Low cost, wide format, wide gamut, inkjet printers are now able to produce a colour gamut double that of offset presses, reports Paul Lindstrom from graphic arts research agency Digital Dots
Colour image quality is clearly of interest with the new standard ISO 12647-2 coming to Australia and being adopted around the world, and the latest 11- and 12-colour wide gamut wide format printers are at the centre of this interest. Looking at samples from these printers in good viewing conditions reveals an astounding colour gamut, really vivid colours and a high contrast that makes the photos almost appear to be in 3D. A question comes to mind: are we now able to reproduce a photo with its entire gamut? We decided to test a selection of wide gamut printers to find out.In previous tests we have found that many inkjet printers can not only reproduce all the colours of high quality offset printing (and for that matter gravure printing), but also many of the spot colours. This is definitely the case with 11- and 12-ink printers on high quality papers. We have tested wide gamut printers from Canon, Epson and HP to find out what gamut volume can be achieved in those amazing machines.
The Canon iPF 5100 and HP Designjet Z3100 have a quite similar ink and printhead setup: an extended CMYK setup with light magenta and light cyan, plus light grey. In addition these printers also have red, green and blue ink, to extend the gamut closer to that of camera and monitor RGB.
The Epson Stylus Pro 7900 has a slightly different approach. You could say it has the capacity of Pantone Hexachrome printing, with the addition of extended CMK (light cyan, light magenta and light black). The Hexachrome part consists of cyan, magenta, yellow, black, orange and green, where the cyan contains more vivid pigments than that of normal offset process cyan. Epson has for some years now used a more vivid magenta, so all in all the Epson 11-ink set reaches a very wide gamut. The light inks don’t add much to the gamut volume, but improve image details in the highlights as well as offering smooth tone gradation without banding.
All of the three printers that we tested reached about twice the colour gamut of offset printing, or around 800,000 colours, versus the approximately 400,000 colours possible when printing according to ISO 12647 or SWOP on high quality coated paper. Even if you use high-pigmented ink, like the Toyo Kaleido ink, for CMYK printing in offset, you would only reach about 480,000 colours. We use the word only here, but that’s when comparing to wide gamut inkjet printers – the visual result of quality offset is of course still very good, especially when not comparing the original image or a photo printout side by side with the offset print. As we can see photographic colour prints reach about half of the colour gamut of high quality monitors, but then we must remember that we are comparing two different colour systems. RGB encoding assumes a self-illuminating image source (light emitting and coloured directly to the eye – the additive colour system), while the printing process creates colour as a result of light reflected from a substrate, coloured by the ink on that substrate’s surface (the subtractive colour system).
The original image, projected onto a calibrated wide gamut monitor, shows a surprisingly good match with the photo quality wide gamut printout from the inkjet printer. That is, viewed in a proper viewing booth, with, let’s say, 5000 K reference white and high brightness (luminance) of around 1500-2000 Lux. What might differ is that a well-calibrated monitor often reproduces shadow details better than a printout, but that has less to do with colour gamut than with correct linearisation of the printer, and correct setting of ink density. The latter is a task for the rip system driving the printer – more on this later.
Calibration challenge
When testing these printers we were facing the same question as any new user of an advanced colour inkjet printer – how far would I get with the standard (and free) printer driver, or should I use an additional software rip for an optimised and more controlled result? We printed some test images straight through the driver, using the default paper settings, and then decided to make some custom ICC-profiles for the same substrates, including new linearisation and ink settings.
The standard driver and default paper settings in general produced a decent result, and actually quite impressive images, since we used high quality papers (suggested by the vendors), and all the printers as such are capable of both high gamut and high-resolution output.
But when analysing the test prints carefully, we thought we might perhaps be able to achieve an even better result, a better match between the original photo, and the printout. Especially in regard to better shadow details, better grey balance, less tendency of banding and so on. Now the question was – which software rip to choose?
Asking the printer manufacturers gave some hints, but in general the printer vendors like to stay neutral in regard to what software rips to use as the front end, and are normally unwilling to clearly favour one single software rip vendor. In our test we came to use three different software rips to compare the results against those of the standard driver. Those three rips were CSE Colorburst, Efi Colorproof XF and Perfectproof Proofmaster. But we have to admit that the choice was quite random (there is a large number of excellent software rip systems on the market), and it will take a new and very elaborate test to cover all of them.
Still it provided us with some insight into how differently each vendor can approach calibration and linearisation of a printer. One would think that the procedure of calibrating an inkjet printer should be quite straightforward and well-established, but we found more variations on the topic than expected. The most challenging part seems to be to determine the ideal amount of ink to use on a certain substrate. The problem most rip vendors have is whether or not to use the default printer driver supplied by the printer manufacturer. While things get easier in many respects if you decide to use the original driver, it’s not always possible to control the printer to the degree you would like. But creating a new and specialised printer driver is of course both demanding and costly, so not all of the software rip vendors take that route. When using the default printer driver you are often limited to using the papers listed in the user interface. You can change the ink and paper and make a new calibration, but you often need to start with some default paper set-up, choosing a paper similar to the actual substrate. In our limited testing we didn’t always reach a fully optimised result with this method. We found that the best linearisation result was when using a specialised printer driver, supplied by the software rip vendor. The key is to be able to use exactly the right amount of ink, for that particular substrate. Not too little ink, which produces washed out colours, and definitely not too much ink, which brings all sorts of problems – such as having too heavy shadows, ink bleeding through the substrate, bronzing phenomena where the ink layer is too thick, very long drying times and so on. One good side effect of having more than three primaries, is that you will actually save ink. Instead of typically mixing yellow and magenta to create red you use the special red ink instead.
