I have a JPEG and PNG cover: 300dpi/1800 x2700
It contain a one sentence blurb in small text. When I compile to Kindle, the blurb comes out faint–about half the strength of the original. See below. Is there anything I might try to improve the situation?
Channel curve adjustment, layer masking, maybe a little spot retouching in Photoshop would be the best thing to do. You could try a simple global black point adjustment in Preview but that gradient to black behind the text will make it difficult to get the text more black without “moving” the gradient rightward, in effect obscuring the text (not to mention making the skin look unnatural).
If you have the source material it might be easier to start with that and produce a slightly different design that works better on low contrast monochrome screen.
Just a change of font might help you get a more satisfactory result.
Just freelancing here, but do cover images get downsampled at some step in the process of kindlization? If so, it might be better to downsample the image to the target size yourself in Photoshop (where the text is still vector art and will not itself be downsampled in the process, and presumably PS has good downsampling algorithms) and then provide smaller version to Scrivener for your compile.
I agree, a bolder variant on that font, or something designed with more weight to it would help. You could probably also mess with the line-height and kerning to reduce the size of the blurb—making it possible to increase the font size, thus expanding it back to occupying the same amount of space it currently does. You don’t want to turn it into a blob, but at the moment there is a bit of space being used that could be used for text size instead.
Only if you have the Downsize and resample inline images to visible size option checked in Layout—but as the wording implies, that doesn’t mess with the cover image. The old standard for cover images is 800x600 portrait, that may be higher these days, though the higher you go the more the device itself has to squish it down to size.
Yes, in the end, since the OP’s image is quite hi-res, it must be getting downsampled at some stage – if only by the kindle device itself at display time, if not before in the chain. So, I think the OP should definitely downsample a copy to the intended view resolution before passing it to Scrivener. It is a good idea from a design standpoint anyway, to see at that stage more what you are really going to get.