Epson Introduces Four New Fine Art Papers
At about the time this review will be published, Epson should be releasing both the roll and sheet versions of its new Legacy line of inkjet photographic papers. The market has become rather well-supplied with a number of high quality papers, hence any company bringing yet more offerings to market should have both good reason and something special to offer.
Epson does. Not privy to inside information, my hunch is that Epson looked at its paper line-up relative to all of its new printer models and decided that there were important gaps in its offerings being satisfied by third-party paper makers, so why not fill these gaps with their own. While Epson has been quite strong in papers using MK ink, its line-up for PK papers lacked Baryta and Platine type papers, and its matte line deserved a freshening-up with matte papers that are thick, quite white without OBAs, and having smooth or textured surface. Hence, for now, the four new offerings from Epson: Legacy Baryta (with very small OBA content), Legacy Platine, Legacy Fibre and Legacy Etching – the latter three Epson says have no OBAs.
One statement on the sample packs clarifies, the origins of the papers: “Working with famous European paper makers, art galleries and chemical engineers, Epson created the Legacy paper line to be the finest papers available.” Epson notes in particular the increased longevity and depth of Black these new papers are capable of providing. I’m not able to review longevity and no independent estimates were available at time of writing, but the papers are under test at Wilhelm-Research. I do, however, have results discussed below on the subject of Black rendition.
My approach to evaluating these papers is pretty much the same as I have done for papers and printers in the past: (1) examine paper/printer profiles to see what they say about gamut and maximum Black, (2) examine outputs of paper/printer combinations using measurable test targets for creating an output data base, and (3) making prints of real-world photographs and evaluating print quality for the usual parameters of colour gamut, colour rendition, tonal gradation, maximum Black, sharpness and rendition of fine textural detail.
Because I’m discussing four papers (plus several comparators), to keep the research and conversation manageable, and assuming that the greatest interest would be how these papers work with the latest printer models for which they are said to be designed, I am focusing this review on their use in the new Epson SureColor P800 printer, with some added material on the use of these papers in the new Canon ImagePROGRAF Pro-1000 printer, for which I created custom profiles. I do also mention profile performance for the Legacy paper/Epson 4900 combinations. I did not, however, reproduce a set of test target prints using these papers in the 4900.
Insight from Profile Analysis (Epson and Canon Printers)
Some interesting observations emerge from these numbers:
(a) The 4900 remains the gamut leader for the PK papers (Baryta, Platine), while the Pro-1000 shows slightly higher gamut than the P800 for all four Legacy papers; however, as I discussed in my review of the Canon Pro-1000 printer, the practical relevance of this extra gamut to most people may be quite limited.
(b) Looking at the P800 set alone for the remainder of these observations, the gamut volumes of the Legacy PK papers generally approximate or mildly exceed those of their closest Ilford or Canson comparators.
(c) The gamut volumes of the MK papers (Legacy Etching and Fibre) are within the expected range for such matte papers (and do show quite differently in prints).
(d) The white points of all these papers are very similar.
(e) The Black points for the PK Legacy papers are generally lower than for their Canson or Ilford comparators; the difference of the visibility between a value or 2 and 3 may be very low, but the difference between 2 and 4 begins to show, and the blacker the maximum Black, the more one can expect subtly richer tonality in the prints.
(f) As well known, the maximum Black values of the MK papers are considerably weaker than for the PK papers, partly accounting for the different character of matte prints.
(g) The L* boundaries of the three primaries differ little between all the PK papers; the more interesting observation being for the relatively higher L* values of the MK papers for all three primaries.
(h) Differences of a* saturation within any primary for all the PK papers are unremarkable.
(i) Differences of b* saturation for Red begin to get more interesting, the Legacy PK papers being more saturated (warmer Red) than their Canson counterparts, but no different from IGFS.
(j) The Legacy PK papers, with some exception, tend to show marginally less b* saturation in Green (i.e. slightly cooler Green) than for the Canson and IGFS comparators.
(k) For the PK papers, differences of b* in Blue are unremarkable.
(l) Colour saturation is less for all the MK papers than for the PK papers.
