Re: Request for specifications
These days, Sheetfed Coated V2 is your best bet. However, this produces separations in line with the old film/MatchPrint standard. Not a terrible thing, but a slight bit heavier ink lay than what most offset printers are heading toward today -- GRACoL, more specifically, G7, the latest revision.
The last 10 years of the litho industry has been some ride. An uproar since film was replaced by CTP and whether or not to start printing sharper than the competition (since CTP provided the capability), or curve data to make CTP print like film, and therefore match legacy data (and have reprints match old jobs).
US Sheetfed Coated V2 is that curved up version to match legacy data. G7 is somewhat of a compromise. Straight linear CTP prints very sharp (that's great) but it also loses a lot of print contrast. For printers that took the sharper linear CTP route, many clients complained of flat imagery. Or they push ink high in the hope of overcoming it. Some even altered the input data, but later sharing of that data with others makes for upset (things no longer match).
But to answer your question, what profile to use when converting to CMYK (I think that's your question). The best would be what the printer recommends. But it appears you're not getting a straight answer. That's a clue, maybe find another supplier. Your second choice is good, US Sheetfed profile. That will give you a "decent" separation.
Now, on to your question of tonal value limits, 2-3% is fine at the low end, but I find it odd to speak of 90% unless you're referring to a grayscale (K only) image. When speaking CMYK we talk total ink coverage, the sum of all colors combined, and typical answers range from 280% (web, for example, what US SWOP will produce, lacking punch) up to 340% (what we use). This is not a equal amount of each color, that's the art of separation, undercolor removal and gray component replacement, and is beyond description within a single posting.
This total ink value also does not automatically determine shadow detail, much of that is dependent on the content. But the higher number will give darker (richer) solids. And muddy up lousy content. There is no magic button that makes it all work. A good scan, separated by an experienced color professional, set to an effective total ink coverage and produced on a good press by a competent operator... you see, there are many factors, and explains why every magazine doesn't look like National Geographic.
Not having control, personally, over all these factors, you have to compromise, and choose something that will get you close enough. US Sheetfed works well for that, but it's dated. You could Google for GRACoL and download the free 2006 profile, that's pretty good. Or ask me and I'll e-mail you my G7 profile set to 340% TIC.
Next, the subject of "embed profile or not embed profile." A sticky subject. Here's one answer: if every user understood what a profile was and its effect and knew what profile they were selecting..." I'd say yes, embed profiles. The problem is embedded profiles are like a loaded gun in the hands of a five-year-old. The Photoshop default is US SWOP. Color-wise it's close to sheetfed, but with lower total ink. So the result is flatter than it could be (made for web, no surprise). When we get these files with embedded profiles, we have to ask ourselves, "Did the user mean for this to be embedded? Or did it just happen by default?" We lean toward the accidental and strip out all profiles, unless the source specifically tells us, "Yes, I embedded, and please use my profile." Photographers are typical of this response, the only users so far that seem keen to profiles. They have to, in order for good color from their digital cameras.
But there is another problem with profiles -- within most workflows, any profile triggers re-separation. Which sounds fine, even CMYK can have differing color spaces and need color management to convert from one flavor of CMYK to another (G7 to Fogra27 for example, when a book printed in the US is reprinted in Hong Kong). The problem is that workflows don't realize when the incoming and target profiles match, in which case color management isn't needed (no need to re-separate data that exists in the proper color space to begin with). So here exists another reason to strip profiles -- to prevent unwanted re-separation, the worst result of which is that K only objects come out the other end a build of CMY. Not to mention a finely tuned CMYK separation can turn to garbage.
So the real answer is, you need to talk with your prepress department. If they haven't got a clue, that's your clue. Or use a trade prepress that does know (what we do, we're not printers), and have files prepped and proofed by craftsman, and printed wherever you please (we determine how they print). Communication is the key, and many printers/prep departments WILL communicate with you, even though at first it may seem they don't understand you, or you don't understand them. Keep trying, and be pleasent, avoid any urge to show off how much you think you know, and ignore if they try the same. It's normal human behavior when confronted by strangers. Get past all that and listen to how they do things. It may be different. The defintion of "right" is that it _works_. If their way of doing things gets the job done, listen and work with them. However, if what they do doesn't work, yes, move along.
What's important is the proof, as it's always been in the litho industry. A good printer can make a proof and match that proof. Once you're confident of that, get the proof, adjust if you don't like it (NOT at the press!) and re-proof until you're satisfied. Then print to match that proof. That's how the process is supposed to work.
