A Feast for the Fridge: Printing Digital Pictures

0 Shares

Reprinted from TidBITS#700/06-Oct-03; reuse governed by Creative Commons license. TidBITS has offered more than fourteen years of thoughtful commentary on Macintosh and Internet topics. For free email subscriptions and access to the entire TidBITS archive, visit www.tidbits.com.

by Charles Maurer

Photos on a computer may look nice, but they're hard to tape to a refrigerator door. Sooner or later, most people who buy a digital camera hanker for prints and a photo printer – and then for aspirin once they start trying to figure out which one to buy. Every model sounds wonderful and every article reviewing them says that different models are best. In this article I shall to try to sort out some of this confusion. I shall explain how photo printers work, what to look at and what to ignore, and how to get the most out of them. Toward the end I shall discuss my own purchasing decisions and review several printers. Along the way I'll mention some useful software as well.

Overview of the Technology — Let's begin with some basic technology. Heat a transparent yellow ribbon until dye comes off and transfers to some paper, then heat a magenta ribbon, then a cyan ribbon, and finish off with a transparent one. The three dyes will form every colour and the transparent one will apply a protective film. Since the dyes are transparent, the colours will not be particularly dense, but no dots will be visible and the pictures will look like conventional photographs. This technology is called dye sublimation.

Dye-sublimation printers are the simplest to use and maintain. You merely snap in a cartridge containing a parti-coloured ribbon and replace the cartridge when the ribbon runs out. They cost more per print than ink-jet printers but only if you always print enough to use up your ink cartridges before they dry out. They cannot produce the ultimate in quality, but the prints resemble those from a commercial photofinisher and they are tough. They are ideal for snapshots and for 8″ x 10″ prints that will be handed around.

Although dye-sublimation prints can be very good, the highest quality possible comes from inks. Ink-jet printers use the same three colours but their inks are more opaque, so these printers do not superimpose the colours, they apply dots of colour adjacent to one another. This means that dark greys formed by three inks would have to incorporate yellow dots. Those yellow dots would limit contrast and soften edges, so ink-jet printers incorporate a fourth ink, black, into dark greys and blacks.

Cheap ink-jet photo printers use just those four inks: yellow, magenta, cyan and black. Those four are sufficient to form a complete range of colours and can be fine for snapshots, but if you look closely at light tones, you will see dots. That's because …

… to make a light tone from a primary colour requires surrounding a brilliant dot with a lot of white. To avoid dots requires using light-coloured inks for light tones, so better photo printers add light magenta and light cyan to the other four.

Those six colours can be sufficient to produce colours approximating the state of the art, but to achieve this quality the inks must be formulated and applied with unusual subtlety and consistency. It is possible to do this. HP manages it with a line of printers that are self-calibrating for colour. However, it is more profitable to accept weaknesses in various colours and to design printers that incorporate additional inks to compensate. (The more colours required, the more ink will be wasted in cleaning and the greater the probability of waste from drying out. Printer companies' profits come primarily from the ink. According to The Economist, HP's profit margin on ink is 80 percent, a markup of five times the cost.) The most common extra colour is a light grey, which is used in lieu of any colours to avoid tints in black and white.

http://www.economist.com/ displaystory.cfm? story_ id= 3817267

Inks can be made with either dyes or pigments. Dyes soak into the paper, pigments lay atop it. Both can be made comparably bright and durable, and for maximum durability, both require papers that are specially mated to the inks. This mating is not at any level we can see, it is intercourse at the level of physical chemistry. Dyes are more finicky about this than pigments, so pigment-based systems work with a wider range of papers, but pigments are more likely to clog printheads, so pigments tend to be more expensive to use.

Sharpness — All photo printers can print lines as fine as the human eye can see, and they can all produce a black that is reasonably black, so there is no physical reason for a difference among them in detail or in sharpness, assuming comparable paper and (more about this later) a comparable colour profile. However, whenever a printhead prints a line, it is following instructions from a printer driver. For large prints, the printer driver must tell the printhead to print more pixels than exist in the file. The driver interpolates those extra pixels by applying some form of running average. At a sharp transition – at the edge of a line – any form of averaging mixes the two sides of the transition, so a running average is guaranteed to blur edges. One averaging algorithm may suit one photo better than another, so one printer may appear to be sharper on any given test.

