Written by Ctein, TOP Technical Editor
I promised last week in my P800 review that I would follow up with a detailed explication of my test results regarding printing from Photoshop using custom-made profiles vs. using printer-managed color. If you don't have any idea what I'm talking about, read the section of that review entitled, “Color management and profiling” along with the first featured comment. That should bring you up to speed.
Before I dive into my results, I'm going to issue a bunch of caveats as well as respond to some commonly asked questions about my “conversion” (ahem) to printer-managed color. Then we can get to the nitty-gritty.
First, and this is really important, printer-managed color won't be a superior approach if you're running Windows. What I'm recommending only works under Mac OS, which allows you to specify the output printer color space as AdobeRGB. Under Windows you're limited to sRGB, which throws away way too much of the color gamut of the printer. Some people have proposed various hacks to circumvent Windows' sRGB limitation. They may or may not actually work. If they do, they're exploiting loopholes (effectively, bugs) in the code. There's no guarantee they'll continue to work in the future.
Printer-managed color is obviously unsuitable if you're having to coordinate with outside clients who aren't running exactly the same set-up as you are. For example, you can soft-proof even if you're running printer-managed color (I'll get back to that), but the results won't be exactly the same as when you're running with profiled color. If you're working with someone else who needs to see soft proofs, what they'll see on their monitor won't be quite the same as what you see.
Similarly, if you're running a whole bunch of different makes and models of printers in one environment and you need them all to behave as consistently similarly as is possible, you need Photoshop-managed color. You can't be having each printer's own drivers calling the shots, with their individual idiosyncrasies.
Finally, I'm confident of my results when Epson x880series and P600 and P800 printers are being used. That's all I tested. Printer-managed color might very well be superior with other printers as well. Or not. I suspect it is with many, but you'll have to run your own comparisons to determine that.
Okay, those are the limitations in scope. Understand that for me this is not some theoretical discussion. I am so convinced of these results that my future work will be printed with printer-managed color. You get to decide for yourself. It's a tough decision; it took me years to get to this point. Old habits die hard.
I get asked three big questions about this:
Q: Don't you have to give up soft proofing?
A: I find soft proofing a useful approximation of what a print will look like, with an emphasis on the word “approximation.” A soft proof fails to convey the nuances of the print for me and the precise interrelationships of the tones and colors on paper. All it does is give me a ballpark estimate. This is not a monitor limitation. It's true I'm running only a calibrated sRGB monitor, but the differences I see have nothing to do with the color space.
Consequently, I find that choosing any appropriate canned profile (whether provided by the printer or the paper manufacturer) works well enough. Different profiles make subtle differences in the on-screen rendering. The differences are small compared to the discrepancies between what's on the screen and what I see in a print. So, no, I don't care that I'm not using exactly the right profile for creating the soft proof; the ballpark estimate I'm getting isn't much different.
Q: Doesn't this mess up if you're using a non-Epson paper?
A: I mostly print on non-Epson papers. On all the ones I've tried, choosing the closest Epson equivalent in the printer driver works just fine. I imagine there could be subtle rendering differences between different papers on the market of the same type, but they're a lot smaller than the differences between printer vs. Photoshop-managed color.
Q: You're throwing away the part of the color space the printer can reproduce that is outside of Adobe RGB. That's a bad thing!
A: That would be true if there were no downside to retaining the entire color space. In this case, there is—worse distortion and incorrect rendering of tones and colors. What it boils down to is, do you want the maximally large color space, or do you want the most correctly rendered color space? Because, you can't have both.
To make an analogy, it's like choosing between two sound systems, one of which can output 200 watts and the other of which can output 500 watts. The 500w system is good, but the 200w system has much lower distortion. Which one do you buy? Me, I'll opt for poorer performance on the extreme transients in exchange for less ongoing distortion.
That's the situation I find with printing. Those outside-of-Adobe RGB colors have much less of an impact on my photographic prints than the distortions created by custom profiles.
Now, on with the show!
