printer test grid

[cthornhill]

All this discussion of resolution has been pretty interesting to me. This AM I broke down and created a couple test targets for myself to run some quick test on my Epson 2200. What I did (it only takes a few minutes in Photoshop if you want to do it) is to make a set of black and white line pairs - strait lines, ever other one black at 360ppi and at 200ppi. Across the top I filled in the pixel below the top one on the white lines so I had a one pixel stairstep effect - so I can see one pixel detail.

I then sent the targets to the printer at native PPI and then tried interpolation just for fun.

Your results on your paper stocks and with your settings may differ, so DON'T TAKE THIS AS ANYTHING BUT ONE GUY'S TEST!

Please, I am not doing anything but reporting my results - you would have to run your own test and interpret your own results!

What I found was that the 360 ppi target with 180 line pairs per inch is a little above what the 2200 can put on paper. I used a semi-gloss stock and the Epson ICC profile in Photoshop 7 and found that with that driver t 1440 high speed and microweave on, I could not clearly print 360ppi worth of B&W fine detail. The printer got lots better when I used 2880 and turned off high speed and microweave, but it still did not render the white lines cleanly on that stock. I could see them as grey areas under magnification, and the stairstep was slightly present, but degraded.

The 2200 was able to run the 200ppi test target with 100 line pairs per inch at both 200ppi and interpolated to 360ppi. The results were (as expected) clean reproductions at most settings on that stock.

This meshes pretty well with the experamental data Norman Koren got for Epson printers, and with my experiance of the device. It may or may not match your experience.

I will try other RIPs and setting when I get a chance, as well as some other paper stock. I should also build a target at 150 line pairs per inch and at 120 line pairs per inch to test with.

My point in posting was just to point out that it is pretty easy to build a quick test target for a device and then you can see how many line pairs per inch you are going to actually be able to get back. Of course most people don't take pictures of fine black and white lines, and there is lots of other testing you could do, but it gives you a handy way to estimate the minimum details you can resolve on a print and is a fairly objective test target (you still have to apply judgement to the viewing so nothing is totally objective about this method).

I plan to use my targets to test paper and settings combinations, and new RIP's.

[cthornhill]

Later in the morning I re-ran my 360ppi test making sure that high speed was off at 2880 sending the 360ppi data at that resolution. I got a bit better result, so I think my first pass was not perfect. The results at 100 percent black and white were still marginal - I would not pass the output as a good reproduction of the test pattern, but they were showing a little white space in the white bars and the stairstep was visable if degraded.

I did the same test with the 360ppi target set to 50 percent intensity so that the maximum black was at about RGB 127 (close enough for me). The result was a much better reproduction of the pattern.

Again, this tends to match what Norman Koren has reported for Epson printers and the Photoshop print function - lighter tones are better at reproducing fine patters. At least that is what I read in Norman's data - you can read it for yourselves and decide.

For me the implications are that there is indeed a quality improvement in using 2880 as a print mode with high speed turned off and no microweave. The difference is slight but visable in very fine detail. I find that the printer does indeed handle fine detail in lighter tones better than darker tones, and that is to be expected since the ink and paper issues contribute to this as well.

I will be interested to see what I get from another RIP...

[cthornhill]

I did two more test inspired by people here and by some data on the Qimage site (I have used Qimage for a while now and like the product - IMO good value for money and nice people).

I created a 300ppi test target with 150 line pairs per inch. I ran this in Photoshop as is (sent the print driver 300 ppi) with a printer set to Semi-gloss (ICC), 2880, no microweave, and high speed OFF. The results were not a lot better than the 360ppi test grid. A little better, but not much.

Then I tried upscaling the 360ppi grid to 720ppi in Photoshop (bicubic interpolation), and printing that at the same settings. This is a type of test suggested on the Quimage site. The results were much better. Clearly the Photoshop driver is not optimizing at some point.

I then sent the same file (original 360ppi file - not upsized) out to the printer with the same settings (2880 no microweave, no high speed) and the results were indeed much better. I agree with the Qimage people that I see a little better detail than I got upsizing in Photoshop. You would have to judge for yourself.