Extracting maximum sharpness from a printer requires using a more sophisticated interpolating algorithm than any printer driver is likely to employ. For the Mac, the best interpolator I know of is PhotoZoom Pro. (There is also a basic version of PhotoZoom, but its sharpening cannot be switched off, so it cannot be used with FocusMagic, FocusFixer or the “smart sharpen” feature of Photoshop CS2.)

http://www.benvista.com/ main/ content/ content.php? page= ourproducts&section= photozoompro_ 1

To achieve maximum sharpness, you need to feed the printer files that have been interpolated to the resolution that it can handle best. For a dye-sub printer that resolution is the number that is advertised, typically 300 dpi (dots per inch) or 314 dpi. For an ink-jet printer it will be a similar number but there is no way to ascertain it from the printer's advertised specs. To determine the effective resolution of a printer, I prepared (and George Reis cleaned up) a set of test files (linked below). Follow the instructions on the image and print the files without scaling them to fit the paper. There is a good chance that 288 dpi or 300 dpi will look the best.

http://www.tidbits.com/ resources/ 810/ PrinterSharpnessTest.zip

Note, by the way, that the finest resolution of a printer may not give the best photographs. Different resolutions are likely to produce different densities of ink on the page, which can have unpredictable effects. You'll need to try different settings on photos to see which you like the best.

Colour — At first blush it seems as though the first factor in choosing a photo printer ought to be its capability to reproduce colours. However, there is no practical way to determine this because there is no accurate way to measure colour, as I noted in “Colour & Computers” in TidBITS-749. Moreover, even if there were a way to measure colours, different dyes and pigments produce different sets of colour, so accurate reproduction would still be a chimera.

http://db.tidbits.com/ getbits.acgi? tbart= 07840

To understand why, let's pretend that it is possible to measure colour accurately and invent some measurements. Let's say that the sensor in your camera can produce voltages in response to 95 units of red, 100 of green and 103 of blue, and let's say that the inks of your printer can generate 90 units of red, 101 of green and 122 of blue. How can do you translate accurately the camera's response into the printer's? How do you translate 95, 100 and 103 into 90, 101 and 122?

To generate the colours on a printer (or on a monitor), somebody sits down at a computer and draws up tables that try make that kind of equation. Somebody figures out what mixtures of colour look best overall – whether it's best to mute everything to the limit of the printer's red and keep all the colours proportional or whether it's best to keep some of them proportional but give the blues full rein. The process is not mathematical, it is cooking fudge or translating poetry from Greek to Portuguese. Inevitably the results will look better in one circumstance than another, depending upon the subject, the room light and – critically – the surroundings, for the context of an object is critical to the colour we perceive. Without contextual cues we would see red instead of brown.

Once somebody comes up with a scheme for equating these colours, he puts the tables into an application or the printer driver or the operating system, where menus enable you to access them. Usually, although not always, the colour-equating tables fit a standard format of the International Color Consortium and thus are called ICC profiles. Most companies selling hardware provide ICC profiles for their hardware, and so do third parties. The Macintosh also comes with some and Photoshop installs a lot of them. The multiplicity of profiles and their means of access is thoroughly confusing, but you need to sort it all out. You need to sort it out not only because you want to find the right one to use, but also because there is nothing to prevent you from instructing the computer to apply multiple profiles atop one another, which yields a mess. You need to determine the source of each profile that appeared on your drive and experiment to find the best ones for your purposes and the most convenient way to assign them. You will probably tear out some of your hair doing this – mine is decidedly thinner now – but at least in the future you will spend less time at the barbershop.

Whenever you compare prints from different printers, you are actually not comparing the printers so much as colour-equating tables. Moreover, since no colour-equating table can always be optimal, a printer that looks better with a landscape may look worse with a portrait. From what I have seen, every photo printer on the market is capable of good results if the colour-equating tables are appropriate, but no colour-equating table can always be appropriate. On top of that, printers are mechanical devices that inevitably differ slightly in ways that affect the application of ink. Because of these differences, any two printers of the same model, using the same colour profile, are likely to give different results (unless the printers calibrate themselves to a uniform standard, as a few HP printers do). For these reasons, when shopping for a printer, I can see no reason to compare the colouration of different prints.

