Digital Prints

Testing Scanner Resolutions for Print Quality
Reading words and theories about scanning resolution guidelines are one thing, but seeing
and evaluating printed test results in terms of detail rendering and overall image quality are
even more valuable. I asked advanced-amateur photographer Mark Segal in Toronto to test
two different scenarios with a film scanner: (A) starting at optical maximum resolution, at
what point does the resulting printed image start to deteriorate? and (B) can you really see
differences between odd and even integer divisor resolutions on prints?
To do the tests, Segal took a 35mm color negative he shot of the Cairo Tower (Egypt) a
few years ago and scanned it on his Konica Minolta DiMAGE Scan Elite 5400 film scanner
scaled to print 6.7 by 10 inches on his EPSON Stylus Pro 4000 desktop inkjet printer.
This is one of his normal workflows. This image (see Figure 3.5) was selected because its
details, diagonals, highlights, and shadows lend themselves well to observing the consequences
of different resolution settings.
The scanned images were produced as “scanner raw”
files with no sharpening or repairs and no color correction,
and they were output in 8-bit mode and printed
at 1440 dpi. What you see shown are scans of the actual
prints made on the Epson 4000.
(Note: Konica Minolta has a slightly different way of talking
about scan resolution terminology and of setting up
their scanner software. There are separate Input and
Output resolution list boxes (see Figure 3.6). The Input
resolution shows how many pixels per inch are scanned.
The Output resolution shows what the resolution of the
printed image would be for the print size and magnification
factor indicated. What ultimately matters with resolution
is this Output resolution or how many pixels
there are spread across how many inches.)
In Test A, Segal scanned his negative at several Input resolutions starting at the maximum
(5400 dpi) and decreasing them with each scan, maintaining his desired output sizing
(6.7 × 10 inches). As you can see in the cropped and enlarged portions of the resulting
prints, the images only start to degrade at his 1704/240 dpi test with the biggest deterioration
at 860/120 (Figure 3.7).
For Test B, Segal tested an EVEN integer divisor of 5400 dpi (2700/383 dpi), an ODD
integer divisor (1800/255 dpi), and a non-integer resolution of 2800/395, again with all
scaled to print at the same desired output size (see Figure 3.8). Studied at a normal viewing
distance, the ODD 1800/255 is slightly inferior to the other two, although it’s unclear
if that’s a function of its odd-integerness or of its lower resolution. The other two look identical.
However, looking through a 6x magnifier, there is an almost imperceptible improvement
with the 2700/383 over the 2800/395, which indicates that there actually is a
difference when scanning with integer divisors, although the practical benefit is debatable.
So what’s the conclusion of these tests? Says Segal in his analysis, “If you don’t look at
prints with a powerful magnifier, there is a wide latitude of scanner settings to produce
similar quality results.”
Chapter 3 ■ Creating and Processing the Image 83
Figure 3.5 The test image of Cairo
Tower in the center of Cairo.
Courtesy of Mark Segal