A friend of mine has some very important photos that he’s having a rough time getting printed to his satisfaction.
He’s spent hours on each one, and has each photo tuned exactly as he wants it printed. He worked with an expert to get the image just exactly right. At this point, no changes to the photo files are allowed… not even copies!
Now, I don’t understand what his point is here, honestly. You can simply duplicate the “adjusted reference file” and work on the copy to your heart’s content, tweaking it in all manner of ways, and printing out copies until you end up with one that looks like the unaltered reference print. That’s the “normal” way of doing it… but (given that a digital copy is 100% the same as the original) for reasons I don’t understand, he insists on using the original file, and without making any adjustments.
Needless to say, then, he’s having a devil of a time achieving prints he likes. He’s taken them everywhere, and can’t get a print that satisfies him. And he’s no dummy: so I assume he knows he’ll never get a print that looks as vibrant as the image on the monitor. Reflected light vs transmitted, etc etc. But even given that, he’s still not happy.
I’ve even offered to do a free print for him, but for some reason (perhaps he thinks friends couldn’t possibly know as much as “pros” – despite the fact that I taught this stuff at the graduate level) he’s declined my offer.
So, instead, I’m inspired to write this entry. Perhaps he fully grasps all this already, in which case, this blog entry will likely prove useful to some of my other readers. Honestly, I don’t know if he fully understands color spaces in the first place. (During a conversation, he said “Straight through prints seem to be the best” by which I’d take it that he managed somewhere along the line to find a shop with a properly calibrated setup.) So, I’m inspired to write this bit of an introduction. (If you are he reading this, and you do understand all this, at least you can decide if I know what I’m talking about. And if instead you are a regular reader of my blog, perhaps this will demystify something for you.)
First, regarding his photo files, here’s what I don’t know: I don’t know what monitor was used to adjust the photos in the first place. That is, I don’t know if it, like the vast majority, was not capable of encompassing the Adobe RGB 1998 space, or if was one of the newer, better ones that can handle 95-110% of the ARGB space. (People tend to think they want what’s on the monitor to show up in the printer, and it can’t. Period. It can’t. Among o ther things, you’re most likely looking at less than what the printer can print, not more. )
Yes: that means that the print (at least one off a good printer) will have a wider tonal range and more subtlety than you can actually see on most monitors (except those designed to show the ARGB color space.)
I don’t know if the monitor recently and correctly calibrated. Heck: I don’t even know if the print files have been prepared for output.
But more importantly are these two things: I don’t know if the files are 8 bit or 16-bit; and I don’t know what color space they are saved in: sRGB, ARGB, or ProRGB.
Color Spaces
They are called “spaces” because they come in different sizes. Think of a circle, with white in the middle, and gradually going outward toward the edges to reach fully saturated colors on the perimeter. Red off in one direction; blue off in another; green in yet another direction.
Now, say we have three of those circles, each doing that same thing, but one circle is small, one medium size, and one large. (Think dime, nickle, quarter.) The white in the middle of each is the same white; the saturation level of the colors on the outer edges is the same. What’s different (because of the different sizes) is how far it is from the center to the edge.
That is, the larger the circle, the greater the number of subtle levels of gradation, from the center to the edge. So, if you want a wide range of tones, say a sky ranging from light at the horizon to the dark of night overhead, you want a wide color space to be able to present that perfect gradual change smoothly. If the color space is too small, you’ll see “bands” in the gradation.
OK, those small, medium and large color spaces are sRGB, Adobe RGB (ARGB) and ProRGB.
A good camera and a good scanner can exceed all of those in tonal width.
No monitor can show all the ProGRB space, because the monitor’s color space size is somewhere in between small and medium.
No printer can print the ProRGB space. Epson printers can print most (but not all) of the ARGB space. Almost every printer can print sRGB (and every monitor can show it.) sRGB was designed to be the smallest color space on purpose: because every camera can capture it; every monitor can show it; every printer can print it.
On the other end of the scale, ProRGB is so wide that no monitor or printer can come close to encompassing all those tonal ranges.
Here’s an analogy I used a lot in teaching: if the printer color space is the size of a dime, ProRGB is the size of a quarter. Imagine laying the dime on the quarter, and you’ll see what I mean.
What is ProRGB for then? Adjusting photos in (usually) Photoshop. When you’re making adjustments, you want all the tonal detail you can get.
But when you’re done making those adjustments (which, remember, you cannot see because no monitor made can show them … and yes: that’s why experience doing this sort of stuff is the only substitute) you MUST save a reduced color-space copy, likely in Adobe RGB, for the printer to use. Keep it 16-bit if you’re fortunate enough to have access to an Epson Professional printer, otherwise, drop it down to 8 bit. (That’s entirely printer dependent.)
So, with that background, the worst possible kinds of files for him to have to hand off to a master printer would be ProRGB. By definition, the printer driver will be transforming the image like mad, altering shades and tonalities to fit the space of the printer itself.
Next worst would be 16 bit files going to an 8-bit printer. Same thing… the printer driver will have to toss out information to get the 16-bits down to 8 (that is, by the way, 32,000 shades and tones down to 256.)
The “guaranteed-to-work” version of the files is, of course, sRGB, and while that’s fine for “drugstore” prints, no fine artist would choose it because of the limited tonal steps. (There are exceptions; see below.)
That leaves what everyone uses: Adobe RGB.
So, if he was treated right by his advisor, he came out with files using the Adobe RGB color space, either in 8-bit or 16-bit. To get the best inkjet printing, he’d want 16-bit with Epson Pro printers. (There are other printing methods, of course, but I’m not discussing those here.)
If he came out with ProRGB files, or sRGB files for the printer, then he’s virtually never going to get a print he likes. (That said, notice that I said “for the printer.” What he should have come away with is several files for each image: one ProRGB for editing; one or two or more for printing, and of course, the unaltered scan/or original file.)
