Digital Camera Home >
The Imaging Resource Tip for Makeing a Sharper Image
A Sharper Image |
In many of the articles on this site, we covered a number of
ways to improve your photos digitally, including adjusting tonal
balance, and "cloning out" distracting image elements.
This time, we're going to drop down to a less obvious level,
and deal with the much-maligned and misunderstood technique of
image sharpening.
I say "maligned and misunderstood" because many people
only use the "sharpening" tools in image editing programs
to try to correct for blurry or mis-focused original images.
Worse, some manufacturers of low-end editing programs (e.g.,
Microsoft, with PictureIt) felt that "sharpening" was
too technical a term, and so changed it to the more (mis)understandable
term "focus." Aarrgghh! (Sorry, I had to get that out
of my system.)
Image sharpening can do a lot of things for your pictures, but
it can't fix an out-of-focus original! IT.IS.NOT.FOCUS
! We'll look a bit deeper at what image sharpening can
do, and how it works, so you'll be able to apply it properly.
The results can be impressive if you have a good sharpening tool,
good images, and know what you're doing!
A bit of background: What is "Sharpness?"
To understand what "sharpening" routines in image editing
packages do, it's helpful to understand what we mean by "sharpness"
in the first place. Keep in mind though, that we need to explain
the concept in terms a mindless image-editing program can understand.
With only a little thought, it's easy to see that image sharpness
has a lot to do with the edges
of things; those places in a picture where the image content
is (or should be) changing rapidly from one tone or color to
another. In a "sharp" image, these transitions happen
quickly, over a small area. In a "soft" image, transitions
between objects are blurred across a larger area.
From this definition, it's easy to see that what we need to do
to increase the perception of sharpness is to make transitions
between one color or tone and another in the image happen more
rapidly. Of course, we only want to do this where there's a fairly
abrupt transition or edge already present. In computer-ese, we
want the program to find places where it looks like the color
or tone is changing rapidly, and make that change more rapid.
In fact, this is exactly what sharpening routines in image-editing
programs do.
Why stop there? Fooling Mother Nature.
In actuality, most image-sharpening techniques go a bit further
than just making tonal changes more abrupt: They deliberately
over-compensate a bit, and push tone and color on each side of
the edge a bit beyond where they're really supposed to go. Then,
a little further away from the edge itself, they let the values
come back to what they should be.
It turns out that this over-compensation is just the ticket for
fooling our eyes into thinking that there's more detail there
than there actually is. (More than meets the eye?) Actually,
it can't insert detail that doesn't exist, but it can make us
more aware of what's there. The figure titled "Sharp up
Close" shows an extreme magnification of an edge "sharpened"
in Corel PhotoPaint, using the "unsharp masking" filter.
The top half of the figure shows the original tint blocks, while
the bottom half shows the impact of unsharp masking. Notice how
the (grossly exaggerated) unsharp masking emphasizes the edge.
Unsharp Masking: A flash from the past.
Actually, for all our digital sophistication, all we're doing
is mimicking a darkroom process that's been around for the last
50 years or so, and that's still actively practiced today. (Albeit
by darkroom perfectionists with way too much time on their hands!)
The technique of "unsharp masking" in conventional
photo processing involves making an underexposed, soft ("unsharp")
contact-print from a negative, sandwiching this "unsharp
mask" in precise registration with the original negative,
and then making a print with the combination. The result is an
artificial enhancement of fine detail in the print that can produce
exquisite results in the hands of a skilled practitioner.
A quick run through the controls
While many of the low-end image editors include sharpening controls
of one sort or another, we're going to focus on the "real
thing" here; the unsharp masking operator, as found on professional
packages like Adobe Photoshop and Corel PhotoPaint. If you don't
have an image-editor of this level, don't despair: Most sharpening
software implements the same basic techniques. By understanding
how unsharp masking works, you'll be better able to recognize
what the lower-end sharpening tools are doing, and use them appropriately.
The figure labeled "Unsharp Central"
shows the unsharp masking control panel from Photoshop. (Corel
PhotoPaint's is quite similar, differing only in that Corel doesn't
permit decimal values for the radius.) At top left is a small
preview window showing the effect your settings will have on
the image. You can scroll this around the picture simply by clicking
anywhere on it and dragging in the direction you want to go.
You can also zoom in or out by clicking the + or - button below
the window, although I find it most useful to work at 100%. The
"preview" check-box lets you view the effect of your
settings on the entire image. (This can take a while if your
image is large.)
The controls themselves appear at the bottom of the window: The
Amount slider dictates how strongly the effect should be applied,
while the Radius control tells the program how wide an area (in
pixels) to scan over when looking for an edge. The area specified
by the Radius control also governs how large a region will be
affected on each side of edges that are found. The Threshold
slider is more subtle: It sets a minimum "edge contrast,"
below which the unsharp masking operator won't have any effect.
In other words, minor variations in contrast will be ignored
if below this value. (The Threshold control can be very useful
when sharpening portraits: You want the hair and clothes to snap,
but don't want to emphasize every tiny blemish on the subject's
skin. By playing with the Threshold setting, you can often keep
the sharpening from affecting the subject's face, yet still work
where you really need it.)
Proper Use
I said at the outset that sharpening can't do much for an out-of-focus
original. This is because the blurring in poorly focused photos
spreads the picture information out across too wide an area to
have any hope of reconstructing it, unless you happen to work
for NASA. On the other hand, scanners and particularly printers
tend to blur images only slightly during the reproduction process,
and in a very controlled fashion. This sort of blurring is relatively
easy to compensate for with sharpening, and the results can be
impressive, indeed. The ideal amount of sharpening to use is
that which just compensates for the effects of the scanning/printing
process: Any more will produce artificial, coarse-looking images.
|