The differences being discussed here between similar paper types, as you can see, are not really large ones, but some could have a subtle impact on print appearance.
Of greater interest are the comparative outcomes printing on these papers compared with other similar offerings. The two ways of going about this of course are (i) to measure outcomes using printed test targets with known colour values, and (ii) just look at the prints and see what qualities they show. I have done both for this review. It generated many sheets of paper and Excel spreadsheets, resulting in a respectable database and set of reference images, both test targets and real-world photographs. I made measurements from prints of the GMCC target (24 patch) and the Outback target (25 patches of grayscales) (Figures 3 and 4), and I observed outcomes by comparing prints of the Atkinson Printer Test Page (figure 5) and the same twelve real world photographs displayed in my review of the Canon Pro-1000 printer.
The main questions of interest are:
The papers’ feel, surface texture, and handling quality;
Accuracy of colour rendition from the printer(s) in which they are being used;
Maximum Black and grayscale neutrality;
Smoothness of tonal gradations;
Print longevity, on which I cannot comment.
Feel, Texture and Handling
All four papers have a nice full-bodied “fine art” feel, and aren’t fragile to handle, say compared with the thinner stock such as Epson’s Ulta Premium Photo Paper Luster or Canon’s Pro-Luster Pro. All four are 314 gsm weight and range between 12 and 20 mil thickness, depending on the paper type. None of them gave any trouble feeding through the Epson SureColor P800 Front Fine Art Feed and the Canon imagePROGRAF Pro-1000 top or rear feeds in either letter-size or 13 x 19 inch sheets. I did not try any roll paper. They are all acid and lignin free, and all except the Legacy Baryta are OBA free. Epson says that the Baryta paper has “minimal levels of OBAs”. I have no further information on exactly what that means. The base of the Baryta paper is Alpha Cellulose, while that for the other three is Cotton Fibre.
Looked at under a 10x loupe I couldn’t detect any difference of surface texture between the Baryta and Platine papers I examined. That doesn’t necessarily mean there are no other respects in which coating characteristics may differ between them. For example, the Legacy Platine finish may be a bit softer than that of the Legacy Baryta, to judge from the feel when I write on them to identify the prints.
Accuracy of Colour Rendition
Using the same measurement procedure explained in the Canon Pro-1000 review, this time applied to the Epson Legacy papers and some comparators, I came up with the following results (Figure 6).
To remind, a resulting average dE below 3 for any printer/paper/profile combination is a good outcome, with results up to 2 very good outcomes. The average dE values of Figure 6 are in each case an average of the dEs for each of the 24 patches in the ColorChecker. The dE for each patch includes the dE values for each of the L*, a* and b* values defining the colour of the patch. In the exceptional case where the average dE exceeds 3, the range of dE values for each patch can be anywhere from below 1 to around 7. In 12 of the 20 results reported in Figure 6 (6 of which are Epson Legacy papers), the average dE is below 2, and in these cases the sizes of the deviations from the reference values for any one channel are generally small. On the whole, the Legacy papers perform well from either printer used in this test. I also noticed that in some cases, custom profiling helped to improve results.
Maximum Black and Grayscale Neutrality
Given many readers’ interest in B&W printing I documented the printed B&W qualities of these printer/paper combinations, shown in Figure 7.
Each of the 23 samples is a different printer/paper/profile combination, including results using Epson’s ABW mode and Canon’s similar B&W mode. The column for L*min shows the deepest black that combination produced. The data set “Averages of Absolute Variance from Zero” indicates the neutrality of the grayscales at the deepest shadow and brightest highlight areas of the target shown in Figure 4. The dataset “Values of 50% Gray from GMCCTW” is what it says: L*a*b* values for middle gray for each sample. Each of these data sets deserves some explanation and comment.
L*min is the measured value of the black patch of the Outback target (Figure 4, upper right corner). Sample 16 is the blackest measured value. Within the luster papers, none of them exceed 2.8 and six are well below 2, which is quite stellar. My Epson 4900 does not produce L*min below 4. Legacy Baryta and Platine do particularly well at about 1.4 in the Epson P800, ABW mode. The values in the range of 15~16 for the Legacy Fibre and Etching papers place them at the blackest limit of what one can expect from matte papers in these inkjet printers. While these values appear high relative to the PK papers, appearance-wise it is a different story, discussed further below. I remind I am discussing all results from drivers, not RIPs.