So… now we’re at the point where for printing photos (not books - photos, on pro inkjet gear) he needs ARGB, 16 or 8 bit files and a printer that can print as much of that space as possible.
Notice I said, “as much as possible” since printers can’t fully encompass the ARGB space either. Going back to the coin example, think of a dime on a penny. The penny is just slightly larger, but not by much (and nothing like the dine-on-quarter example above.) Now instead of having that dime exactly centered on the penny, push it off to one side, but just barely overhanging the edge of the penny.
The dime covers 95% of the penny, but 5% or so is outside the dime’s coverage, like a crescent moon.
The dime represents the printer, which can print about 95% of the ARGB space, represented by the penny.
So now here we are: photo files that are “client perfect” and I take him at his word on that. I assume that if one of those files was given to me, I’d find it to be AdobeRGB 1998, 16-bit. (Photoshop will let you “assign” a color space, or “convert to” a color space. The files must have been “convert(ed) to” the new space, not assigned. “Assigned” only works within Photoshop.
So, (back to the dime on the penny) what do we do about the part of the color (the penny) that is not encompassed by the printer’s range (is not covered by the dime)? (The phrase for that is the penny has a larger “gamut” than the dime, so the penny has parts that are “out of gamut” (for the dime.)
Well, there are two possible solutions.
1) you can imagine the penny is made of rubber, and you shrink it to the size of the dime. In that case, every single point on the penny moves when it compresses. The result is a smaller space (dime-sized) but everything moved proportionately, and so it looks “right” to the eye when printed. The color relationships themselves remain the same as in the original, but every single color has changed very slightly; in most cases, imperceptibly. To the viewer who has never seen the image before, everything looks perfectly correct.
Put it up next to the original, unaltered file, and it will look different.
OR…
2) you imagine that everywhere in the penny that is covered by the dime stays absolutely fixed. In that area, nothing changes; every pixel in the dime is directly over the exact same pixel in the penny. And the stuff that is not covered by the dime; is “out of gamut”s? It all get jammed down to the very edge of the dime. That is, the colors beneath the dime stay exactly the same… but anything outside of the dime gets converted to what’s on the edge of the dime… it’s totally altered in other words. “Clipped” in one sense.
Put it up next to the original, unaltered file, and it will look different.
Those are called “rendering intents” and the first is “perceptual” while the second is “relative colormetric.”
In both cases, the image looks different, but in a different way in each case.
What’s the bottom line here? That the printer is the limiting factor here. That so far, no printer can take a file “straight thru” and print it as is (unless it’s in sRGB form.) It doesn’t work that way. Choices have to be made, compromises taken: you have to deal with out-of-gamut somehow.
There is a way around it, however: just make sure that no colors fall into that out-of-gamut area. And there are two ways to do that. One we have already covered: sRGB.
Some artists have, in fact, chosen to produce their final printer file in sRGB form, since that way they really can get exactly what they see on the monitor on to their print. Nothing is out-of gamut. That is an artistic choice which I would recommend only for images which have strong colors, and do not depend on subtle shades for the correct impact.
Personally, I prefer the wider range of ARGB, even though some of it lies outside the range of the printer, and I therefore have to make choices about how to handle that compromise.
And that leads to the second way to make sure your colors are “in-gamut” – if you know the paper, ink and printer to be used for a print in advance, you can adjust the image in Photoshop using the ICC profiles for that printer/ink/paper/driver. You can bring the greens and blues into range for that kind of print. (That is, although the ARGB space is wider than the printer’s space, you can adjust those “out of gamut” colors to be inside the gamut of the printer.)
Is this tricky? Yes. Is this time-consuming? Yes. Does this require experience? Yes. And this could not have been the case for my friend, since he didn’t know the printer/ink/paper/driver in advance. There’s about a 99.99% chance his images are out-of-gamut.
I can only hope that some well-meaning “expert” didn’t send him out the door with ProRGB files.
Unfortunately such an approach has severe limitations as well, the most obvious one being, now that printer file is for one, and only one, combination of printer/ink/paper/driver and could not be used elsewhere without undergoing significant color shifts.
So without that kind of initial file-adjustment, one is left with choosing a rendering intent because some of the file’s color space will invariably end out outside the range of the printer, and a decision about what to do with that out-of-gamut color will simply have to be made.
In theory, a perfectly calibrated monitor will allow for photo adjustment. Pass that on to a master printer who has a perfect ICC profile for his printer/ink/paper/driver combination, and what should come out the other side is a perfect representation of the image as the client wants it.
What happens in practice, however, is different. It’s reflected light on the print vs transmissive light on the monitor. It’s RGB on the monitor but CMYK on the printer. The monitor shows deeper into the shadows than the print. If you’re not a master printer, you can’t look at a monitor and “see” what the print will look like… and make the changes necessary to compensate.
And in my friend’s case, since he refuses to alter the files (and he should not, if they really are printer ready) then the only adjustments left to be made are during the printing process. One can change the paper and ink. Change the ink density. And for critical work one can create modified custom ICC profiles for each file. (That is non-trivial work, and again would leave him in the position of the custom profile working with only one combination of printer/ink/paper/driver.)
My first suggestion was to work on a copy of the original file, but that was, as I said, nixed at the outset, for reasons I don’t understand. Since you can only tweak the file or the printer however, ultimately that was my suggestion to him: get custom ICC profiles for each photos. Stay with one printer-bureau who will make them; stay with the same inks; the same printer.
Better yet, when you finally get “the right print” run off a bunch of them all at once. Today’s inks and papers will last a couple of hundred years, so there’s no point in doing them one at a time.
Meanwhile, the offer of a free print still stands. :-)
