On Jun 17, 2015, at 10:06 AM, Mike Stewart <mstewart@embassygraphics.com> wrote:
I realize colour would not be an exact match for each sheet; but should be close enough to allow a printer to "match" the proof.
Only you (or your clients) can determine what "close enough" means. You're talking about compromising, which is something often necessary...but how much compromise are you comfortable with? Do you live by the "80/20" rule, or are you a "six sigma" shop, or do you eyeball it until it's "close enough"? All have their place; the important part is figuring out what that place is. I do have some practical suggestions, though. First, do you have an IT team available with a developer group? Building applications to manage the sort of thing you're dealing with is the bread-and-butter business of the vast number of programmers who actually get paid a living wage. Anybody competent should be able to work with you to design something that automates away at least the worst of your headaches. Second...for the empirical part, I'd starting just by examining reflectance graphs of the papers from the i1 Pro. I know Argyll has some good tools for that sort of thing, and I'd expect both BabelColor's PatchTool and the original X-Rite software to also be at least adequate if not excellent. There're some patterns of paper spectra that start to become very obvious very quickly. Especially, there's a very distinctive dip and spike in the blues that directly correlates to the amount of OBA, and the shape of the rest of the curve follows various patterns as well. You could certainly load the raw data into a spreadsheet or statistical analysis tool...but just eyeballing graphs may well be all you need to figure out the patterns, as well as which specific papers you have are the most representative of the particular pattern. Another good tool for analysis is to compare 3-D plots of the gamuts of the profiles. Lots of ways to generate those, with a very common one the ColorSync Utility built into Mac OS X since forever. If you have two papers with similar reflectance graphs and printer gamuts, you're probably pretty safe in using the same profile for the both of them. It's not a guarantee, but, in practice, it's pretty likely. (The sorts of things that can throw it off are ink spread and absorption and the like...but, in practice, there aren't that many different knobs paper manufacturers can fiddle with to come up with papers that have similar gamuts and reflectance that don't also result in similar ink handling. It can happen, but it's not what you'd expect.) Lastly, if you find a cluster of papers that are all similar enough to use the same profile for, you'd be best off building the profile from an average of the data for all the papers. There are various ways to do this...one would be simply to merge the readings from all the scans into a single file and build the profile from that, being sure to use whatever smoothing level is necessary to avoid the profiling engine from trying to fit the data too exactly to disparate devices. Argyll also offers a tool that lets you average multiple data sets before profiling, which is the approach I'd personally take for something like that. But, really...see if you can't find a programmer (or a team of programmers) to build you something that makes life easy for you despite having a profile for each and every paper. It'll give you and your clients the best quality, it'll make your life easier, it'll help keep the programmers gainfully employed...best all around for everybody. Even if your company doesn't have anything like that in-house, there're all sorts of ways to hire somebody for an one-off special project like this. Cheers, b&