“Averages of Absolute Variance from Zero” applies only to the a* and b* channels of the L*a*b* colour system. Recall, a value of zero in either channel means “neutral” for that channel and zero in both channels means gray, hence numbers differing from zero indicate a colour bias. (The values in Figure 7 are absolutes, i.e. without the +/-signs, designed only to show the extent of a bias, not its tint or temperature.) The overall average includes the 25 measured patches (in the Blacks and Whites ranges taken together) for the Figure 4 target, circled areas. The closer the overall average to zero, the more neutral these ranges should appear in print. For avoidance of doubt in reading this data set, I’ll explain sample #1 reproduced below:
The overall average (yellow line) is 0.55, which is a very good result (Epson P800, Legacy Baryta). For the a* channel, the average deviation of the Black range is only 0.18, growing to 0.35 for the white range, producing a composite average for the a* channel of only 0.26. Likewise for the b* channel the average deviation of the Black and White ranges and both together is a higher 0.84 (still very good).
Looking over these outcomes, the Legacy papers do very well on this parameter, whether printed in the Epson P800 or the Canon Pro-1000. (That said, they are not alone in doing well.)
“Values of 50% Gray from GMCCTW”
This data set is not variances from zero – it shows the actual measurements of the middle gray patch from the extended gray set below the 24 patch GMCC shown in Figure 3. Hence in this data set you can see both how far the reported Luminance (L*) deviates from 50% gray and the read values for a* and b* channels, the latter indicating the direction of colour bias; a “-“ sign indicates bias toward the cooler tones (green in the a* channel, blue in the b* channel) and no sign bias toward the warmer tones (magenta in the a* channel and yellow in the b* channel). This data point is useful, because a range around middle gray probably reflects a good part of the tonality in most B&W photographs.
For the most part, including the Legacy papers, the deviations from 50% gray seldom exceed 2 L* steps. This means that intended tone is pretty well preserved in both printers on all but two of the papers, where the deviation is about 3 steps. The a* and b* values show that none of these samples are strictly neutral, but the departures from zero seldom exceed 2 steps, and many are below 1. This means that in terms of human visual perception these results would all look quite neutral, especially each seen in isolation, and quite a few even compared. The Epson Legacy Fibre paper in the P800 using the Epson profile shows the most accurate and neutral result of the sample set, insofar as both the a* and b* values deviate from zero by well less than 1 and L* is only about half a step different from 50.
Colour Vibrancy and Smoothness of Tonal Gradations
Both these variables are best assessed by looking at real world photographs, or printer test targets that include a judicious combination of real photographs and tonal ramps. My favorite for this is the Atkinson printer test page shown in Figure 5 (courtesy of Bill Atkinson).
Numbers can provide some comparative indicator of potential vibrancy by examining the a* and b* values from the profiles analysis of Figure 2. The higher the absolute values, (ignoring +/- signs), the greater the saturation potential. I have commented on saturation in the discussion under Figure 3.
All the papers tested, upon close examination of numerous 13*19 inch prints of the test image, show very similar handling of tonal gradation, and all of it very good.
I would like here to revert to two observations I made about saturation and Blacks under Figure 3: “Colour gamut is less for all the MK papers than for the PK papers”, and the lighter maximum Black (higher L*minimum) of the matte papers. While these observations emerge from the data, what they really mean for the visual appearance of prints can differ markedly depending on the photograph and how one edits it. There are many photographs whose colour and tonal range look beautiful on matte papers because they don’t need the maximum colour saturation or black levels of other photos that benefit more from being printed on the luster papers. I’m using the word “luster” as a broad catch-all for Baryta, Platine and similar textured PK-type papers.