As far as I can see, there are some limits to the quality of the results given by the Photoshop 7 driver with 360ppi data. Using only the Photshop driver I feel I do not get as much fine detail as exists in a 360ppi data set. I think the output from the qimage print RIP looks a little more detailed.

Now for me, this is interesting, but only part of the picture. I do not feel (and this is just for me!) that the output of the Epson 2200 is fast enough to be viable for business use at 2880. I do find most prints at 1440 to be of good quality, and the speed is not good, but I can live with it for small runs of prints.

For me (and again this is just me!) I don't have resolution issues on most prints that would force me to worry about the difference. On some jobs I might need the extra resolution offered by the other RIP or higer end slower settings, and I will use it as needed.

I knew from my experiance with ImagePrint on larger Epson systems I use that the Photoshop RIP is not the best, but now I feel I have a better handle on what I might be loosing and where I can go to adjust it.

Again, these are just my results, everyone would need to make their own decisions based on their own tests!

Cecil

[crabby]

Thanks for your research. This is very useful information for me. I'm trying to understand how you created your test patterns. For instance, how do you print a white line on a white sheet of paper?
Frankly I'm tired of running tests on my printer. I think I have spent as much money on printer supplies running tests as I have for actual prints.;) So I look at other peoples test results and opinions with great intrest.

[cthornhill]

All I did was to supply the black lines...The white background takes care of itself. I just created files that were one inch square in Photoshop 7.01 at the required DPI, then used a 000 (100%) black one pixel pencil tip at 300% magnification to produce the lines top to bottom. Once I had enough lines I copied them to fill the square. I use the shift key to keep the lines strait while drawing them. There are other ways, but that was easy for me.

I am also pretty tired of test patterns, but I did not know any other good way to see what I would get in the way of DPI reproduction. I just resist using photos as a first test since there is so much room for interpretation (heck there is plenty of room for argument with a test patern). In the end though you have to judge photos not test paterns. I just wanted to see what the upper limit of small dots would be for me...And to try the suggestion at the QIMAGE site.

Interesting stuff. I think there is some room for improvment in the default Epson driver with Photo Black. I suspect it is tuned a little dark to make sure the blacks look good, but that is just a guess. In any event, it sure does reproduce finer detail when the black is not totaly saturated (100%).

I still get great response to images printed out of Photoshop 7 with the Epson ICC profiles and the default driver, so I am not having problems. For me speed of printing and ink use are also important issues as is fine detail, so my workflow choices are a balance between all these factors.

If I just had unlimited money I feel I would use ImagePrint, but I don't think I am shortchanging my customers when I use the Epson print driver. I do feel I understand what is going on a bit better for the testing, so at least that was useful.

Hope it was of some use to others.

Cecil

PS - I really should make a printer tourture test with diagonals, and color patters, but it is just not at the top of my list right now. If I do I will post the results.

PPS - Crabby, what happend was that the 360 ppi pattern printed at 1440 with high speed on blocked up - the print did not show a black line and a white line, just a messy mostly black square. Too much ink came out and the space between the black lines (the so called white line) was lost. When I re-sized the image to 720ppi in photoshop I allowed photoshop to use bi-cubic interpolation so the pattern has shade of grey, not just sharp black and white lines. This is partly why (at least I think so) the print looks better - I know from tests that the 2200 will print finer detail without blocking up if the subject is not 100% black (that is if it is shdes of grey). Tone is affecting detail reproduction since the system as a whole (who knows what part exactly just now) is putting more black ink on the paper than it should with a fine pattern that is totaly black and white. Perhaps another ICC profile would adjust this, or maybe it is in the driver...Right now I can't say. I do see that other rips (qimage) on this printer can do a little better and they claim that upscalling is one thing they do to help...ImagePrint is such a totally different screening algorythm that I know it will work differently - I just haven't tested it on the 2200 yet. I do like it on the larger Epsons (24 inch or 44 inch).

Cecil