Effective print output resolution

a question comes up once in a while about the real effective geometric resolution of print outputs, there is a simple answer - the printer manufacturers don't tell you but blind you with numbers like 4800x9600 or 5760 or something.

There is a simple test, with a line test pattern in RGB Bk by Roger Clark, a small file with lines 1 to 6 pixels wide wlhich you can print on your printer


you print this file with your printer, disable all image enhancing functions like interpolation, sharpening etc and need to find a way to print them with the genuine print resolution of your printer - with 300 or 600 dpi on Canon printers, or 360 or 720 dpi on Epson printers, a program like Qimage would tell you and let you print with the embedded resolution in the file, as stated in the file name with 300,360,600,720 dpi. You disable sharpening and interpolation in Qimage, or any other program you use for printing.

The resp. files will print these sizes
1,666 x1,00 inch/42,3 x 25,4 mm with 300 dpi
1,389 x 0,833 inch/35,3 x 21,2 mm with 360 dpi
0,833 x 0,5 inch/21,2 x 12,7 mm with 600 dpi
0,694 x 0,417 inch/17,6 x 10,6 mm with 720 dpi that's about the size of a small stamp, not more, and you
can check the size of your print output against these numbers

And when you print on some glossy paper you'll get prints like this, I took macro photos of the printouts


This is the printout of a L800 with the settings Ultra Glossy Photo in 720 dpi, you see that there is not
much detail in the bottom line with the 1px patterns, the limit is around the 2px patterns - about 360dpi effective. You can see stray droplets diffusing the line sharpness and ink bleeding that ink dots spread wider and closing the gaps between lines.

You can compare this output to an L300 4 color office type printer with a droplet size of 3.5pl vs. the 1.5 pl of the L800


It's not bad actually - there is no visible difference in output resolution between a L800 and L300.
The L300 has less nozzles and prints slower for color output, does not print borderless but overall performs quite well in this test.

Other resolution tests longer time ago with the Siemens star yielded about similar results that better inkjet prints do about 280 - 320 dpi effective resolution, that test calculated an average for the resolution in different line directions.

It might be interesting to see how other printers perform with this RogerClark test patterns, but I don't have access to any. I tested as well an Epson R265 which gives about the same results as the L800.
The pixel race on cameras goes on at this time into the 50 Mpix region, the printer output resolution has not changed since a long time, not at Epson or Canon , and the 300/600/360/720 dpi output is there since more than 10 years with the R265, IP5000 and others at that time. Would we need more - I don't really think so, but smartphone displays already outresolve the eye in its angular resolution - so what will happen with printers next ?

This Roger Clark test pic allows you to test the impact of various driver settings onto the resolution of the print output. Epson offers typically various quality settings but does not provide any detail of the differences.
Let me give you an example, the L800 with the Ultra Glossy paper selection offers the choice of 'Photo' or 'Photo RPM' with RPM = resolution performance management, but would you know what that means or does to your output - most likely not .

This is the actual L800 printout with the RPM setting which you can compare with the 'Photo' setting above.

Differences are barely visible in regards to resolution, the print output in RPM mode takes much longer, and you see a very slight improvement in the gamut, but not that much that the additional printing time justifies the use of this setting for me.

If you do the same with the L300, setting the print quality to extra high via the submenu 'more options' you get this print output

you see a slight improvement in the definition of vertical lines but the separation of horizontal lines gets weaker, so it is a mixed bag, and for me not a usable option in light of this irregular performance.

This just shows the performance of two printer models, every printer/driver behaves differently, and don't believe the manufacturers claims and flowery wordings of driver settings like 'best Photo' or something, test it and find out the most useful settings for yourself.

There is another setting with Epson printers/drivers - the 'high speed' switch in the advanced settings tab to turn on/off the bidirectional printing. All prints above are done with bidir. printing on, turning it off does not make a visible with these printer models, but I saw improvements with older printers which could
not be aligned well enough - so it all depends, and you only know if you go and test it.
I don't know whether the bidirectional printing can be turned off on Canon printers, that option is available as well in some Brother drivers, pretty well hidden.

Here is a sample of Pixels that I made up and printed out on my 9500..

There are lots of articles in the internet about print resolution in combination with sharpening, in Photoshop and with plug ins, and they all use different images with enough detail to show the intended effects. But only a few deal with the hardware performance of the printer only, the fixed resolution driven by the hardware properties of stepping motors, the coding on positioning bands etc. Printing with Qimage allows you to separate the effects and offers various algorithms for the interpolation between the image resolution and the active printer resolution as defined by the driver settings. Part of the pre-shapening is compensated by the pretty strong softening effect of the printer driver when resampling an image to the print resolution. Printer drivers need to be fast, not today but 15 to 20 years ago on processors with a much lower clock frequency than todays. The softening impact of a printer driver vs. the interpolation is directly visible in Qimage when interpolation is turned off in Qimage and is left to the driver. But in order to overcome any effect of resampling, either in the driver or in Qimage, it is necessary to directly feed the image data with the active print resolution - thus the 300-720 dpi settings of the Roger Clark test image, to print a 2 dot width line with two print pixels - with no resampling in between.

I won't spoil this thread with toner dust.....
http://www.printerknowledge.com/threads/samsung-clp-300.11000/

I printed the Roger Clark resolution pattern with an L800, with RPM setting and 720dpi as shown above. The posted images above are macro shots with an old 50mm lens in reverse position and about 75mm extension tubes on a Nikon D810 and some cropping to the size of the little printout- about 11x18 mm. I rescanned the same printout here,


scanned with an Epson Photo 3490 scanner with 2400dpi, the scan is not as sharp as the photo, you barely can see any stray droplets, everything looks smoother, that's about the practical limit for this scanner.
(Colors are off, I didn't do any white point adjustment/corrections etc)