Printers
Epson 11880 vs. Canon ipf6100 - which is faster?
Tuesday, February 12, 2008
I’ve been using both of these printers for several months now. Canon was kind enough to send me an evaluation unit when the ipf6100 was released in early September of 07. I ordered an Epson 11880 the day they were announced, and it arrived at the beginning of November. Currently I have the Canon at my office and the Epson is in my workroom at home. The printer I use is based on which location I’m working at, and I must be honest, I do it without much regard to output quality, since I feel they are both outstanding with great gamuts, rich saturated colors, smooth subtle transitions, and great detail. Both are much better than my previous printers, an Epson 4800 and 9800.
On several occasions, I have been asked which one is faster. In this article, Canon claims openly that the ipf9100 is substantially faster than the 11880, and while I don’t have that particular printer, the only difference between it and my ipf6100 is paper width ... the printers share the same heads, the same head transport design and speed, as well as quality settings and options. My assumption is the 9100 will see some increase in speed as you move to wider material, much like the Epson, since the printer can spend more time laying down ink with each pass and less time changing directions. But to me the Epson seems faster than the Canon when printing on 16” and 24” rolls, which is what I use the most, and in no way does the Epson seem 60% to 70% slower, as claimed in the article.
It’s pretty easy to find speed ratings for the 11880 ... Epson openly publishes times to print various prints at various settings and compares that with other printers such as the 9880. It is a little more difficult to find such information on the Canon. It is also pretty easy to understand what overall dot matrix and density the Epson is using with each quality setting, something which has been a challenge to track down on the Canon. In fact, after running my initial series of tests, I found the Canon has an additional level of quality available with certain paper types, of which their “special” paper types are not able to produce. Initially it appeared at the maximum quality setting of each printer the Epson was slower, but testing this additional quality level on the Canon dramatically reduces it’s throughput, and it is indeed the slower of the two printers (at the maximum quality setting anyway).
I believe it is somewhat significant to note that among other technological advantages the Epson may have over the Canon, the simple fact is it can lay down more ink dots per square inch. The maximum resolution of the Epson is 2880x1440 dpi, whereas the Canon achieves a density of 2400x1200 dpi at it’s maximum setting. If you calculate that out, the Epson is printing 4,147,200 dots per square inch of media versus the Canon at 2,880,000 dots per square inch of media ... that’s about 44% more. So logically at the highest quality settings, the Epson perhaps should be slower, after all it’s printing a far number of individual dots on the paper. It also can produce drops as small as 3.5 picoliters, where the Canon’s smallest is 4 picoliters (wow ... both are pretty tiny... a picoliter is 1 trillionth of a liter !). According to Epson, the 11880 head can produce ink drops that are more consistent in size and shape than any other printer, and can place them more accurately than any other printer. Add that all up, and you end up with a printer that should produce a higher quality image when using very high quality paper surfaces.
My methodology, as usual, isn’t very scientific, but to me is logical and pretty simple. To test the speed of each machine, I created a file by cropping Bill Atkinson’s Lab Test Page to a 24x15” print ... 2.5 square feet. I also used this same page and created a 60x15” version, duplicating the image so the entire page contained image data, to see what differences this made on the 11880 in throughput. This test page was 6.25 square feet. I opened the top of the Canon and used a key to by-pass the cover interlock switch so I could observe the head, and timed each test based on when the head began moving to lay down it’s first pass of ink until the head stopped moving. I made sure my Mac had no other tasks running, and watched carefully that the printer didn’t stutter or hesitate while printing the test. On the Canon I ran the tests with both the OS X driver, as well as the Canon 16-bit plugin. Since the network sometimes has problem handling the data on either printer, causing it to pause, I used the USB connection for the tests, which eliminated this problem.
After spending a few days calculating output speeds, I suddenly found myself in a quandary ... this isn’t an apples to apples situation. Both printers are capable of a multitude of speeds based on the desired level of quality of the output, and both trade “quality” for speed as you go to lower settings. Since there is no correlation in quality between the output settings of the two printers, to evaluate which one is “faster” requires a somewhat subjective opinion on which quality setting of each machine provides similar results.
So is the Canon faster? Is it substantially faster as claimed in their “competitive brief”? Here are the results of this test showing the square feet per hour each printer achieves at the various quality settings.