To get the best results out of any printer you need to control the colours yourself. This requires making the monitor's colours resemble the printer's colours closely enough that by controlling the one you can control the other. To achieve the optimum resemblance between monitor and print requires custom-made colour profiles for your monitor and/or printer. Many people will be pleased to take your money to make these profiles or to sell you special hardware to help you make your own. If you have several employees running several machines, then you may want to pay them to achieve a consistent result, but for most people there is no need. I suggest that you make page-sized prints from the first link below using your print driver's built-in defaults and any additional ICC profile that may have come with the printer. If you have a choice, keep the nicest of them and use that profile for all your work. Next use the $19 SuperCal to calibrate your monitor. Eventually SuperCal will ask you to balance the colours on a portrait: when it does this, substitute the test file for the portrait and make that test file match your print as best you can. When you make the best match, your monitor will be calibrated as well as it can be – calibrated especially to your eye and to the colours that you deem the most important.

http://www.drycreekphoto.com/ tools/ test_ images/ DCP- TestImage.zip

http://www.bergdesign.com/ supercal/

You must still learn how the pictures on the monitor translate into prints. This will vary with the qualities of your printer, your monitor, and your eye. It is not a physical match, it is a perceptual approximation that must be learned and that interacts with the size of the print. Also, to compare prints properly to the monitor, you must have a suitable lamp. Use either an old desk lamp that combines a circular fluorescent lamp with an ordinary 60-watt bulb, or buy a fluorescent tube that meets graphic-arts standards (like one in the Ott-Lite TrueColor line) and put it in a cheap fixture. Use this lamp when you calibrate your monitor.

http://www.ottlite.com/ productsview.asp? product_ type= bulb&product_ phosphor= truecolor&category= truecolor

If all of this sounds less precise than advertisements promise – well, it is less precise. It is so imprecise that Photoshop and Preview offer two different matches of monitor to print, the ordinary way you use while editing and a soft proof. The colours on your screen look one way when you are concentrating on them alone and another way when you are comparing them directly to a sheet of paper. Conventional techniques for profiling monitors implicitly assume the former; soft proofing applies an alternative set of colour-matching tables geared to the latter. If you profile your monitor as I suggested, you will see soft proofs all the time.

One last parameter that causes confusion is colour space. Colour space is the gamut of colours “understood” by your computer. It is a theoretical range of colours denoted by numbers. If you want to match a particular colour, then if you could measure it accurately, you could denote it by a specific number, and if your printer could produce colours accurately, you could specify that number to reproduce it. This technique can be useful when trying to print a catalogue (although you know how useful if you have ever ordered clothing from one) but it bears no relevance to pictorial photography. As I explained last week in “Reality and Digital Pictures,” the eye interprets contrasts, not absolutes. If your camera records a red of one-half its maximum, then it would sensibly be printed as approximately one-half of your printer's maximum, however red that happens to be.

http://db.tidbits.com/ getbits.acgi? tbart= 08365

If your printer offers profiles for different colour spaces and they give different results when used with the appropriate colour space, that is not because the colour space is different, it's because the different colour-matching tables match colours differently. This is likely to happen because different colour spaces are used in different markets. Most people tend to print small pictures and stick to their computer's built-in colour space of sRGB; professional photographers tend to print larger pictures and service the graphic-arts industry, which commonly uses Adobe RGB. Smaller pictures tend to benefit from more punch at the expense of subtle gradations in mid-tones; larger pictures tend to benefit from more gradual transitions across the tonal spectrum. In any case, what matters for prints is the colour profile, not the colour space. Use whatever combination of colour space and colour profile that gives the results you prefer. If in doubt, choose sRGB. It is the industry's standard so its use is likely to cause fewer complications.

Blacks and White — Although most photos are in colour, many pictures are more effective in black and white. Most printers produce black-and-white prints in the same way they produce colour prints. The only difference is that the computer instructs the printer to print equal amounts of all three colours at each intensity. Unfortunately, since there is no way to measure colour, there is no way to generate equal amounts of three colours at every intensity, so the greys inevitably sport tints. Printing with black ink alone is not satisfactory because it looks spotty wherever it is applied thinly enough to form light greys.