As well, one may often find oneself editing a photo to be printed on matte paper so that appearance-wise it will come as close as possible to resembling its luster rendition. Sometimes this succeeds nicely, but sometimes, it just can’t. I made the excursion into this territory because in producing the test prints of real world photographs I found that the Epson Legacy Fibre paper comes closest to preserving the matte texture and feel, while doing a really good job of handling colour saturation and Blacks, yet containing no OBAs and having a bright native white point. This is a very good recipe indeed. Legacy Etching behaves similarly.
However, as I mentioned, this process works better on some photos than others. Hence I provide here softproof screen-grabs of three worked examples (between Legacy Baryta and Legacy Fibre papers) to show what I mean. The first is a case where the attempted convergence between luster and matte doesn’t satisfy completely, the second where it does, and the third a B&W comparison.
(1) Limited Convergence
Notice the orange/red saturation of the engine, and the contrast between the wing and the background. This is the outcome I aim for on Legacy Fibre paper. The real challenges are the rich saturation of deep orange/red on the aircraft engine and the contrast of the wing against the sky.
Comparing Figure 10 with Figure 9, one readily sees what the issues are: it looks kind of flat; the orange/red on the aircraft engine is comparatively weak, sky contrast is lower and the aircraft wing is somewhat buried in the sky, weakening its effectiveness in leading the eye to the moon above. So the challenges are to make this print on matte paper while trying to achieve as much of its depth as we can based on the luster rendition. This turned out to be not fully achievable because some colours in this photo simply exceed the gamut limit of this paper/printer/profile by quite a visual margin.
Figure 11 shows the extent of adjustment I could achieve in Lightroom 6.4 without incurring saturation clipping and smashing of important tonal gradations. By reducing Exposure and Highlights I created “room” for increasing Contrast, Blacks and Clarity very considerably without saturation clipping. I had to leave Vibrance and Saturation alone. Not shown here, I also made a slight positive adjustment of Dehaze. The result worked pretty well on the aircraft wing and the sky, but was only partially successful at rendering the richness of the orange/red aircraft engine seen in the Legacy Baryta version. In sum, this is a photo I would print on a luster rather than a matte paper. It “pops” very nicely on both Legacy Baryta and Platine papers.
(2) More Successful Convergence
The original editing of this photo required boosting of shadow exposure while increasing deep shadow contrast and reducing highlights to preserve the late evening lighting effect. The overall tone and hue range of this photo is more subdued than evident in the previous example. Nonetheless, when softproofed without any amendments for Legacy Fibre paper, the result is that of Figure 13.
The adjustment of this photo to more fully replicate the tonality of the Baryta print didn’t need very radical editing. The main changes were moderate increases of Exposure, Contrast, Blacks and Clarity, with a very small dose of positive Dehaze and a change of Rendering Intent to Perceptual (from RelCol).
This rendition is at least as well contrasted and saturated as the Legacy Baryta version, the shadow detail is good and the photograph looks fine on Legacy Fibre paper. (In fact, the saturation of Yellows is even a bit better in this version – they can be converged more closely one way or another.) By the way, apparent sharpness is every bit as good on the one paper as the other.
(3) Black and White
I converted this photo to B&W in Lightroom and adjusted it to produce a rather “weighty” effect while preserving good tonal differentiation to keep it alive. The sharp reduction of Highlights brought out the clouds without risking highlights elsewhere in this photo. Shadow boost and clarity were both needed to improve micro-contrast in the very dark areas and between slightly darker and slightly lighter stone blocks.
I was particularly interested to see whether I could preserve this subtlety in a matte print, because I thought the subject matter suited itself particularly well to rendition on matte paper.
Figure 16 shows that without further adjustment, the matte rendition looks rather flatter than desirable, hence the main challenge was to see whether contrast and blacks could be improved without smashing shadow detail.
The key adjustments to achieve the result of Figure 17 were to add contrast using the tone curve, along with a plus two point adjustment of Dehaze (not shown). For this photo I actually prefer the rendition on Legacy Fibre paper, notwithstanding my usual preference for luster finishes to achieve higher dynamic range.
Wrapping up this section, the purpose was to show that when these Epson Legacy papers are well matched with the subject matter and the photos given appropriate editing, the outcomes are very good. The papers (along with the inks, profiles and the printer of course) do what we want them to do.