Epson 11880 2880 dpi uni-D, Microweave on, 12.33 sq. ft/hour
Epson 11880 2880 dpi bi-D, Microweave on, 21.42 sq. ft/hour
(increases to 25.5 sq. ft/hr on 60" wide paper)
Epson 11880 1440dpi (uni-D), Microweave on, 23.43 sq. ft/hour
Epson 11880 720dpi (uni-D), Microweave on, 31.47 sq. ft/hour
Epson 11880 1440 dpi bi-D, Microweave on, 33.08 sq. ft/hour
Epson 11880 1440 dpi bi-D, Microweave off, 40.9 sq. ft/hour
Epson 11880 720dpi (bi-D), Microweave on, 43.9 sq. ft/hour
Epson 11880 720dpi (bi-D), Microweave off, 50.56 sq. ft/hour
*Epson 11880 360dpi (bi-D), Microweave on, 116 sq. ft/hour
(all tests are Finest Detail off. This setting is not designed for photographic images and is not recommended for large files)
Canon ipf6100, 32 pass, 600dpi Highest (uni-D), Precision on, 8.55 sq.ft/hour
Canon ipf6100, 16 pass, 600dpi Highest (uni-D), Precision off, 17.14 sq. ft/hour
Canon ipf6100, 32 pass, 600dpi Highest (biD), Precision on, 17.45 sq.ft hour
Canon ipf6100, 12 pass, 600dpi High (uni-D), 22.22 sq.ft/hour
Canon ipf6100, 16 pass, 600dpi Highest (biD), 34.75 sq/ft hour
Canon ipf6100, 12 pass, 600dpi High (biD), 44.11 sq/ft hour
Canon ipf6100, 8 pass, 600dpi Standard (biD), 66.6 sq ft/hour
*Canon ipf6100, 300dpi Standard (biD), 133.3 sq ft/hour
(Precision is only available at the Highest setting on certain paper types, unless noted precision was off)
*Please note that on both printers the fastest setting is listed as a reference, is a setting I would never use, and is not even available on any high quality paper types.
As you can see, each printer's speed is affected dramatically by the quality choices selected. At the maximum quality setting available to each printer, the Epson is faster (12.33 vs 8.55 sq. ft/hr.) As you move to lower quality settings, both printers trade quality for speed. But here's where it gets complicated ... which settings on which printer are "equivalent" quality? The quote from the article specifically states SuperPhoto mode, and 24 sq. ft/hour as being 60% to 70% slower than the Canon iPF9100. In SuperPhoto mode with 24" paper my tests shows throughput of 21.4 sq. feet per hour. If I extrapolate out the speed being referenced by this article, it sounds like they are comparing the Epson to the speed of the Canon at its 600dpi Standard mode.
Let's be honest ... if I want to print things faster, I can also do so on the Epson ... I can print things much faster. But after printing extensively on both printers, one thing I'm quite confident of ... the Epson at 2880dpi SuperPhoto mode is substantially higher in quality than the Canon at 600dpi standard mode. Why they think it is acceptable to compare the speeds at these two quality settings is beyond me ... not appropriate by any stretch.
I decided to make some attempt at establishing which settings might be comparable between the two printers. This is a pretty difficult task, and I'm sure many "experts" will challenge my methodology, but I'm not an expert, so I devised a somewhat simplistic approach .. print a bunch of prints and compare them visually. Sure, I could scan them and look at dot patterns and density, but to be honest, when you do that you are dismissing the very property that makes the printing process work ... the way our eyes blend the dots into colors and details.
One other thing makes this task very challenging ... both printers produce outstanding quality, and it is difficult to find anything in any of them to say one is better. This even applies to prints from the same printer ... prints at various quality settings on each printer yields remarkably identical output, and what differences can be found are extremely minor.
But I at least needed to try, or the speed tests have little meaning. As I said, my methodology was pretty simple. I resized Bill's Lab Test Page.tif WITHOUT interpolation to a 20" tall print. This yielded an image file that was not native to either printer. I printed the print at most of the quality settings available. All prints were printed on a brand neutral paper, Kodak Professional Glossy. Both printers were profiled using Bill Atkinson's 5202 target, read with an EyeOne i0 table, and generated with ProfileMaker 5.08. The Canon profile is a new one, and does use the correct "color match" option. Finally, I decided to use the Relative Colorimetric Intent with Black Point compensation on to avoid variations that may be due to the way the Perceptual Intent was created in the profile. The profiles were made on the Epson at 2880 dpi, on the Canon at 600dpi Highest with precision off. (16 pass). I will admit for "perfect" results I would need to profile both printers at each setting, but I don't have that kind of time, and to be honest, most of the differences are not seen in saturation and gradations, but more in visible dithering in some regions ... and I believe the dithering would remain unchanged despite the profiles. Additionally I had each printer perform an auto head alignment before printing the test prints.