The best and most expensive workaround is to print with several shades of black and grey. Some of the latest Epson and HP printers do this. Second-best is to use the normal inks but balance them differently, fudging the balance in a way that may produce bizarre colours but will produce greys that look more consistent to more people under more circumstances. This approach requires more consistency than is available from most printers but it can be done well with the models from HP with self-calibrating colour.

Paper — As I mentioned earlier, for maximum life expectancy, ink-jet paper needs to be mated to the ink at the level of physical chemistry. Tables you can find at the link below show whopping differences in longevity from one paper to another. Other important differences are physical toughness. For example, Epson's best glossy is easier to crease and rip than HP's, which is weaker still than the paper for the Olympus P-440 dye-sublimation printer. Also – this is usually less important – prints on some papers are highly soluble in water unless sprayed with lacquer. HP's paper is bad in this regard, Epson's is better, dye-sub paper is excellent.

http://www.wilhelm- research.com/

The most obvious attribute of photographic paper is its surface. If you want to maximize attention on the image, then you want no texture on the surface to distract the eye. If you want to maximize the density and saturation of colours, then you want no diffuse glare to wash them out, you want the glare localized enough that you can adjust your position to let your eyes can see around it. In short, to bring out the most from a photograph, the paper should be smooth and glossy, just diffuse enough to prevent it from looking like a mirror.

If this doesn't square with the notion that fine art requires a matte finish – well, walk around an art museum and look at the pictures on the walls. Every picture done on paper will show a glossy surface. The paper may be matte but at normal viewing distances, the surface you notice is the surface of glass, glass in a picture frame. Curators could use so-called non-glare glass, which has a matte finish, but they do not because diffuse glare detracts more than specular glare. For the same reason, oil paintings are coated with glossy or semigloss varnish, not matte.

Most people print photos on paper that's cut into sheets but if you print a lot, you might consider using paper that comes on a roll. Roll paper is half the price of sheets. The pictures come out curled, but the curl is easy to tame. Just let prints sit overnight rolled backwards around a tube sandwiched between two sheets of 10-mil polycarbonate.

When printing a lot of pictures I find the $50 application Portraits & Prints Pro to be invaluable. Like several other applications, it employs templates for printing pictures, but Portraits & Prints Pro differs in enabling a user to create a template of any size. If the template is for one picture filling one sheet of paper, then you don't need to set Page Layout for each picture, and large templates allow gang-printing pictures on roll paper. Unfortunately, Portraits & Prints is not particularly stable. I have learned to save my work after every change.

http://www.econtechnologies.com/ site/ Pages/ Portraits_ Prints/ pnp_ overview.html

Size — Prints look surprisingly different when enlarged to different extents. To see an example of this, download the two files below to your hard disk, launch Preview, select them both, then drag them atop Preview's Dock icon. This opens them in two pages, so that you can flip between them to make an instantaneous comparison. Make sure Actual Size is selected in the View menu, switch to the larger image, and then click the green zoom button to make the window the same size as the image. Look at the larger picture with your eye as far from the screen as the diameter of the photo – that is the perspective I intended – and then, without moving or looking away, click to the smaller image. The spider suddenly becomes flatter in colour, flatter in contrast, and flatter in depth. In addition, you can no longer see many strands of the web and the background becomes more distracting.

http://www.tidbits.com/ resources/ 810/ 00_ 03557_ 1200×800.jpg
http://www.tidbits.com/ resources/ 810/ 00_ 03557_ 600×400.jpg

To the eye, the size of a photo is not centimetres or inches, it is the proportion of the visual field that it fills, the visual angle. Since angular size is what matters, I like to make my pictures large enough that people will usually view them from similar angles. Since I am now taking pictures for pleasure, not to fill holes in advertising layouts, I aim for the size that is optimal perceptually, which is about 45 degrees, the same distance as the picture's diameter. I have found this not to be practical with page-sized prints because unless I am showing them to a single person sitting at a desk, they end up being viewed from too far away. My normal size is now 11″ x 17″ (with broad borders). This is a standard size in North America, roughly equivalent to A3 elsewhere, the size of a tabloid newspaper. For this size the diagonal is close to 20″ (50 cm) and the prints are large enough that two or three people can and will naturally arrange themselves to see them from reasonably close to that distance. It strikes me as the point of diminishing returns. Compared to the next larger size, usually 13″ x 19″ or A3+, 11″ x 17″ prints are more economical and less awkward to handle yet hardly different in effect.

