专讗砖讬 English EPSON R3000 Review

EPSON R3000 Review

1
0

We received the Epson R3000 in late December 2010 – almost an entire month before it was announced. The new Epson Photo printer is designed for advanced amateur and professional photographers, it allows for A3+ prints and includes many of the most recent print technologies introduced by Epson. Today we present to you Dr. Boris Oicherman's review of the R3000. Dr. Oicherman spent the past two months with the printer and besides a comprehensive review also included some important insights regarding printing.

Printers' evolution

"Things have changed in the inkjet printers' world鈥 – I thought to myself while unpacking the package: the shiny black device looked nothing like its' predecessor on my desk from ten years ago (no names…).

It was quite a traumatic experience back in 2000 which made me think more than once why on earth did I not stick to my printing service shop. Do I really need to go through all this just to print pictures? To begin with, every print had to start with the ritual of head test and cleaning, unless I could tolerate those nasty banding lines in my pictures. Cartridges that ran out of ink before I could print hardly anything, partly due to excessive head cleaning. Color that faded in front of my eyes unless put under a museum grade glass to shield it from UV.

Epson R3000 in action (credit: Dr. Boris Oicherman)

At that time bootleg pigment inks could be purchased on the web in bottles, along with the hacked print heads and a bunch of infusion tubing (the whole thing was called 鈥Continuous Ink System鈥), all this has promised longer print life and huge savings on ink cartridges, but in practice delivered microscopic color gamut and even more head clogging, to the point that no prints could be made at all.

Black and white pictures was impossible because of the extreme sensitivity of inks colors to illumination ( 鈥illuminant metamerism鈥 in color science parlance). Holding a black and white picture close to a window will result in gray colors turning to an ugly shade of green. This also could be solved by buying inks specifically designed for black and white, six gray inks of different lightness, but I never got this far. It didn't take long until I gave up and got back to my printing service shop.

What changed? The Epson R3000 is a machine capable of printing up to A3+ prints with 9 ink cartridges and more DPI than you will ever need. The printer is designed (according to Epson) for "advanced amateurs" and professional photographers which means that it should be capable of producing high quality exhibition grade prints and generally save the owner the trip to the printing service, as long as the paper size is no larger than 329×483 mm.

By now I can spoil the bottom line of this review: the new Epson R3000 does print exhibition quality prints鈥 and it does so with the minimal effort on the owner's side. There are things I did not like, all to do with design, and none has to do with the quality of the prints. Now for the review.

A remark on WIFI

One feature which I actually did not have a chance to check was wireless printing (the R3000 has a built-in WIFI/LAN). The printer I received was a pre-production unit and there were some limitations regarding the WIFI.

Printing directly from the Camera

It just so happened that I did not have a computer to plug the printer to on the first day, so the Epson representative suggested: 鈥渨hy won't you just print straight from the camera? The machine has a USB input!鈥 OK, so I took it as a warm-up: here I go, plugging my D90 into the printer, the PictBridge interface popped up on the camera and 鈥 who would believe, within couple of clicks it actually printed. There are even some options to play with: paper size, number of copies, even cropping. I can't think of too many cases where direct printing from the camera is useful (besides kiosks and, perhaps printing wedding photographs on-site), but it worked and I trust someone will find what it's good for.

External design and paper feeding

I did have one problem though. Look at the printer's picture; see that small white mark at the bottom left? This is the PictBridge logo. Can you see the USB port just below it? I couldn't either. I'd love to find out what was the human engineering basis for sticking that port in a deep socket at the very bottom of the printer. Was it in purpose that the port is virtually invisible from the height of a standing human? Is the requirement to lay one鈥檚 head down flat on the desk to see which way to stick the plug listed in the product design specifications?

Design was always an issue with printers. The focus on quality and speed made the engineers forget that users are human beings, and the R3000 isn't different in this respect from other printers.

On first glance the black shiny appearance of the R3000 looks great, however, a closer examination reveals a rather flimsy plastic feel; this is especially notable due to the need to open and close different trays to load the paper. There are four trays to be opened and closed at different times to tune the printer for different kinds of paper, and these do not feel very secure. Also, because of some sort of an electrostatic reason the R3000 black plastic attracts dust in industrial quantities, which is rather annoying.

The R3000 in Japan CP+ expo – read our full coverage (Netanel Yadrai).