Once the prints were made, I asked a colleague of mine who happens to be a great photographer but also very technically savvy to help me examine the prints. We looked at several "sets" of prints, but were unaware when looking at the prints which printer and settings was used. The first set was to take all of the prints from each printer individually and order them by quality. I knew this would be challenging, and indeed there are no visible differences in almost the entire print, but we discovered a couple of areas in each print where dithering could be seen to change visibly (this is without the aid of a loupe, only reading glasses in my case). When we were finished it was apparent that both printers increase in quality directly corresponds to it's decrease in speed. While each step the difference is quite miniscule, there were differences.
We then compared the top 2 prints from each printer. All four of these prints are remarkable quality, but the Epson at 2880dpi uni-directional print had the least amount of visible dithering in the skin tones in the 4 portraits, with the Canon 600dpi Highest precision on uni-directional print next, followed by the Epson bi-directional and the Canon bi-directional. Again, we placed these prints in this order without knowing which print was printed on which printer. We felt the Epson prints were very slightly sharper in a couple of very select regions, such as the printing in the face of the clock.
We then compared the Epson midrange quality against the Canon, and here we both order the 1440dpi prints from the Epson very slightly ahead of the Canon prints at Highest or high. Another set of prints included the 720dpi uni-directional print against the Canon Highest and Epson 1440 bi-directional, and we were surprised the 720dpi uni-directional print seemed the best.
Finally we compared the Epson 720dpi with microweave on against the Canon at Highest and High. In the first case, they are very close. The Epson still seemed to have an edge on sharpness, but the word miniscule may be overstating the difference. However, the Canon seemed to have a slight edge in the visible dithering in the skin tones. We ordered the prints Canon Highest, Epson 720dpi, and Canon High, feeling that the dithering in the Epson and Canon High were nearly identical, but the Epson had that miniscule edge in sharpness.
The Canon at Standard had increased dithering, less sharpness, and a light loss of saturation. It didn't seem to measure up to any of the other prints in the test. I will admit this saturation loss could be from the profile. The Epson profile worked great at every quality setting, (which surprised me a little). This perhaps didn't work as well for the Canon, and I plan on testing it. So while the profile admittedly may resolve the saturation issue, the profile would be much less likely to help with sharpness, and I can’t imagine it would help with dithering.
So there you have it. I believe Canon's article I linked earlier is misleading, and even somewhat irresponsible. When I try to "match" quality levels, the Epson is a faster printer. At the top end this is very apparent, especially if you want to use uni-directional mode where the Epson has a distinct advantage, because it returns the head at about 3x it's printing speed, whereas the Canon head moves at a constant speed whether it is printing or not. The speed differences were less so as you move to a notch below maximum quality. Only if you accept the Canon's lowest high quality setting as good enough will it win over the Epson in speed, but 66 sq. ft/hour vs 50. sq. ft/hr isn't any 60% to 70% faster, and is only obtained by again slightly decreasing quality ... in fact this is a setting which the Epson has no equivalent. Personally I never use that setting, and never will.
In conclusion, let me state one last time that the differences are extremely small. Even comparing the Epson at it's highest quality settings available at 2880 dpi against it's own much faster settings available at 720 dpi yields differences so small that most comparing them side by side may not even see them, or may dismiss them as insignificant. My goal wasn't really to prove the 11880 was faster, but more in the response to the ridiculous claim quoted from the article. However, at the ranges I tend to use in both machines on my various choices of paper, my personal impressions stated earlier tend to be true, the Epson is faster.
This was the first image I tried at 40x60 on my new Epson 11880 printer. I was really pleased with the final print.
“Rainbow Eucalyptus
of Maui”
Hassellblad H1 with Phase One p45+ back
HC 300mm f/4.5 lens with HC 1.7x TC
1/6 secs at f/6.3, ISO 100