In virtually all digital cameras, colour images are generated by mixing down the single Red-, Green- and Blue-sensitive photosites into micro-blurred pixels of continuous tone colour (the only exceptions to this rule are the few cameras based on Foveon sensors, which record full colour information at all photosites, and Leica's Monochrom camera, which only records B&W images).
Additionally, many digital cameras come with a low-pass 'anti-alias' filter in front of the sensor itself, which can be thought of as a necessary evil in order to avoid/reduce the occurrence of ugly false detail ('moiré') in the final images.
What all of this actually means for the practical photographer is that all digital photographs need to be sharpened somewhat to 'un-do' the blurring induced by the software algorithm that mixes the three primary colours, and by the anti-alias filter (if present), and thus make the most of the sharpness that one's lenses can deliver.
This first round of sharpening can either be applied directly in-camera to the JPEG file, or done in the 'digital darkroom' to the RAW file. Either way, it is best to be very subtle at this stage, in order to avoid any artifacts and leave the images ready for subsequent resizing. For instance, the following Unsharp Mask (USM) settings are usually good ballpark values for straight-out-of-camera RAW or unsharpened JPEG files:
Amount = ~ 100% , Radius = 0.5 - 1.0 , Threshold = 0 - 5
(Amount may have to be increased or decreased depending on the strength of the camera's anti-alias filter, and on the specific image content - always check the result on screen at 100% zoom to avoid over-sharpening halos! Threshold depends on the amount of high-ISO noise, and on the specific image content)
2) OUTPUT SHARPENING
A second round of sharpening may then be required when the photograph is prepared for printing.
My 'quick and dirty' recommendations are as follows:
Resize (and crop, if required) the image to the desired print size (e.g. 60 x 40 cm), resampling it (uning the 'Bicubic' method) to the nearest integer submultiple of the native printer resolution (i.e. 360 dpi for EPSON Inkjet printers, 300 dpi for HP and CANON inkjet printers, and generally 300 dpi for lab C-type printers), and then apply USM with:
Amount = ~ 100 - 200% , Radius = 1.0 , Threshold = 0 - 5
(In this case it is best to check for the correct Amount on screen at a zoom ratio that best approximates the print output; the most useful ratios often being 50% and 25%)