For each of the photos, I made five prints—four with my custom profile and the one Andrew Rodney made for me, with both perceptual and relative colorimetric rendering and one using printer managed color. Some photos I tested on the 3880, some on the P800, most on both. Just doing my part to make the paper and ink manufacturers rich. Here's what I found with five of the dozen representative-of-my-work photos I used for testing:
• Xmas Spirit—This one is totally a torture test, but it's not atypical of photographs I do of artificial light sources (not just Christmas lights). There's plenty of out-of-gamut color in the magenta-reds, greens, and cyan-blues.
Relative colorimetric rendering fails badly with both custom profiles. It slams those primaries into the wall, with poor color and tone discrimination. It's especially noticeable in the large red areas. Perceptual holds much better separation, but distortions show up in the near-saturated colors. They don't stand out strongly, but they're there.
Printer-managed color does the best all-around job. It holds better gradation in the saturated areas than the custom profiles did with relative colorimetric rendering without introducing the outright distortions that perceptual rendering did. It retains subtle gradients with better fidelity and more delicately that either profile with either rendering intent. It does a better job of holding detail as well as tonal and color variations in the yellow-orange part of the spectrum (especially obvious in Pooh). I've observed this in other test prints of things like sunsets, with all those golds and yellows.
• Stream Bank near River Polly—This is one of my standard photographs, very typical of my nature photography. I even use it to optimize ColorMunki profiles. It has very small spots of out-of-gamut color in the dark greens with my profile, none with Andrew's.
It was interesting how little difference there was between mine and Andrew's profile. With the custom profiles, perceptual looked better than relative colorimetric rendering. I don't think objectively the custom profiles are inferior to the printer-managed color. Printer-managed color is slightly closer to what I created on the screen, so technically it's more accurate, but it would only be subtle tweaks of curves and hue saturation layers to make the prints interchangeable. I'm calling it a draw.
• Grasses in Liquid Sky—this is another one of my calibration photos because, although the blue sky colors aren't out of gamut, poorly calibrated printers and poor profiles render them distinctly too violet-magenta. I've seen this problem pretty consistently for years across printers and photographs for sky-bluish stuff. I could guess it has something to do with how the RGB space gets transformed into the CMYK space for output, but I really don't know.
I got mixed results on this one. On the 3880, printer-managed color slightly beats out either custom profile, although Andrew's is a bit better than mine. On the P800, printer-managed color is little worse than either custom profile and mine is better than Andrew's. In other words, the order is exactly reversed. I should mention that any of these prints are much better than any print from a factory-canned profile, either for the printer or from a paper manufacturer.
• Ripples in Stream—There are no out of gamut colors, but a good range of fine color and detail cover a large fraction of the color space.
Much like "Stream Bank," any of the prints would be aesthetically acceptable, but printer-managed color was more accurate, in particular in holding the correct hue in the orangish reds and the cyans-to-blues that dominate the lower half. The overall effect is that printer-managed color makes the print more (correctly) vivid, and the different areas are better differentiated.
• Tiffany Dome, Chicago—A lot that's out of gamut here—the blues and the olive green ribbons in the dome glass, almost every bit of the golden glass in the "sun" light.
All the prints render the golden glass more orange than it appears on the monitor (I can see that in the soft proof). Andrew's profile with relative colorimetric rendering does the best job of holding the hue, but it produces the worst overall result—differentiation is absent in the out-of-gamut areas and the print looks flat and lacking in texture. All four profile/rendering intent combinations do a worse job than printer-managed color in holding detail in the sun lamp, in producing accurate shades of blues in the dome and in keeping the greens olive-colored instead of pushing them towards a darker, pure green. Printer-managed color is a clear win.
For the heck of it, I printed out this photo at full 17x22 size on the P800 with printer-managed color, to compare with my earlier 3880 print made with my custom profile (perceptual). OH MY. It is soooo much better. Between the extended density range and color gamut of the P800, plus the better handling of the out-of-gamut colors by printer management. Hard to describe—it's like the difference one sees between looking at an Adobe RGB and sRGB rendering. The print looked great before; now it looks better than great.
