There is always “buzz” in the community of photographers who still make prints when a major new printer model hits the market.
Year 2000 and the Epson Stylus® Photo 2000P marked the first assurance of long-term fade-resistant photographs from a compact home-based digital inkjet printer. Since then, through the Epson Stylus® Photo 2400, Epson Stylus® Pro 4000, 4800, 3800, 3880 and 4900, we appreciated the successive incremental improvements Epson introduced from model to model. We have seen this technology mature very quickly over the past 15 years, such that for any new model appearing over the past few years one is tempted to ask: “does it get any better”? The mark of a mature technology is that one is looking for the finer points.
The 3880 is now history, the P800 will replace it and 4900 owners may be pondering whether to renew with a P800 or keep their 4900s. Hence this discussion views the P800 in the context of the 3880 and the 4900. Yes we know, the 4900 is a “one-up class” printer with two more inks (orange and green), 360 nozzles per inch rather than 180, it weighs a bit less than three P800s and costs at least twice as much (but you get more ink). That makes it all the more interesting as a “stretch target” and some of what you read below may surprise you.
This review is based on tests with a pre-release version of the P800 printer, which Epson America made available to me for this purpose.
As it turns out, differences between some papers overwhelm differences between printers. That raises the question about how the printers perform comparatively with different papers. My standard printing paper is Ilford Gold Fibre Silk (IGFS – about the same as Canson Baryta Photographique and offerings from one or more other manufacturers) hence a fairly broad community of interest in printer performance with this class of paper. What used to be called Epson Premium Luster Photo Paper (PLPP – the latest name for the same thing is Ultra Premium Photo Paper Luster, but I’ll call it PLPP for short) is also a popular luster medium and being an Epson product should be tested in a new Epson printer. Not to be completely neglected because many people like them – matte papers. I selected Epson Hot Press Natural, being an Epson OBA-free smooth cotton fiber matte paper, but I had a passing look at Velvet Fine Art (VFA) and Watercolor Paper-Radiant White (RWC) because they have OBAs. I needed to make my own profile for IGFS in the P800 because none are readily available elsewhere at time of testing.
So the main comparison matrix looks like this:
The object of the comparison is to observe comparative and stand-alone print quality. What do I mean by “print quality”?
- Visible resolution,
- Smoothness of tonal gradations,
- Richness of grayscale and colours,
- “Blackness of Black” and
- Shadow detail/distinctiveness of levels in the quartertones.
Some of these factors are related. For example, gamut affects richness of colour, while “Blackness of Black” can affect tonal separation of shadow detail. Gamut, Blackness and tonal separation are amenable to quantified analysis, however it’s really important to understand that there isn’t an obvious one-to-one correspondence between observed differences in data and what the prints look like when compared. Limitations of our visual perception prevents us from seeing certain differences that instruments can detect, and our image editing applications can adjust images so that the final results differ – considerably – from an unedited comparison between two sets of differing initial conditions.
There are targets for appreciating printer resolution, but even harder to relate findings from these targets to the apparent sharpness of a printed photo. So there are no objective analyses of resolution in this review – only comments on perceived sharpness from looking at prints of real world photographs with no visual aids, and through a (3x/10x) loupe. More than this is unnecessary for most purposes.
The Blackness of Black:
I measure this directly from prints using the Outback Photo printer test page (Figure 1). It holds a black patch (R=G=B=0) in the upper right corner, and a shadow ramp from (ProPhoto) RGB2 to RGB24 in the lower left. I use an X-Rite Pulse spectrophotometer and the Color Shop X Color Scratchpad for the readouts of the Lab values from each patch.
The story here is about not only the printers but also and very importantly the papers. There is a very large difference of maximum blackness between the luster papers on the one hand and the matte papers using MK ink on the other. There are smaller differences between printers for the same paper and patch. While laundry soap commercials tout the brand that produces the whitest whites, here we discover the brand (of those tested) that produces the blackest black. The clear winner is the new Epson P800 printer used with IGFS in ABW mode. L* = 1.63 – very remarkable, considering that the next best contender is the P800 again, this time on PLPP coming in at L*=2.33 (non-ABW mode; I infer from other results L* would be below 2.33 with ABW, but PLPP/ABW was not directly tested). Then we enter the range of about 4.5~5.4 for IGFS or PLPP in the 3880 and 4900 models (Figure 2).