The aforementioned four trays serve the paper feeding system: two input trays and two output ones. One pair of input and output trays is for 鈥渟imple鈥 paper types, and one pair is for 鈥渇ine art鈥 papers – the heavy inflexible type which tend to be problematic with all printer brands. The feeding of the latter ones is a 鈥渟emiautomatic鈥 process, i.e. you push the tray at the front so it springs out, push the paper carefully all the way through the printer and press OK. At this stage the printer pulls the paper in and adjusts it. This process worked fine in most cases but occasionally the printer reported a misfeed and the user had to try again. These problems occurs less frequently after you get familiar with the process, however, the feeling of unnecessary over complication remains: why can't things be simpler? Can鈥檛 the paper feeding be engineered to do all papers through the same path?

The ink system

The R3000 has 9 inks: CMY with light cyan and magenta, two grays, and two blacks 鈥 for matte and for glossy papers. These two give the only ink-related trouble with this device: for whatever historical or engineering reasons, the two blacks share the same heads and pipes. The practical effect of this is the need to switch between the two blacks every time one changes paper type from glossy to matte. This process is automatic and takes a couple of minutes of funny noises and a tiny amount ink. I鈥檓 being told that this is an improvement from the previous model, where one had to switch the printing heads manually. Well, this is a kind of improvement that makes things less bad rather than better. Whatever the constraint that has led to this situation is irrelevant: wasting two minutes to throw away ink in order to replace paper type is something of a nonsense, I expect this will disappear in the future models. Otherwise the ink system is trouble-free: the installation is easy, the labeling is clear, no hints of even a slight head clogging. The ink cartridges last for quite some time: I printed more than 40 full-size A3+ prints, about 30% of which B&W, and I just replaced the matte black once.

Print quality

OK, I'm done griping: design imperfections aside, this is a great printer, it does print high quality pictures, in color and in black and white. In fact, over the course of using the R3000 I could not make up even a single significant complaint related to the print quality. It seems very much like these things are getting to the peak of their technological development; I honestly can't see too much room for improvement.

I have done all my prints in Photo 1440 dpi mode, with 鈥淗igh Speed鈥 option unchecked. I never saw any advantage in using the 5760 dpi mode: it takes about twice as long to print an A3+ and the improvement is hardly measurable (except maybe in sharp straight lines). The driver supports 16 bit printing, but I also couldn't find much quality advantage in using this feature.

In the mode I used the prints were invariably clean with not a slightest hint of screening dots, posterisation, or any other visual disturbances. The tone and color transitions are smooth, with plenty of details anywhere on the tonal scale. The transition to shadows on glossy paper is something of a masterpiece: it has sense of glass transparency. I can't remember such a phenomenon on any other prints I've seen.

Black and white printing

Choose 鈥淧rinter Manages Color鈥 in Photoshop's printing dialogue, go to 鈥淧rint Settings鈥 driver's window and in 鈥淧rint Settings鈥 OS dialogue, choose 鈥淎dvanced B&W Photo鈥, choose paper type and…

A comment on the driver interface

Sorry, one last complaint. As inkjet printing technologies advanced in giant steps during the last decade, the human interfaces of inkjet drivers remained conceptually stuck in the last century, making things easier for the machine and for the programmer, but not for the user. The poor user is till forced to prints in 鈥35 easy steps鈥, while trying to find his way through a maze of setting, menus and sub-menus. The most famous example was the need to remember to disable the color management in the printer driver if it's enabled in Photoshop, and vice versa; thanks God this one has been solved.

Now, my problem is this: the list in the 鈥淧aper Size鈥 menu contains no less than ten options for paper sized A3+, of different types. These papers are all, I hope, A3+, they are just different types called by unintelligible names. Not only that, there is another menu below called specifically 鈥淢edia type鈥 which lists, indeed, media types. There are two lists where the user chooses media types, one above the other, and the user should coordinate between the two. You will not be able to choose your 鈥淰elvet Fine Art Paper鈥 in the bottom list if you do not choose A3+ of a kind 鈥(Fr.-FineArt(Borderless, Retain Size))鈥 in the top one. And next to it in the list there is a paper called 鈥(Sheet Feeder 鈥 Borderless(Retain Size))鈥. And you shall not confuse those, for then your 鈥淰elvet Fine Art鈥 option grays out. It seems to me that the technological challenges involved in developing 5760 dpi inkjet process with variable drop size and wide gamut inks are far more complex than the ones involved in building a human-readable driver interface.

A million paper types on the Mac driver interface

By the way, I鈥檓 being told by Epson that this is a purely Mac phenomenon.

