Over the weekend, Red River’s Drew Hendrix posted their latest results, covering Canon’s Pixma Pro9500 and Pixma Pro9000 inkjets. Their results show that the dye-based Canon Pixma Pro9000 offers roughly the same cost per print as the pigment-based R1900, while the Pixma Pro9500 has a slightly better cost per print than Epson’s R2400. (The Stylus Photo R2880, which was introduced in June, is probably much closer to the 9500 in cost per print, based on our tests, which use the same test image and similar methodology to Red River’s.)

We’re currently testing the HP B8850 with some slightly different methodology, but one that we hope will give us slightly more accurate results. Stay tuned—we think this is pretty important stuff.

This report addresses concerns and arguments about the true cost of ink in desktop photo printing. Using the Epson R2400 and Epson R1900, we conducted a series of print tests to determine how much ink is used in a full coverage 8”x10” print. From that figure we extrapolated ink usage per square inch. The objective is to share a realistic cost per print vision with inkjet users. The choice to pursue photo inkjet printing is in the end an individual economic choice.

We think this is pretty important stuff, and it’s good to see others working on similar tests — the data regarding the R1900 is especially welcome, and it looks like Red River’s results on the R2400 are very similar to ours, which speaks well to this style of test’s repeatability. If you’re interested in this topic, it’s worth going through the report.

Based on some of the comments we’ve received, we think there are a few tweaks we can make to get the test methodology a bit more secure, and be extended to HP and Canon printers. Stay tuned.

“in our initial testing, we were able to print nearly twice as many photos using the same amount of ink on R2880 as we were able to do with the R2400.”

Our results — which were based on observations from our second round of installed ink cartridges — weren’t done with any specific methodology in place. It was based on our general performance testing, alongside standard print quality test prints that we generally use. Since then, our initial findings haven’t held up, and while we stand by what we saw in that specific round of printing, it was obvious that we needed to look at this issue a bit deeper.

In the past week, we have spent a fair amount time trying to come up with some type of test methodology to give a realistic sense of ink usage on the new printer. This week we ran some initial tests, and we’ve gotten some interesting data, but this is an issue bigger than the R2880. We wanted to share our results as a starting data point for a wider discussion of ink yield in photo printers.

The ink problem

Determining ink yield isn’t an easy task; there are a number of variables that come into play that make it hard to come up with a reliable number that works in the real world. Some of the variables aren’t just built into the printer. They also include how the printer is used. Here are a few of the factors that are involved (in no particular order):

• The range and type of images you print. If you print more portraits, or in black and white, you might find that you’re using a much narrower set of inks more frequently than others.
• Print quality settings. If you print using your device’s highest quality setting, you’ll use more ink than you would at a lower quality level.
• Print output size. Relatively speaking, the larger the paper, the more ink that gets laid down on the page.
• Cleaning cycles and maintenance; some printers perform regular printhead checks that use small amounts of ink. (HP posted a firmware update earlier this year that reportedly fixed a problem related to overzealous cleaning of the printheads.)
• The swapping of matte and photo black ink cartridges (in the case of the Stylus Photo R2880 and R2400); this also wastes ink.

Measuring — and comparing — ink yield

We could just say that it’s hard reliably to determine ink yield and leave it at that, but this is a hot-button topic. And, with increased competition among the printer vendors, who are producing much higher quality photo printers, people will turn to other factors when evaluating a printer. It’s reasonable to expect that people would want to know how much ink they’ll go through over the course of time. But looking at the variables involved, it’s darn near impossible to tell someone what they’ll get.

What makes this more frustrating is that, even if you were able to develop a test that will reliably tell someone how many prints they can expect from a particular printer’s inkset, it doesn’t help you compare one printer against another. Each company uses different sized cartridges throughout their printer line, and none of those capacities match up with cartridges from their competitors. So, one can look at getting lots more prints out of an HP Photosmart Pro B8850 from a set of ink than they would from the R2880, but the latter printer’s inks are much smaller (and cheaper) than the HP device. This is one of the reasons we’ve been reporting cost per ml of ink in our reviews, but that doesn’t help people with ink yield.