But the real big jump is the move to the matte papers, where regardless of the printer the blackest black values read from the paper are in the range of 14~17.8 – visibly a deep gray and one of the reasons why matte prints, depending on the photo, can look muddy when compared together with luster prints of the same photos. That said, of the lot, the P800 in ABW mode once again scores the blackest black of its class for HPN at 13.96. HPN non-ABW comes in at 14.98 in the P800 and 17.75 in the 3880, indicating that the P800 can generate blacker black than the 3880 on this matte paper. I did not test matte papers in my 4900, because the MK ink has been sitting there unused for a long time.
HPN has no OBAs, while Velvet Fine Art (VFA) and RWC do. Regardless, their black performance in the P800 is no better than that of HPN, and not always as good.
Turning from the blackest blacks to the performance of the shadow ramp in the Outback target print from RGB=2 to RGB=24, the story is a little more involved, but revealing. The object of the exercise was to determine the growing extent of visual and measured distinctiveness of the gray scale values from patch to patch relative to each other and to their pure black background. The principle is that the greater the distinction, the nicer the separation and contrast of shadow tones in the bottom of the tonal scale that one may expect in real photographs. The first question is at what level does one see any distinction at all between a deep grey patch and the black background. The next question is how much more distinctly do these patches appear in the prints from the level corresponding to their first visibility onward.
To answer these questions, I printed this target in all three printers for the three papers except for HPN in the 4900. (I also printed VFA and RWC in the P800 to compare OBA-laden with no-OBA matte papers in the same printer.) I then used my X-Rite Pulse spectrophotometer and ColorShop-X to read the Lab values of each patch in the shadow ramp of each print. As the Outback target values are quoted in RGB (ProPhoto) and the ColorShop X values are in Lab, I corresponded the latter with the former using Bruce Lindbloom’s colour calculator. This correspondence appears in the first two rows of the table in Figure 3.
This same table then proceeds to report the measured value of each patch for each printer/paper combination that I measured. The yellow boxes show the first patch for which the tone is barely distinguishable from the surrounding black. (For context, on my display, calibrated to 110 cd, D50 and gamma L*, I can begin to distinguish tonality from RGB 2 onward – just barely at 2, but no question from 4.) The final column of the table called “RANGE” is the simple difference between the left-most and right-most values read from each print. The wider this range, the more tonal separation the printer/paper combination is achieving in that part of the tone curve. Very few measured printed output values for each patch are the same as the ProPhoto file numbers for the same patches. Some are close. This is expected. It’s the overall tendencies of the results and what they imply for tonal separation that is of interest here. The key results of this exercise are:
- Except for the P800 in ABW mode, none of these printer/paper combinations can distinguish between shades of deep gray until at least RGB=8/L= about 2.
- The P800 in ABW mode can distinguish grayscale from RGB=4/L= 0.5 up, which is truly exceptional and quite path breaking for this class of printer.
- The range is far narrower for the matte papers than for the luster ones. This narrow range and their relative inability to reflect deeper blacks explain why tonality and shadow detail in matte prints can look muddy by direct comparison with luster prints unless very carefully worked in LR/PS.
- In this vein, VFA and RWC, despite their OBAs, don’t perform better than HPN – in fact not always as well.
- The range derived from the P800 generally beats everything tested except for PLPP in the 3880, which is a tad higher – to which I would attach no significance – the difference is likely within the margin of measurement error.
- Bottom line inference from all of this: Epson has a point when they claim the P800 reaches new depths of black and deep shadow tonal rendition. The data substantiates the claim.
Lest readers think I may be too hard on matte papers, such is not so. There are images to suit all papers and papers to suit all images. The print on HPN of the photo in Figure 4, if I may be allowed to say, is very successful – knowledgeable observers have spent time looking at it and liking it; it also reveals qualities of fine texture rendition. That said, in general I’ll stick with my luster papers for their superior range and the easier image editing under softproof that they permit.
