The Single Case of the Epson SC-P800 Printer
and Epson Ultra Premium Presentation Luster Paper (UPPL)
This note owes its existence to challenging statements from Bob David (Bob Rosinsky) on the LuLa Forum, to which some members took vigorous exception, about the merits of PMC for this specific combination of printer and paper. While one can always take sides in a debate based on general impressions, I decided that a more rigorous approach to the question may be warranted. I always like going back to “first principles” when challenged and do a fact check. So this is just that.
In comparing PMC with application-based ICC colour management (ACC), what am I looking for? Most importantly, I’m concerned about the accuracy of hue and tone rendition between the image file and the print, gamut volume, maximum Black, and smoothness of tone gradations where they should be smooth.
I’ve already demonstrated (in “Printing Can be Fun and Easy”) that tone ramps can reproduce smoothly whether with PMC or ACC (but the specific gradations are not identical), so I won’t repeat that demonstration here. As for gamut size – it’s not possible to compare them statistically because none is stated for PMC; however, there is a choice of colour space in the Epson Color Controls menu. It allows for a maximum colour space of ARGB. (Figures 1 and 2).
The question then is whether a P800/UPPL ICC profile exceeds ARGB. That is a tangible question amenable to analysis in ColorThink Pro (Figure 3).
Based on this information, there won’t be much grief being limited to ARGB if using PMC, because only two tiny areas of the profile’s gamut exceed ARGB, and quite likely many photos won’t use those bits of gamut that the profile includes for.
So one of the remaining questions is profiling accuracy: the question being, can the Printer/Paper/Profile (PPP) combination reproduce file values accurately? As my readers will know, I put a lot of analytic effort into this matter, because I think it’s pretty fundamental to the whole printing enterprise. Why spend so much time finely editing our photos to get exactly the tone and hues we want (as viewed on a properly colour-managed display) only to see different tones and hues coming out of the printer because something is amiss in the PPP pipeline? As I already discussed in my Canon Pro-2000 review, especially using the basic dE(76) formula which is valid for proofing purposes, we need to recognize that human visual perception is such that it may not see low dE errors as disturbing differences relative to correct values. How “low” is “low” would vary from paper to paper and colour to colour, but generally speaking, dE(76) errors in the range of 1 to 3 (whether of L*, a*, or b* and depending on which) can be tolerable.
So I performed my PPP accuracy testing for three PPP conditions: (1) PMC, (2) the supplied Epson profile for SC-P800/UPPL (the “canned” profile), and my custom profile. The differences are very stark – so much so that I initially distrusted the result I obtained from PMC and repeated the test, which confirmed the initial findings.
Firstly, gamut volume:
PMC: Indeterminate, but could be ARGB (over 1 million);
Epson profile: 731,428
Custom Profile: 777,840
All the colours in the GMCC 24-patch test file fit within the Epson profile’s gamut (the narrowest of the three above; White sits on the boundary, Figure 4); so perhaps save for a minor bit of White Luminance there are no out of gamut colours that would cause unavoidable inaccuracies of rendition.
The dE results for the three printing approaches are as follows:
PMC: average dE 5.94 (M2). The worst offender is Blue at over dE 15. 11 of the 24 values are above the average, in some cases by a large amount (Figure 5).
Epson Profile: average dE 2.35 (M2). The worst offenders are Green and Red at 4.32 and 4.68 respectively. 10 of the 24 values are above the average (Figure 6).
Custom Profile: average dE 0.96 (M0). The worst offender is Yellow at 2.1. 12 of 24 values are above the average, but apart from Yellow, none exceed about 1.5 (White, mainly on account of lower reported Luminance relative to the file value); (Figure 7).
Based on these outcomes, the superiority of a high-quality custom profile is the best assurance of no surprises trying to match printed outcomes with good display rendition of the file’s colour values. PMC should be a last resort for unimportant work needed in a jiffy.
I’ll finish this comparison with a look at grayscale rendition, in light of peoples’ interest in B&W printing. Using the 21-Neutrals test, the relevant range excludes L*0 and L*100, and the measured results include l*5 to L*95. I remind, that for all these values, the closer the dE values are to 0, the better in respect of both Tone and Chroma.
Firstly, the readings for Maximum Black (lowest L*) show as follows:
PCM: L* Black = 2.98
Epson Profile: L* Black = 2.94
Custom Profile: L* Black = 2.70.
While the custom profile is a tiny bit blacker, there would be no visible difference between the Black points of any of these approaches. Turning to the accuracy of grayscale rendition in general, the results are shown in Figures 8, 9 and 10, and summarized in Figure 11.
There’s no doubt, based on this data, that PMC will yield less satisfactory B&W outcomes relative to file reference values than either the Epson profile (better) or the custom profile (best).
The overall conclusion of this investigation is that PMC is the least preferred approach for making quality prints with this printer and paper combination.