Q: I'm looking to buy a professional scanner to scan my own
artwork. What do you recommend?
When you say "professional scanner," what do you mean by "professional"? Real professionals-- high-end service bureaus-- do not use scanners you will find in a computer store. They use drum scanners, which offer vastly superior resolution and dynamic range when compared with flatbed scanners. These scanners start at about $40,000 US and go up to about $320,000 US. They offer resolutions starting at 8,000 pixels per inch and going up to about 14,000 pixels per inch.
However, resolution IS NOT a good indication of scanner quality. The best indication of scanner quality is "dynamic range"--the range of tones, from light to dark, the scanner can capture. Color transparencies have a dynamic range of about 4.0. Dynamic range is measured on a logarithmic scale. A dynamic range of 4.0 is much, much higher than 3.7, which is itself much, much higher than 3.2.
Typical flatbed scanners have a dynamic range of 2.4. High-end consumer flatbed scanners--what you are probably thinking of when you say "professional" scanners-- typically offer a dynamic range of 3.2 to 3.6.
Printing professionals consider a dynamic range of 3.7 to be the minimum for low-end to midrange print reproduction, and 4.0 to be the minimum acceptable for high-end print reproduction.
is less important, because an image rarely needs to be over 400 pixels
per inch at the reproduction size for offset printing. What
does it mean if the dynamic range is low? It
means the scanner cannot pick up detail in the shadows.Anything darker
than a certain point the scanner "sees" as
black, with no detail whatsoever.
Scans made on a scanner with a low dynamic range are flat, muddy, and dark, lacking in contrast and shadow detail, when compared with the same scan made on a scanner with a better dynamic range.
Consumer-grade scanner manufacturers--even those who have marketing departments suffering under the delusion that they make "professional" scanners-- rarely advertise their products' dynamic range. If your manufacturer won't tell you the scanner's dynamic range, it is probably junk.
Without knowing more about what you intend to use the scanner for, it's difficult to make specific recommendations. Do you use your scanner to make images for print? Do you use it to make FPOs, and let your service bureau make final scans on a drum scanner? How important is image quality? What kind of reproduction will the images be used for?
noise in scanned transparencies
Q: I recall reading ... somewhere .. a technique to reduce noise
in dark areas of scanned transparencies. You do multiple scans, save
each into a separate layer, and then average the layers to remove random
noise. My question is, how to average the layers. I seem to recall
there's some kind of exponential formula for increasing transparency of
each layer, and I know you have to select an option to control how the
A: Yes. This technique can also be used to remove grain from a photograph. The theory is the same: You snap ten identical photographs, under identical lighting, from a bolted-down tripod, then average the photos together.
There are several techniques for doing this. Mine is pretty straightforward: Start with ten scans (or ten photos), in ten layers.
Change the layer opacity on each layer, starting from the bottom up, by setting its transparency to (100/layer number).
So: The bottom layer is Layer 1. Set its opacity to 100/1, or 100%.
The next layer up is Layer 2. Set its opacity to 100/2, or 50%.
The third layer up is Layer 3. Set its opacity to 100/3, or 33%.
And so on up to the top layer, whose transparency is 100/10, or 10%.
Resolution, image size, file size
Q: I scanned my file at 2400 dpi, but the file is too big. It
looks OK on my screen but prints out too small.
A: Resolution, image size, and file size are all related. Think of your image like a tile mosaic. Each tile is a single pixel. Resolution determines how big those tiles are--that is, how big each pixel is. If an image is 72 pixels per inch, then each "tile" is 1/72 of an inch across. Likewise, if the image is 300 pixels per inch, the tiles are 1/300 of an inch across.
Resolution and print size are related in a very simple way: The number of tiles, multiplied by the size of each tile, gives you the total size of the mosaic. If a tile mosaic is 100 tiles wide, and each tile is half an inch across, then the mosaic is 50 inches wide. Likewise, if you have an image which is 300 pixels per inch, and it's 600 pixels wide, it is 2 inches wide.
Similarly, a picture which is ten inches wide and is at 300 pixels per inch must be 3,000 pixels wide. When you create an image for printed output, it is not always true that the higher the resolution is, the better. An ink-jet printer does not print a better image from a 600 pixel per inch image than it does from a 300 pixel per inch image, for a variety of reasons. Similarly, an image printed on a printing press (say, in a magazine) does not benefit from a very high resolution; all images printed on a press are broken up into halftone dots, and the ideal resolution depends on how fine those halftone dots are.
you look at an image on your screen, the image will seem to be very
large, because what Photoshop says is
"100%" magnification is not the size the
image will print! When Photoshop shows you something at 100% magnification,
what it means is that one pixel on your screen equals one pixel
in your image; you are seeing "100%" of the number of pixels
that will fit on your screen. An image that is
two inches wide at 600 pixels per inch is
1,200 pixels wide--it will more than fill your 1,024 x 768-pixel
monitor, even though when it prints out it
will only be matchbook sized.