Wrapping up this review, having printed a good number of test targets and real world photographs on the four Legacy papers and several comparator papers in both the Epson P800 and Canon Pro-1000 printers, I have no hesitation highly recommending these Epson Legacy papers, while at the same time recognizing there are other papers on the market that on the whole can achieve similar outcomes – save for two comments as follows. Despite my overall preference for luster papers, I find the Legacy matte papers have achieved a big step forward on apparent dynamic range and saturation without OBAs, compared to some other matte papers I’ve worked with. On the luster side, the Legacy Platine has a particularly pleasing feel and it gives the photos a bit of a distinctive richness that I believe will make this paper a favorite in the fine art photographic community.
From a user’s financial perspective, it is good to see increased competition, to the extent that in the final analysis a choice amongst one of several offerings may boil down to price and the (perhaps small) distinctions that different photographers may observe in the handling and image properties of one paper versus another. Epson has complemented its paper line-up respectably with these new papers.
General Guidance on Measurement Consistency
As I use objective measurements to analyze the performance of printers, profiles and papers, I thought it perhaps useful to review the reliability of the measurements obtained from my i1Pro2 spectrophotometer. The key concern is the extent to which observed variances of measured values relative to their reference values are due to the behavior of the specimens being measured or rather to instrument variability; hence, whether having taken a measurement, the result is a valid statement of the behavior being measured.
There are (at least) two ways of verifying the reliability or explanatory value of the measurements:
(i) Take a test chart and measure precisely the same spot in the same patch 10 times, calculate the average value of the ten measurements, then calculate the Standard Deviation (*) (STDEVPA in Excel) around this mean. This measures the consistency of the instrument over multiple passes of precisely the same spot.
(*)Standard deviation is a measure of the dispersion of a set of data from its mean (average). The lower the number the less the measurements of the samples in the set deviate from the measured average for the set. In our context, the lower the STDEVPA the more uniform are the different readings of the same patches and the more likely that different results are due to real differences between the samples being measured rather than instrument inconsistency.
(ii) Measure a row of patches (each having a different value) by placing the instrument over each successive patch in the row for all the patches in the row, this whole procedure being replicated a total of ten times. Doing the measurements in this manner, one isn’t necessarily measuring precisely the same spot within the same patch ten times over; one is placing the instrument on slightly different locations within the same patch on each pass; hence variability could indicate that the colour value of the patch is not uniform at every point within the patch (not the fault of the instrument).
I performed both tests using the extended set of gray patches under the GMCC chart (bottom row of Figure 3 in the main article) printed on Epson Legacy Platine paper in B&W in the Epson P800 printer using the driver’s ABW mode. The results are all L*a*b* values. For test (i) I used the 50% grey patch, and for test (ii) I used all nine patches going from paper substrate to Black (L*=0).
Figure 1 shows the result for test (i):
Figure 1. Exact same spot x 10
The STDEVPA relative to the average for the L* channel is extremely low. It is also very low for the a* and b* channels, but less so relative to their averages. The dispersion would be indiscernible to the unaided human eye.
For test (ii), the resulting averages and STDEVPA for each of the nine patches are shown in Figure 2.
Figure 2. Nine Patches
On the whole, the STDEVPA results in Figure 2 are greater than those of Figure 1. This indicates that there is some variance of ink laydown within each of the patches, which according to the Photoshop Eyedropper and Info palette are rigorously uniform. However, looking at the patches, these variances are small enough to be completely invisible to the naked eye.
The dispersions in the L* channel remain very low relative to their corresponding averages; those in the a* and b* channels are relatively higher, but taken in the context of the Figure 1 results, it means that most of the STDEVPA is explained by the factors underlying the printing of the patch, not the instrument. This gives confidence that the readings from the instrument are reliable enough for practical purposes.
It doesn’t mean that errors can’t happen. They do, and they show up as egregious “outliers” in the data set. Before accepting and using a data set, one checks for such egregious outliers and re-measures to verify whether the result was indeed erroneous.
A Forum discussion on this review can be found here.