Hi folks,
Due to the recurring questions about print resolution (limited by printer and or medium), I sat down and created a super (too ?) critical resolution test target which should reveal any shortcomings in the output process. It's merciless, and currently in the testing phase of design, but I wanted to share it already, in case someone can use it in it's current stage for optimizing their output workflow.
There are two versions available for download and personal use, mainly intended for testing inkjet printers:
Test target for 600 PPI printers (e.g. Canon / HP)
Test target for 720 PPI printers (e.g. Epson)
If printed at the indicated PPI resolution, the final target output should have a square size of 130 millimetres (5.11811... inches).
There is a number of patterns on the target that will allow to quite accurately measure the actual resolution that the printer/ink/media combination can produce. That's also useful if one is searching for a medium that has relatively low ink diffusion characteristics. Some media are sharper than others, on the same printer.
There is a central 'star' shaped pattern that should be resolved all the way down to the green concentric circle near its center. There is also a red circle at half the maximum resolution, which should help to estimate intermediate resolution in any orientation (not only horizontal or vertical). Inside the green circle is a pattern of sinusoidal 'rings' which also get higher in resolution as they approach the green circle, and they are very sensitive to ink diffusion which tends to make the rings seemingly darker as the detail gets smaller. Differences in print-head alignment and paper feed accuracy should show by the appearance of aliasing patterns and differing resolution between horizontal and vertical resolution limits.
At the bottom edge of the target is a greyscale. Actually it's a combination of 5 greyscales that each use more steps to produce the full gradient. The top one is 32 steps, followed by 64 steps, 128 steps, 256 steps, and finally virtually continuous tone (16-bit/channel precision). There is no dithering in the gradients, so they should show discontinuities and non-gray profiling issues quite well.
On the left and top edge there are blocks of uniform resolution, each with 4 patches with horizontal, vertical and diagonal sinusoidal patterns of the indicated resolution, marked in cycles/mm (assumes correct print resolution, 600 PPI or 720 PPI, was used). In principle, these should print at equal (medium-grey) density regardless of the pattern orientation. But differences in print-head alignment and paper feed accuracy may cause density differences and/or aliasing patterns, especially on the higher resolution patches. Non-linearities in the match between output profile and media can also cause what appears to be aliasing patterns, but they are actually something else.
The top left-hand side patches are at the printer's maximum resolution, 1 pixel wide lines. That level of resolution is much higher than normally required for good print resolution. Good resolution is achieved at 5 cycles/mm, and excellent print resolution is achieved at 8 cycles/mm, when viewed at reading distance in good (not too dim, not too bright) light levels. The 'excess' resolution can be put to good use when using high quality output sharpening, and for subjects with very fine lines or sharp edges and high contrast.
At the right-hand edge of the target there is continuous sweep of diagonal sinusoidal patterns that should ideally look like a smooth increase of resolution without aliasing patterns or disruptions. They are also resolution marked in cycles/mm, and the range is chosen to be able and encode the sinusoids relatively accurately, and encompass all practically relevant levels of output resolution.
There are also two smaller pattern blocks of 1, 2, and 3 line wide gradients that allow to estimate if and how the highest resolutions will start to lose the ability to mix intermediate tones. They attempt to show the trade-off between resolution and color mixing of intermediate tones by the printer driver's dithering algorithms.
Finally there is a small patch of high and very low contrast lines, 1, 2 , and 3 pixels wide, at the top left corner of the grey square of the star background.
Make sure to first do a head alignment on the actual paper (thickness) that the target will be printed on. It will save you from disappointing results, which may still occur due to the overly critical nature of the target. The patterns on the target itself are virtually artifact free, so any irregularities that show up, must be caused by the output workflow (e.g. profiling), or printer driver settings or mechanical printer tolerances. The target itself is tagged as having an Adobe RGB colorspace, but one can always assign another colorspace to see if the profile conversions to the output medium have any adverse effects on resolution.
That's about it for the moment.
Cheers,
Bart