I played with several more photographs, but eventually better becomes the enemy of deadlines, and they didn't tell me anything the earlier tests hadn't already. My conclusions are that, except in uncommon cases, perceptual rendering is going to be superior to relative colorimetric if you're using a custom profile. It shows more artifacts in the comparison prints from mine and Andrew's gradient torture-test files, but those artifacts rarely up as a problem in real photographs.
Far more often than not, though, the very best results are going to come from using printer-managed color. "Grasses in Liquid Sky" was the only photograph where the custom profile was of clear benefit, and that was only on one of the printers. Sometimes (rarely) printer-managed color is spectacularly better. Most of the time the differences are small. In the few cases where a custom profile is doing better, the differences aren't large.
Whew! I'm done. I'm going to go back to making pretty pictures and pretty prints.
UPDATE, Saturday: Quite a few of you have said that you can get AdobeRGB color out of printer-managed color under Windows. I'm not in a position to check that out; I'm not even in a position to argue with you. What I said about Windows' sRGB limitations comes from Adobe's printer guru. I'm not saying he's infallible (neither would he) but he knows so much more about this than I do. I can't argue his case. It would be like Mike trying to stand in for me in an argument about physics. I just don't know enough.
Instead, what I'm going to give you folks is a simple 20-minute, two-print test that will tell you if you really are getting AdobeRGB color out and not just sRGB color. For this you need an image file with a really wide range of colors and tones. Preferably something that was done in ProPhotoRGB space. You want something that comes as close as possible to filling up the printer's entire color space. If you don't have any such photograph of your own in your collection, grab Andrew Rodney's torture test file. This is a 16-bit ProPhotoRGB file, and it will destroy any printing system out there. (Don't be distressed by any weirdnesses or artifacts you see in prints made with this file–– it's designed to break things, and those things will almost never matter in a real photograph.)
Make two prints from that file, using what you believe to be sRGB output to the printer and AdobeRGB output. You should see a very clear difference between the two, with the “AdobeRGB” print having a greater color range (that is, producing more intensely saturated colors at the extrema) than the sRGB print. If you do, then you've got a hack that works. At least it does until such time as Microsoft rewrites their code. If you don't, then no matter what the settings in your printer software are telling you, that “AdobeRGB” data is still going through the Windows' sRGB bottleneck.
That's all I've got. Have fun!
©2015 by Ctein, all rights reserved
Links in this post may be to our affiliates; sales through affiliate links may benefit this site.
(To see all the comments, click on the "Comments" link below.)
Featured Comments from:
Michael Perini: "Simpler is better, and your results seem very encouraging.
"You mention an Adobe RGB workflow. In your introduction to Andrew Rodney's piece you recommend ProPhoto RGB (which most of us use): '...if you're working in ProPhoto RGB color space. And, really, you should be, because most if not all digital cameras can produce colors that fall outside of Adobe RGB.'
"So do your recommendations hold for a ProPhoto RGB Workflow? Thanks."
Ctein replies: Michael, yes. Most of my work is in ProPhotoRGB space. The Tiffany Dome photograph is a good example of that. It contains a lot of colors that lie outside of Adobe RGB gamut.
In fact, this is a place where soft proofing is useful and accurate. Under the View menu in Photoshop, the third item down is “Gamut Warning.” If you click that, the screen grays out any colors that are outside of the space you're simulating. You can pick AdobeRGB as the “Device to Simulate” under “Proof Set up” (the top item under View) and you'll see what colors in your photograph actually do lie outside of AdobeRGB space.
There may not be any. Some digital photographs fall entirely within AdobeRGB space. Just because your camera can capture colors outside of AdobeRGB space doesn't mean that there necessarily are any in any particular photograph. But, one never knows in advance, and it could change when you start messing with the photograph in Photoshop, which is why I recommend converting from RAW into ProPhoto RGB space.