Upsampling Photographs and Line Art
More and more, interesting
photos and other images are being included in books, advertising
materials, and every other form of printed product.
Look in any design magazine and you will see a plethora
of ads for digital photo libraries from which you can purchase
the rights to reproduce specialized photographic images
and line art. In addition to on-line photo services, you
can shoot images with digital cameras or even produce hard-copy
glossy prints from film shot with traditional cameras. Should
you need to prepare the image you have captured into a form
suitable for offset printing, scanning and/or some form
of image manipulation would be involved.
The general rule of thumb regarding
scanning any photograph is to scan at twice the
photo’s printed line screen. If your offset printer
will print halftones (b/w or color) at 133 lpi (the number
of lines of halftone dots per inch), you should scan your
photos at 266 dpi (dots per inch). Anything higher is a
waste of computer memory because it does not improve the
An important, related item
is to factor in the final size of the image you scan.
If your scanned image will be used at 100 percent of the
size of the original, the formula noted above is adequate.
However, if you scan the image and then reduce it, you can
get away with less than 266 dpi. An example would be to
scan the original at 133 dpi and then make your photo 50
Although scanning at the final
size is preferable, you can reduce or enlarge an image slightly
(maybe 105 –110 percent of the original size,
or 5-10 percent enlargement) without compromising the photo’s
quality. Where people make the most drastic mistakes (which
really compromise the final image) is by scanning a photo
and then enlarging it more than a few percentage points.
When you ask Photoshop (or
any other image manipulation software) to increase the size
of the image once it has been scanned, and then
increase the resolution as well, this is called upsampling.
(An example of this is to take an Internet image saved at
72 dpi and then enlarge it and upsample it to 300 dpi).
As Photoshop does this, it essentially makes up picture
information that doesn’t exist. As the picture gets
larger, Photoshop adds pixels (picture elements, essentially
dots that are averages of the existing pixels) between those
already there. Past a certain point of over-enlargement
(beyond 105-110 percent of the original size), what you
get is a blurred image and/or visible pixels.
The same doesn’t exactly
apply for line art (drawings with only black lines and no
shades of gray).
A scan of a line drawing (with no shading)
should be made at 1200 dpi at the final size (100 percent
size) to not show any visible pixels. However, upsampling
(which is also called interpolating) is often more forgiving
with line art than with halftones. Although I have never
read this anywhere, I have found this to be true in my own
experience, and this is my theory as to why.
When you enlarge a color or
b/w image and use Photoshop to upsample, you are
averaging between pixels of process color, PMS color, or
various shades of gray. When you upsample a line art image,
however, you are not averaging between shades of gray or
between colors, since all pixels are either black or white.
If you were to enlarge the image without increasing the
resolution, your pixels would become visible dots or squares
(this would cause undesirable patterns). But by enlarging
the line art image in Photoshop and at the same time upsampling
(or increasing the resolution, let’s say from 600
dpi to 1200 dpi), you make the pixels smaller (and therefore
less visible, making the image look like a continuous line
drawing and not a pattern of dots). In this case, unlike
enlarging halftones, you are just adding or removing smaller
and smaller pixels that are either black or white. If the
resolution is sufficiently high, the eye will not distinguish
between the separate pixels.
This has worked for me, but you should
always look closely at your laser output before committing
to film or plate at the offset printer.
Another thing to remember
when enlarging or reducing images is that the above only
pertains to bitmapped, or rasterized (dot pattern), art.
Vector art, on the other hand, is created in Illustrator
or Freehand and is made up of PostScript lines, curves,
and fills (essentially mathematical equations) rather than
a grid of dots. “Vector art” refers to drawings,
not photographs. It is superior in its own right since it
never has jagged edges from visible pixels, as does rasterized
art. This is true whether or not it is reduced or enlarged.
Vector art can therefore be reduced or enlarged to any size
with no degradation in quality.
[Steven Waxman is a printing consultant. He teaches corporations how to save money buying printing, brokers printing services, and teaches prepress techniques. Steven has been in the printing industry for thirty-three years working as a writer, editor, print buyer, photographer, graphic designer, art director, and production manager.]