Pictures larger than that must be hung on a wall, which introduces another factor: people sometimes walk up to them closely, to examine details. When they do, they will always come to a point where the detail breaks up. The natural result is disappointment. Oil paintings let down the viewer gently by breaking up into brush strokes that add nothing to the detail but suggest or complement it or provide some interest in their own right. In contrast, photographs usually become grainy and blurred. To make a very large photograph that does not disappoint up close requires the photographic equivalent of brush strokes – it requires concentrating the lowest level of detail into pieces that suggest the normal appearance. This can be done digitally.

You can see seven different approaches to this in the examples linked below. Each of these photos is blown up to show on your monitor approximately the detail that you would see in a 20″ x 30″ (50 cm x 75 cm) print. Reduced to 33 percent, they will show approximately the detail you would see at a normal viewing distance. For comparison, I preceded those seven with a picture scanned from film, from a 2-1/4″ x 3-1/4″ (6 cm x 9 cm) transparency. That one I took for a publicity poster with a small view camera (and supplied the sunrise with electronic flash).

One thing that is not required for large blow-ups is extreme resolution. Of these eight pictures, the scan from film blows up the least well yet it resolves finer lines than any of the others. The seven digital images are enlarged to double their original size, so the finest lines any of them can possibly resolve are two pixels thick, yet this URL shows the scanned image resolves one-pixel lines:

http://www.tidbits.com/ resources/ 810/ 8_ feedingchickensmagnified.jpg

All of those digital images came from a Foveon sensor with black-and-white resolution comparable to a conventional Bayer sensor advertised as six megapixels. (Manufacturers' megapixel claims are grossly misleading. See “Sense & Sensors in Digital Photography” in TidBITS-751 and the errata at the second link below for background on sensors and a realistic discussion of resolution.) The images enlarge well not because they have extreme resolution but because they lack noise, because minuscule details have clear contrast, and because lines remain distinct even where contrast is low. To minimize noise and to control local contrast I used Noise Ninja; to control edges I used “Smart Sharpening” in Photoshop CS2 or unsharp masking in PhotoZoom Pro, either with or without PhotoZoom Pro's anti-aliasing. Images from the Foveon sensor are particularly well suited to this treatment because their microstructure is uniquely sharp.

http://db.tidbits.com/ getbits.acgi? tbart= 07860
http://db.tidbits.com/ getbits.acgi? tbart= 07906
http://www.picturecode.com/

My Own Printers — The first digital camera I bought was a pocket-sized point-and-shoot. From it, I could not see printing anything larger than a postcard. Since my usage would be occasional, I did not want an ink-jet printer with cartridges and heads that would dry out. Instead I bought a dye-sublimation printer. I could find no sensible way to compare the various offerings, except on price, so I bought the cheapest one that I could find in a local shop, both the cheapest to purchase and the cheapest to feed. It happened to be a Canon CP-200. It is simple to use, and it produces post-card sized prints with colouration that is appropriate for their size. I have no complaints with this printer and still keep it around because it does a nice job and because for the odd small print, I prefer to use it than to change the paper in a larger printer.

When I bought a digital SLR, I bought a larger printer. I still expected to use it erratically, so I still did not want an ink-jet, but this time I wanted the largest dye-sublimation printer that I could buy. Two were available that could print 8″ x 10″ prints. Again I could find no sensible way to compare them. Since both were used almost exclusively by professionals, I had to assume that either one would do, so I bought the cheaper, an Olympus P-440. I described this printer last year in “Colour & Computers” in TidBITS-749. I shall not repeat myself here except to say that it also works well. It comes with a ColorSync profile that strikes me as mediocre under most circumstances, but the printer driver offers two good non-ColorSync presets, a normal one geared toward warm pictures of people and another one that intensifies greens and is better for scenery.

