Often, instructions for authors make a fuss about wanting figures in CMYK format (referring to cyan, yellow, magenta and black inks which printers use) rather than RGB format (the red, green and blue light that computer monitors use), e.g.
However, researchers usually produce figures which are fundamentally RGB (e.g. digital photographs, scans, images created in scientific software like MATLAB, or figures put together in Powerpoint).
My understanding of the process of converting from RGB to CMYK is that it is best done with:
(a) expensive professional graphics software (like Adobe InDesign or QuarkXPress, etc.), to which many research groups will not have access, but all journal production offices should;
(b) knowledge of the colour profile (e.g. ICC profile) of the printing/press hardware (e.g. Nature hopefully print their journal hard copy on a better printer than the one in my office...)
Furthermore, many readers view journal articles on computer monitors (via PDF and/or HTML versions) rather than printing them out, in which case the original RGB version should be better than the RGB-to-CMYK-back-to-RGB version.
For these reasons, it seems far more sensible for the journals to ask for figures in RGB and then to convert to CMYK as part of their pre-press pipeline. They could then use the RGB originals in the HTML version. PDF versions could be either RGB or CMYK, or journals could even offer a choice of both (just as some journals offer a choice of PDFs with different resolutions, for printing or viewing online).
Lastly, expecting the journal's production office to spend a little time converting RGB to CMYK doesn't seem unreasonable to me given how high article processing charges are (for open access articles, which aren't paid for via the similarly high subscription fees), e.g. many journals charge USD 3000 for processing paid open access articles.
Am I overlooking some counter-arguments? Or should authors ask/petition journal publishers to remove this requirement from their submission instructions?