What everyone wants is number that measures the cost per page, and that’s what we’ve thinking about. The question is how realistic it is to expect an answer that means something.

In looking at this issue specifically for the R2880 review, we wanted to be clear, and we wanted to be accurate. Our remark in the first look was counter to that reported by Vincent Oliver in his R2880 review on photo-i, and we didn’t want to add any more confusion to the market. (We think having many independent reviews of a product is a good thing.)

Our intent in this instance was to compare the efficiency of ink usage between the older R2400 and the newer R2880, which was complicated by the fact that the ink capacities and prices of the cartridges used in the two printers was different: 13ml for the R2400 (currently $14.24 on the Epson online store) and 11ml for the R2880 ($13.29 on Epson.com). There’s been a lot of invective online about the shrinking size of the R2880 (and R1900) ink carts, but Epson representatives have told us that they’re working on technology to make ink usage more efficient, so one would expect to see some improvement in this area.

The test

The question was how to come up with a test that made sense, and would be repeatable by others. We ultimately decided that printing with a freely available test chart was a good place to start. We would use one representative page, with a good balance of photos and test ramps, printed over and over until we ran out of ink. For our first tests, we chose an image (Profile Test Images) that Bill Atkinson has graciously made available “for free download to assist other artists.”

So we set up a test bed with the R2880 and the R2400, loaded them each with new, full ink cartridges, and resized the Profile Test Image file to 8" by 10" in output size at 240 dpi. We felt that this size was a decent compromise, and would also give people a better idea of how ink usage would scale as you increased page size. We printed each page from Photoshop CS3 in 10- or 20-copy batches until we ran out of our first ink cartridge. We then replaced that ink cartridge and printed until we had to replace the second cartridge, and stopped there.

During the process, we took screen captures (at a size of 200%) of the ink levels at 10-page increments. While we know that the capacity levels aren’t entirely accurate, it was a better option than ripping the cartridges apart and measuring the ink left in them. Using Adobe Illustrator, we built a grid that let us come up with an estimate of ink used in each cartridges. Again, we know that isn’t necessarily the perfect solution, it was better than no data at all, and added a crucial data point to the cost per page metric.

The findings

We found that, contrary to our initial results, the R2880 and R2400 ink yields were actually pretty similar, with a slight edge to the R2880. The R2400, with the bigger ink tanks, ran out of both yellow and light magenta inks at 90 pages exactly. The R2880 ran out of yellow ink after 74 pages (the second cartridge to go out was vivid light magenta, at 96 pages). Using Epson’s current prices for ink, the R2400 used approximately $70.72 worth of ink before we had to replace any cartridges, resulting in a cost per print (of ink only) of roughly 79 cents. The R2880 used approximately $54.40 of ink before requiring a swap, for a cost per print of 74 cents. At 90 pages, the R2880’s cost per print was approximately 75 cents; given the margin of error here, we’d consider it roughly the same.

Does it mean anything?

The point of this post is not that the ink yield of the R2880 is slightly better than the R2400 — that’s something we’ll cover in the final review. We are comfortable with this test for these two printers, and we think this is interesting data, but we’re on the fence about how useful it is. And we’re not even sure that it’s repeatable from vendor to vendor. We do think there is something there, and we’re hoping to get feedback from you regarding this test.

We know that there are plenty of holes in the way we tested this, and we’re thinking about how we would do this in a more varied environment. Ideally, when we tested, we would extract all of the ink from the existing ink cartridges and measure it carefully, but breaking into cartridges, dealing with sponges, ink sacs and other containers is problematic.

Using the ink gauges for the vendors doesn’t make much sense either; at the entry level especially, the gauges are off by a few percentage points, and they’re certainly not going to be consistent from vendor to vendor. On the other hand, it might be good enough as another, albeit inexact, data point that helps you determine whether a printer is the right one for you.

As we noted at the beginning of this lengthy post, ink cost is a hot topic, and we know that reliable, repeatable data that helps determine ink efficiency and yield would be a useful thing. But, as we deal with larger-capacity printers, the time and costs associated with testing every new printer at a deep level becomes prohibitive to us, and we’d like to hear from you on this. Weigh in on the comments below or drop us an email if you have thoughts on the matter.