http://db.tidbits.com/ getbits.acgi? tbart= 07840

I did not expect to replace the Olympus printer but the law of unintended consequences prevailed. The better I got to know the Foveon sensor in my SLR, the more impressed I became with its capabilities. I found myself taking pictures that would have been inconceivable with film. So many photographic horizons opened up that for the first time in 30 years I began to haul along a camera bag while travelling on holiday. Since I happened to be visiting some unusual places, I ended up with a lot of photos that are not your brother-in-law's snapshots. I wanted to be able to print them large enough to share comfortably with other people and to be able to print big ones cheaply enough to feel comfortable giving them away to friends. In short, the Foveon sensor turned out to be so good that it led me to shop for a larger printer.

I also wanted a better printer for black-and-white images. One weakness of the Olympus, like most digital printers designed for colour, is its handling of black and white. I wanted to be able to print black-and-white enlargements like the ones I used to get from my enlarger. For this reason I looked first at printers that offer light grey ink, but I did not like any of them. HP's consumer models with grey ink are very attractive, especially the 13″ x 19″ Photosmart 8750. This supplies almost every feature that anybody could want, including an Ethernet port, at a remarkably reasonable price ($500). However, it is also slow, and I really did want something larger.

Epson offers larger and faster models with grey ink, but they can cost an outlandish amount to feed. Epson's printheads are notorious for drying out, cleaning them uses a significant amount of ink, changing one cartridge wastes a lot of ink from all of them, and Epson states the life expectancy of an opened cartridge to be only six months. If you make prints constantly, then an Epson's running costs might become reasonable, but if you print infrequently or erratically – well, for eight ink cartridges replaced twice annually at $70 each, the guaranteed minimum cost of ink is $1,120 per year. (Although those figures are for large printers used by pros, the same problems are likely to be evinced at a smaller scale by smaller models. I'll talk more about consumer printers below.)

Aside from those printers, the only ones well suited for black and white are three professional HP models that are self-calibrating for colour: the Designjets 30, 90 and 130 ($700, $1,000 and $1,300 without Ethernet ports or roll-paper feeds). These three are the same except for the maximum width of paper they will handle, which is 13″, 18″ and 24″ (34 cm, 46 cm and 60 cm) respectively. Compared to Epson's equivalents, their ink costs slightly less per drop, changing a cartridge wastes no ink, the cartridges hold less ink so that intermittent use is less likely to see cartridges dry out, cleaning is rarely necessary and uses negligible amounts of ink, and the printheads are not guaranteed to dry out twice a year. On top of that, their purchase price is less. Epson's 17″ (43 cm) Stylus Pro 4800 costs $2,000. The 18″ Designjet 90 equipped similarly costs $1,500.

I was tempted by the 13″ Designjet 30, because compared to the Photosmart 8750 it is much faster, it costs only $200 more to buy and, as we shall see below, it promised to save that in running costs in a few months. The Designjet 30 would have been the most sensible choice because I would only occasionally want to print larger pictures and it would be cheaper to send them out than to buy a larger printer. Indeed, I wouldn't even need to send them out because I have after-hours access to a 44″ (1.1 m) Epson. However, whenever I have used that Epson, I have found it awkward to rebalance the colours to take full advantage of it. After a lot of dithering I decided to buy the largest Designjet with a roll-paper feeder and an Ethernet card (model 130nr, $1,900).

For my erratic usage at home, the difference in purchase price and running costs between HP Designjets and similarly sized Epsons promised to be so extreme that every other consideration paled by comparison. However, if I were printing daily in a business, then the running costs would be more similar and I would have considered other factors. The Epsons are more heavily built and can handle more kinds of paper (although I happen to like one of HP's papers better than any of Epson's), and HP's papers are more easily damaged by water. On the other hand, HP's best glossy and semi-gloss papers are stronger and stiffer than Epson's.