Blacks and shadow detail are two aspects of print quality; the larger picture though is the overall gamut a printer/paper combination can achieve. We look into that by profile inspection using ColorThink Pro – the 2D and 3d gamut mapping and Profile Inspector tools. Here two factors are most relevant: the overall gamut volume (the larger it is the bigger the “crayon box” for drawing colours”) and the shape of the gamut – in which parts of the colour spectrum is one combination stronger or weaker than another. Here again, the comparisons of interest will be between the P800 and the 3880 on the one hand, the P800 and the 4900 on the other, and between the matte and luster papers.
The table reproduced in Figure 5 reports the gamut volumes achieved for each printer/paper combination, the higher the better.
The 4900 achieves the highest gamut volumes of the three printers for all the papers, but by much less of a margin for HPN than for PLPP and IGFS. As expected, the gamut volumes of the matte paper are low compared with those for the two luster papers. Comparing gamut volumes between the P800 and the 3880, for HPN (Epson profile) that of the P800 is slightly larger, for PLPP the P800 shows slightly smaller and for IGFS the P800 shows smaller. None of these differences except the last one would matter to image editing or print quality and even for the IGFS the practical impact of a 9% difference (P800 vs 3880) would be very slight. Only the 4900 shows a materially larger comparative gamut for the luster papers. The HPN (Epson profile) gamut volume difference is small between the P800 and the 3880.
The following figures show two and three-dimensional Lab portrayals of this information for the three printers and three papers. These help to show differences of not only gamut volume, but also gamut shape, as no one printer/paper combination is necessarily stronger than the comparators in respect of all colours.
In the 3D plots, the colour that shows through refers to the printer with the wider gamut in the area of the visual spectrum shown in the “True Color” ring below.
From Data to Prints:
Leaving the rather sterile world of profiles, data and diagrams, it’s time to talk about how these printers comparatively print real world photographs. Here, the selection of the photos and the approach to the comparison are rather important.
For the selection, I was looking for photos that would reveal capability in at least one of the following areas:
- Quality of reds;
- Quality of yellows;
- Quality of greens and blue sky;
- Smoothness of tonal gradations, particularly skies;
- Quality of shadow tones and detail;
- Vibrancy (where wanted) and subtlety (where wanted);
- Resolution (image detail);
- Quality of B&W rendition.
I happened to have been printing a group of photographs I made in Edinburgh, Scotland last year and found several that I considered suitable for printer testing in respect of these criteria (Figures 15 – 21). I used all of these photos to evaluate the printer for all the aforementioned qualities, but have selected two of them for detailed analysis below, as they all present the same general comparative picture.
Both Edinburgh; all photos co. Mark D Segal
These photos are all in “print ready” state. They did not look quite like this as unedited raw files just opened in Lightroom; more about that now, as it lies at the heart of printer evaluation: the main message being that apart from printer characteristics, technique, outcomes and evaluation go hand in hand.
There are basically two ways of approaching the evaluation. One I’ll label the “non-interventional” approach, and the other the “interventional” one. The difference is about what one does with the files before printing them. The principle underlying the non-interventional approach is that one should see what the printers produce without profile/printer specific editing intervention, so that the performance of the machine is not comingled with the proficiency of the image editor. For this approach, one optimizes the file based on its display appearance without softproofing, then prints it in the various printers each with their bespoke paper/printer profile.
The principle underlying the interventional approach is that a serious printer will start with a raw file, printer and paper, and edit the photo, eventually under softproof for the chosen printer/paper combination, to whatever extent necessary for achieving the best possible print the printer and materials can deliver. While the non-interventional approach may be of clinical interest, I assume the interventional one is how most serious users of these printers would go about their printing, so I tend to prefer focusing on this approach, but look at both. In an article published over the Internet you cannot see the prints, but you can see the screen captures of the non soft-proofed and softproofed display versions. For brevity, I shall make this demonstration for two of the above photos, because the story is essentially the same for all of them and points to the same general conclusions, discussed below.
Figures 22 to 27 inclusive show the raw material and outcomes for photo DSC4356 “Rose & Crown”, where the interest is in reds, shadow detail and highlight con