Back to the B&W printing

Neutral gray color in the process (i.e. CMYK-type) printing can be created in two ways: by printing black ink only, and by mixing all CMYK inks in proportions that make 鈥済ray balance鈥. The 鈥渂lack only鈥 version will obviously create a neutral gray tone 鈥 black can't be anything else. It will be the most stable to printing variations and illumination change: it will look perfectly gray under an incandescent lamp as well as under a fluorescent one or daylight. The downside is that the number of gray levels will be limited to whatever is achievable by a single ink 鈥 256 in 8 bit printing.

Adding intermediate gray inks in addition to black helps: instead of putting less drops of black ink on paper to create some mid-gray tone, one can print mid-gray ink. Now we have a much larger number of gray levels to play with since every mid-gray adds another 8 bit to the scale, so we've dramatically improved the number of printable tones and made richer prints. This is one of the reasons why the R3000 has two gray cartridges – gray and light gray. Another reason is reduction of the visibility of black dots in light color areas such as skin tones.

Test targets used in the review. The images sources are: JIS set of test images (top two) (LINK); SHIPP image set (middle left) (LINK); the CIE 鈥淪ki鈥 image (middle right) (LINK); detail from the PhotoDisk target (bottom right) (link); synthetic gradients (bottom right).

The CMYK version can produce rich B&W prints, but will be extremely sensitive to process and illumination variations. Our eye becomes a high a precision instrument when it comes to detecting even the slightest variations in neutral grays, so we'll easily see the smallest deviations resulting from imperfection in ink composition or, more famously, from the change in the color of the light 鈥 metamerism (illuminant metamerism, to be precise). Dye-based inks are more prone to this phenomenon, pigment-based inks like the ones used in the R3000 are less, but it's always there.

The golden path always lies in the middle: to compose the B&W images mostly from black and gray inks, and add a small amount of CMY inks to add richness and precise tone control. This way we get stable prints along with the ability to tone them with, say, sepia. This exercise is not trivial on itself but the Epson guys have done it beautifully. In the 鈥淎dvanced B&W鈥 setting the prints on all papers I tried come out uniform, slightly-warmish neutral, with plenty of detail at any tonal range with no bronzing at all. The color was stable gray at any illumination I tried 鈥 tungsten, fluorescent, shadow direct sunlight.

The default toning settings (warm, cold, sepia) came out too saturated for my taste, but are easily modified by the color picker tool in the driver window. The same window also has sliders for lightness and contrast control for those who are not happy with the aesthetic choices of Epson's color engineers.

I have had one rather esoteric problem though. In fact, I have always had it since I have started to use inkjets to print scanned black and white negatives. When printed, the B&W negative grain gets this harsh look at the near-black part of the scale, like sand, as if there are not enough bits to reproduce the transition smoothly. I鈥檓 not sure how to explain it technically; it is not R3000 specific, and anyway who knows how many psychos like me still scan B&W films. It would be interesting learn though what Epson engineers can tell and do about this.

Black and white setting window

Color management

The immediate question arising when considering the topic of color management in inkjets is "to profile or not to profile?" The answer is 鈥 mostly not, but before I explain why – an historical note is in order.

About a decade ago photographic inkjet devices have introduced a new concept to color science: 鈥淩GB printer鈥. Anyone who remembers anything from the color class in one鈥檚 photography studies would remember the subtractive and additive color mixing diagrams, and would shout: 鈥渘onsense, RGB is for displays and cameras, printers print CMYK!鈥. You'll be right: inkjets did not change anything in the basic physics of light. They did, however, introduce color printing technologies to places where they were not common before such as offices, homes and photography studios. This resulted in a rather rare coincidence: the interests of the industry and of the users converged. The users did not know anything about printing technology and color management and most of them did not want to learn. The manufacturers on the other hand wanted to be free to control the printing quality and introduce changes in the printing process as they please, but could not ask the users to adapt to all these changes.

The solution sort of emerged by itself: users would send their images to the printer in RGB color (all the consumer and professional cameras use RGB anyway), and the printer would do whatever magic it has to in order to convert the RGB to whatever printing color mode it needs. This way the photographers could keep their working practices, and the manufacturers could add color inks to expand the gamut, add gray inks to improve the black and white printing, fine-tune the color rendering algorithms independently of the user-side color management, optimize the printing parameters for different media types, and so on.

The bottom line of this story is this: If the color scientists in Epson have done their job well, i.e. have built good transformation from some standard RGB space (sRGB or AdobeRGB) to the printing space (i.e. combination of Epson inks with some range of papers), the user 鈥 however advanced – should have absolutely no reason to fiddle around with building ICC profiles for the supported papers.