The Designjets come with two sets of colour-matching tables, a set of ICC profiles buried somewhere in /Library/Printers/hp/ and a set of proprietary profiles accessible only through the Print dialog. They appear in the Print dialog as ColorSmart/sRGB. The two sets give slightly different results. ColorSmart/sRGB seems more suited to snapshots but I prefer the ICC profiles for large prints. Both work well for sRGB and Adobe RGB alike. The printer also handles black-and-white images well using a different profile, a profile that fudges greys instead of colour. HP's Web site offers a broad choice of profiles for black and white free for the downloading. You can choose whatever tint that you like. Alternatively, Neil Snape sells profiles to produce prints on the Designjets 30/90/130 that are photometrically equivalent to black and whites from the 8750. I prefer his.

http://www.neilsnape.com/

The weak part of the Designjet 130nr is the pair of paper feeds on the rear, the feed designed for rolls and a feed just above it that can be used for cut-offs from rolls or for paper too stiff to be fed in through the normal sheet-feeders in front. The printer grabs and feeds the paper automatically but this mechanism is fussier than a two-year-old child. There ought to be some manual way to get the paper started, but there is none. This is particularly problematic if you need to rip the paper to clear a jam, because the printer wants the end of the paper to be cut dead square. Indeed, sometimes it has deemed an end that it trimmed itself to be out of square.

For most people even the Designjet 30 will be too expensive but HP makes a range of consumer photo printers using similar technology that ought to give similar results. These are models with at least six colours that use cartridges from this list: 84, 85, 94, 95, 96, 97, 99 and 100. As I mentioned, the Photosmart 8750 Professional is particularly interesting. According to people who own both of these printers, their prints are not quite identical but are so close as to make no difference. Their speed and running costs are very different, however.

Although HP calls the Photosmart 8750 “Professional”, anybody in business would happily pay $200 more for the Designjet 30, because the extra cost would rapidly be recouped in lower running costs. To print a colour photo on good photo paper with 90 percent coverage (which is probably representative of the real world), HP claims the cost of ink consumption and printhead wear on a Designjet to average $1.01 per square foot. That does not count wastage, but there is virtually no wastage with the Designjets. HP's numbers seem in line with others' measurements and with my own usage, so I am prepared to believe them. With a similar rate of ink consumption by volume, the equivalent figure for the Photosmart 8750 calculates to be $2.03, but the cost would in fact be even greater, because the Designjets use one colour per cartridge while the 8750 uses three: since the three colours will never be consumed at the same rate, a certain amount of waste is guaranteed with the Photosmart 8750.

http://h20000.www2.hp.com/ bizsupport/ TechSupport/ Document.jsp? lang= en&cc= us&objectID= c00238851

I have been able to find no comparable information for any printers by Epson. Epson does provide some specifications, but they are highly misleading because they are based on 40 percent ink coverage and take no account of cleaning cycles. A realistic average coverage is double that, and cleaning cycles can waste more ink than printing. Depending upon how often you clean a consumer Epson printer – which can mean how often you switch it on – it may cost less for ink than HP's equivalent or it may cost a good deal more.

Wrapping Up — In sum, for most people and most purposes there is no valid way to compare print quality – obvious differences in quality come primarily from the photographer, not the printer – so I cannot see any reason to try to compare print quality when choosing a photo printer. Any modern ink-jet printer with at least six colours ought to be capable of excellent results, as should any dye-sublimation printer. I also have no idea how to tell what printer is likely to be more reliable and durable than another. I don't even know how to begin to find out, so I cannot see any reason to worry about that either. It seems to me that the only discernible differences among photo printers are (1) size and speed, (3) durability of the prints (as tested by Wilhelm Imaging) and (3) running costs.

Of these three, running costs are likely to be the most important criterion. It is not possible to learn how to make good prints without making a lot of bad ones. A print ought to be cheap enough that you won't mind throwing it away and trying again. Moreover, if you don't print a picture because the ink dried out and you don't want to buy another set of cartridges because you don't expect to use them up either – well, it does not matter how many centuries the photograph might have lasted if you never print the thing, and the printer's size and speed become immaterial as well. Good ink-jet photo printers can produce stunning results but they are not always practical. If you do not expect to print a lot, consider a dye-sublimation printer instead.

PayBITS: If you found Charles's explanation of printing helped you choose or use a printer, please support Doctors Without Borders. Read more about PayBITS: http://www.tidbits.com/ paybits/