when in monochrome, e.g. when printed out in black and white (cf. the issue specifically for logo design). This could naturally be seen as also applying to academic publications (journals, books, conference proceedings, etc.), but judicious use of color can be effective when accounting for e.g. colorblindness and vision impairment. How, then, can I best include photorealistic images with many colors in a way which retains meaning both in cases where colors are available and where not available?
In my specific situation, I am illustrating sample items used as stimuli in an experiment which is strongly but indirectly related to color (i.e. models I've conditioned on the data include features which encode color). For this reason, it would be helpful for readers to get the best possible perception of what the stimuli look like in real life. Unfortunately, however, some of the colors are hard to discern from others even on an RGB monitor, let alone printed in black and white (which is actually part of the reason for doing the experiment).
Currently, I've tried adjusting brightness and contrast measures for the images, which has helped to a small degree, but I'm afraid to try anything more radical out of fear of misrepresenting the sample items.