Given that profiling software and hardware (and color consultants) don't come cheap, this is a great deal. In fact, the Epson scientists are in much better position to build the RGB-to-printer transform than any super-advanced user, since they know their machine with all its glitches inside-out. What I wanted to find out is whether they've actually done this.

I built ICC profiles for Photo 1440 dpi mode for all four papers I had on test. I printed my test page once using the profile I built, once letting the driver do the color conversion (i.e. using Epson color mode with default settings) and once with the ICC profiles that came with the driver (鈥渃anned profiles鈥).

According to what I saw the Epson color mode seems to use the perceptual table of the canned profile, doing some mild image enhancement in addition that makes the images slightly more saturated and contrast. Comparing the canned profiles and my custom made ones, most of the color differences I could see are not worth mentioning.

There was one problem though. The canned profile's perceptual rendering intent likes to shift dark saturated blues to cyanish tones quite a bit, whereas the profile I created preserves cleaner deep blue. Both decisions are valid, and the choice is the matter of aesthetic preference; the problem is that that cyan shift is not previewed when proofing colors in Photoshop. This result was the same on all papers; eventually the RGB-PCS and PCS-RGB tables in all canned profiles do not match. Pity, softproofing is one of the main reasons the canned profiles are there in the first place. This trouble aside, my conclusion is that unless you are really unhappy with the default dark blue rendering of the perceptual intent, or you're printing on non-Epson papers, you can safely stick to the canned profiles.

For those who intend to build profiles 鈥 a quick note on color stabilization. For any ink on any paper, it takes some time for the color to settle down, and we really want to profile the printer after the color it's just printed stopped changing. Now the question is 鈥 how long do we have to wait? Have a look at the graph below:

I printed some 140 color patches on Epson Premium Glossy paper and measured them every hour for 24 hours (well, I cheated at night and went to sleep). The X axis on the graph is the time, the Y is the average color difference between the measurements taken on that hour and the final (24th hour) one (in 螖E00 units). The changes in color during the 24 hours were minimal. The initial value of 0.9 means you most probably will not see any difference even on direct comparison. Value below 0.5 means nothing at all for any practical purpose in photography, so you're completely safe to measure your profiling target after about 4 hours.

Color gamut

I'll wrap up with a discussion of the color gamut 鈥 a bit of a tricky issue to discuss. The dry definition of color gamut is the range of colors printable by a given device, with the given inks on a given paper in a given environmental conditions, computed for a given type of illumination and a given type of observer. For practical reasons we usually ignore some of those "givens": we assume that the printing process is stable within a reasonable range of conditions and we assume standard illuminant conditions and the observer. So we're left with three variables: printer, inks and paper.

The next question to clear out is this: how do we know whether the gamut we measure off a given set of printer/ink/paper is good or not? The answer is: 鈥渨e don't鈥, unless we add an additional variable here: we need to define what we use the printer for. For example, if we use it to do color proofs for an offset press, we'd want our gamut to enclose the gamut of that press. In fact, we'd love the inks to be as similar as possible to the offset inks we use; it will save lots of headache in preparing the proofing profiles (anyone who used an Indigo digital press to proof for offset would appreciate that). Another example could be that all you ever print is black and white pictures; in this case you'd want your gamut to be as small as possible. One can buy inks dedicated to this purpose (known as 鈥渟mall gamut inks鈥). In fact this is the effect achieved by the R3000 when using gray inks with small addition of CMY.

Now, obviously in this review I have to figure out how good is the gamut of the R3000 for photographers who print pictures they shoot in random conditions by random cameras. I need to find some reference gamut I can compare the R3000 to: if the reference gamut is smaller than R3000 by a reasonable amount we're good; otherwise 鈥 we're not. AdobeRGB's gamut is often taken for such evaluations, because it is the de-facto photographer's standard working color space. Here is the result on Epson Premium Gloss.

Top: 3D visualization of gamut boundary in CIELAB. Cyan solid: Epson R3000, Epson Premium Gloss, 1440 dpi. Purple meshed solid: AdobeRGB. Bottom: 2D cross-sections of the same plot at different lightness levels 鈥 from the left: shadows (L* = 15); midtones (L* = 50); highlights (L* = 85).

The top 3D plot gives us the general impression about the relationship between the two gamut volumes, which is 鈥 AdobeRGB is way bigger. Who would think otherwise. The bottom plots are more informative as they give a more detailed account: it is bigger, but not everywhere. Most notably the R3000 misses lots of blues and purples in the midtones, and lots and lots of saturated light colors in the highlights (鈥減astel colors鈥).