You can change the image size of an image in two ways: by using interpolation, which changes the number of pixels, or by not interpolating, which leaves the number of pixels the same but changes the size of each pixel.
For example, let us suppose that you have an image that is 10 inches wide and 300 pixels per inch. It is 3,000 pixels wide. If you shrink it to 5 inches wide, but you have interpolation turned off, the number of pixels will not change. The image is still 3,000 pixels wide, but it is only 5 inches wide; therefore, it is now 600 pixels per inch.
Now, let's suppose you shrink the same image to 5 inches wide using interpolation. Photoshop will change the number of pixels. The image will now be 5 inches wide, and it will still be 300 pixels per inch, so it will now be 1,500 pixels across, instead of 3,000 pixels across.
Good Photocopies of Images
Q: I am making a newsletter that I want to photocopy. I
have some pictures I want to put in it. They look good when I make my
printout. Problem comes when I go to try to photocopy these. As this
production is only a small rum (100 or so), it is not really worth going
to a printer to make plates etc. Hence the desire for photocopyable
A: Being a veteran of small-press publishing, this is an area where I've had a lot of practice. :)
For good quality photocopiers, you should use an 85-line halftone screen for your images. A tighter screen will reduce the number of shades of gray your image will have, especially on a 600 dpi printer, and will not photocopy as clearly.
Very, very important to photocopy reproduction is the quality of the image going in to the process. Photocopying introduces significant "dot gain" into your images; what that means is that photocopied images become much darker.
To get good results when reproducing an image on a photocopier, you must get familiar with Photoshop's Curves command. This command is found under Image-> Adjust-> Curves. You also must have your Info palette open and be familiar with how to read the numbers in it.
.Start with a grayscale image. Set your Info palette so that one of the two readouts displays "actual color." Open the Curves window (Image-> Adjust-> Curves). Since photocopying darkens an image, you are going to lighten it. It should look too light and slightly washed-out on your screen. If it looks perfect on your screen, it will look muddy and dark when you photocopy it.
position the mouse pointer over the darkest part of the image which
still contains detail. This part of the image should not be 100%
printing according to the Info palette. Instead, it should be no more
than about 85% printing; the shadows in your image will darken
dramatically on the photocopier. If the darkest shadow detail in your
image is darker than 85%, pull down the shadow end of your curve (the
right-hand part of the curve in the Curves dialog) until your Info
palette shows it to be about 85%.
Now look at the lightest part of the image which still contains detail (not the part which should print pure white; the lightest part that shows detail). It should be 5-6% in your Info window. If it is lighter than this, that part of the image will be pure white (no detail) on the photocopier. If it is darker, lighten it by moving the left-hand point on the curve to the right until the Info palette shows you it is about 5-6% printing.
Now pull the middle of the curve down slightly. The image will look much lighter on your screen. This is what you want. Click OK in the Curves dialog. Now print the image to your laser printer, using an 85-line halftone screen. Your results when you photocopy should be much better
|Changing the color of an object in Photoshop
...is relatively easy if the original color and the color you're changing to are both about the same
lightness.The farther you have to go in changing the lightness, the more
difficult it becomes.One of the most challenging retouches I've faced along
these lines was to take a photograph of a model wearing a white tank-top for
a swimwear catalog and make her tank-top black. Doing this, and making the
result look natural and unretouched, took many hours.
Here's one way to make a light object darker:
Create a new layer. Fill it with white. Set its Mode to Multiply. This will make the layer seem to vanish.
Now, switch to the layer underneath. Examine each channel in turn, until you find the channel which offers the greatest level of detail in the object (in
this case, the eye) you wish to darken. Make a selection around the object, and copy it.
|Switch to the while layer. Go into each channel in turn and paste that object
into the channel. This will have the effect of darkening the object (because
the layer is set to Multiply) while preserving fine detail. (Note: If your
image is CMYK, not RGB, you may not want to paste into the black channel,
depending on the quantity of black generation you want.)
Now switch back to all channels. The object won't be the right color, but it
will be darker and will still preserve detail. If necessary, adjust the layer's Opacity if the object is too dark.
The color of the object can be changed by any of the normal techniques: Hue/Saturation, painting or filling in Color mode, etc.