So instead of using up to 200 per cent ink in pure red areas you use only 100 per cent. The same applies to blue and green, if you have special ink for that in the ink set-up.
We hope to be able to test more rip systems systematically, and report our findings, but for now we can only conclude that a given printer can produce quite different results, depending on the front end driving it.
For less demanding workflows and applications the standard driver produces a reasonably good result, but for high volume printing, and for a high level of control, an advanced and high performance software rip is clearly a necessity. Such a rip needs to be able to properly calibrate the new 6- and 7-colour ink sets of wide gamut printers, or the printer won’t reach it’s full potential.
The future
It’s very difficult to envision a dramatic increase in gamut in the near future – with 5 or 6 ink-sets for the primaries, CMY and RGB, or CMYOG (CMY + Orange and Green), the gamut on high quality paper is astounding. All of the printers tested here have several ink heads for black (called light black, but actually grey), which allows for excellent black and white photo output. There will be development of higher pigmentation and perhaps an increased portion of fluorescence effect in the ink to give an even higher colour intensity. Development of substrates will also continue, while it’s perhaps not so much to reach higher colour gamut as such that is the goal, but to reach high gamut with less costly pigments and stock.
A development that is clear already now is to integrate a spectrophotometer into the printer itself. This both speeds up and facilitates the calibration of the printer, as well as helps maintain a high and even print result. Both HP and Epson have a built-in spectrophotometer in their top models, and we expect Canon to follow suit. But for high end ICC profiles you can’t rush the calibration process too much – before you measure the final test chart it should have dried some 24 hours. In that respect it’s perhaps less crucial to have the spectrophotometer built into the printer – it can just as well be a good stand alone spectrophotometer.
Another thing on the wish list is for Adobe to bring preview capacity of multicolour profiles into the Creative Suite. As of today you can create n-Color profiles in, for example, X-Rite ProfileMaker, but you need to install a special plug-in in Photoshop to preview the image, colour converted from RGB to the print output colour space. An accurate visual preview of how an RGB image will look when printed is something one would think had been solved by now, but there is actually room for improvement from all the parties involved.
Canon’s work on its Kyuanos technology is supposed to achieve a better match between screen preview and final output, even under different viewing conditions (colour temperature and spectral distribution in the light source). For this test we had hoped to include some initial testing of Kyuanos, but haven’t received any software from Canon to try it out. But any work on a better visual match between screen and final print is welcome, and we will come back to the topic as soon as we have more to report. The colour gamut from wide gamut inkjet printers, about twice that of regular offset printing, is a good starting point in the chase of the photorealistic reproduction of high quality images.
The maximum gamut of human vision
There are different ways to calculate the gamut that it is possible to see with the human eye, and over the years we have come across very different statements as to how many colours we humans actually can detect. Numbers varying from some thousands to many millions of colours have been suggested. At the moment we are quite convinced that the way the colour scientist Bruce Lindbloom calculates colour spaces makes sense (see www. brucelindbloom.com).
Somewhat simplified, because we are not fully capable of following all the advanced mathematics behind this, of all the possible tone values, Lindbloom only counts the colours that differ by 1 ∆E. When using this methodology, the human vision is capable of detecting approximately 2.4 million colours. This is far from the theoretical 16.7 million colours possible to encode with 24-bit image depth (eight bits per colour channel, as in, for example, RGB and CIE Lab). Using 10, 12, 14 or 16 bits per channel doesn’t increase the number of detectable colours – those extra bits are just used to make finer divisions between the tone values in the image processing.
The next question is, whether or not it’s possible to capture all the ‘real life colours’ with a digital camera? We use ICC profiles to check this, but when analysing an ICC-profile from a high end digital camera, we find its gamut to be quite a bit smaller, but very similar in shape, to that of the ProPhoto RGB, developed by Kodak. But the ProPhoto RGB has a gamut volume of 2.9 million colours, so quite a few of the theoretical values can’t be detected in real life by human vision, and are not likely to be reproduced with any known technology.
Even the new monitors with LED back-lit technology can’t reach a larger gamut than around 1.5 million colours. This number will probably be slightly higher within a few years, but we can conclude that using the Adobe RGB colourspace with its 1.3 million colours is still valid as a production RGB working colour space. But already now it will clip colours that can be reproduced by the best LCD monitors on the market.
This is one of the reasons why Bruce Lindbloom has suggested a new, and slightly larger production RGB, or working space as Adobe calls it in the colour settings of Creative Suite. The suggested colour space is called Beta RGB, and uses a white point of 5000K and a reference gamma of 2.2. The gamut volume is 1.7 million colours, and shaped so that printable colours are within the colour space (some printable colours fall outside the Adobe RGB).
For image processing it’s a good choice to work in Camera RAW and/or ProPhoto RGB, and 16-bit image depth, but processed production image files need to be saved and delivered in 8-bit format and in a colour gamut that is close in size to the gamut of the reproduction technology. For colour printers and printing presses this is again a substantially smaller gamut than for cameras and monitors.