What do we learn from this exercise? Not much, really, except for the fact that we should be careful while playing in Photoshop to not introduce these midtones blues and pastels into the pictures we're going to print (this is why the softproofing is so important). We also can't come to Epson complaining and asking to enlarge the gamut to these areas, because we are comparing here gamut of reflective object colors created by subtractive process with the gamut of self-luminous colors created by an additive process. Epson engineers are good, but the laws of physics might interfere with their attempt to please us.

Another option is to compare our R3000 to some other printer of the same class. This would probably be the most useful thing to do as it may also help us make a purchasing decision. But, as this is the first review in the series we have nothing to compare it with, so will have to wait.

But what if we try and evaluate the R3000 gamut in some absolute terms: is this gamut makes the machine a good tool to print photographs? What could be the criterion for this? Can we find some reference gamut that represents the world we photograph, and not some device or abstract color space? Fortunately we can: the color scientists in the University of Leeds spent some years measuring every material they could put their instruments on (in fact about the measured almost 86,000 surfaces), and built the Gamut of Surface Colors. If the R3000 has a color gamut which comes close to that created by the scientists at Leeds University its says a lot about its potential (you can find the original Leeds article here – paid download)

Here is what I've got from Leeds compared with the R3000:

Top: 3D visualisation of gamut boundary in CIELAB. Colour solid: Epson R3000, Epson Premium Gloss, 1440 dpi. Red dots: Gamut of surface colours (REF Leeds Uni). Bottom: 2D cross-sections of the same plot at different lightness levels 鈥 from the left: shadows (L* = 15); midtones (L* = 50); highlights (L* = 85).


I wonder whether there was any connection between the Leeds scientists and the Epson engineers: the match between the two gamut volumes is quite extraordinary. The biggest difference between the two is in light saturated colors, which means Epson still has some room for improvement in the color quality. If the Leeds Gamut of Surface Colors is indeed representative of all color objects around us, then the Epson R3000 can print colors of almost all the objects around us, at least on the Epson Premium gloss paper. This conclusion is a bit bolder than the one I would want to conclude my review with, so I'll ruin it a bit: there is some cheating in how this immense gamut is achieved. Have a look at the following graph:

Fluorescence graph – Epson Premium gloss paper

The graph shows the amount of light reflected from the paper relative to the light that falls on it, for each wavelength in the visible spectrum. Note that the purple line goes above the value of 1 almost all the way, meaning the Premium Glossy paper reflects more light than falls on it. How come? Fluorescence. The Optical brightening agents in the paper absorb energy in the invisible UV part of the spectrum and convert it to visible light energy. This is what makes this paper so brilliantly white and, in combination with the high-saturation Epson inks, this is what helps achieve the large gamut. The implication of this is that in a UV-free environment (controlled illumination in museums, or under UV-blocking glass) some of this color brilliance will vanish. Does it make the Epson R3000 a bad printer? Really not: the R3000 still does print darn good pictures.

Conclusion

The R3000 is an excellent printer, capable of printing in the highest standards. It would be nice if now after the print quality have reached such a high level some of Epson's resources will refocus on improving the design and the driver user interface of their products.

Pros:

  • Very good printing quality.
  • Large color gamut.
  • Generally trouble-free operation.

Cons:

  • Design can be better 鈥 of the printer itself and of the driver interface.
  • The switching process of black inks for matte to glossy paper must disappear.

Setup:

  • 15鈥 Macbook Pro, i5, 8GB of RAM, OS 10.6.
  • Profiling with Xrite ProfileMaker + I1 Pro.
  • Printing from Photoshop CS5.

Papers on test:

  • Epson Premium Glossy Photo Paper: high-gloss 255 gr/m2, high whiteness.
  • Archival Matte: relatively inexpensive 192 gr/m2 paper for proofs, smooth matte surface, high whiteness.
  • Hot Press Bright: heavy 330 gr/m2 fine art paper, smooth matte surface, neutral white.
  • Hot Press Natural: 340 gr/m2 fine art paper, slightly textured surface, warmish white.

Special thanks to Epson Israel for providing the printer and to Getter Graphics for the Monaco GamutWorks software.

讟注谉 注讜讚 讻转讘讜转 拽砖讜专讜转
讟注谉 注讜讚 讘 English

转讙讜讘讛 讗讞转

  1. Lola

    21/06/2011 讘 1:55 PM

    great article, now I'm more confident in my next printer purchase, guess what will it be:)

    讛砖讘

讻转讬讘转 转讙讜讘讛

讛讗讬诪讬讬诇 诇讗 讬讜爪讙 讘讗转专. 砖讚讜转 讛讞讜讘讛 诪住讜诪谞讬诐 *