The best explanation of the process I can find is from Digital Prepress Complete by Donnie O’Quinn and Matt LeClair, published by Hayden Books, 1996. It is a review of the process from the scanned image to the printed piece.

In scanning, a photographic image is converted to squares called pixels that make up a map of the image. This map is used to print film, like a mosaic. The pixellated image is placed into a page layout document, which translates the image to an output device using PostScript code through RIPping software.

This code translates the square pixels into round or elliptical dots. The dots usually take the place of several pixels and have neither the color of the pixel nor do they cover the entire area the way the pixels did. The dots create an optical illusion by changing their size so that the tone appears continuous when viewed from a distance. They are able to do this because pixels described the space.

The dots are printed on film as a negative of the original layout. The film makes a plate, which prints to a blanket on the press that prints to paper. The image flip flops negative to positive to negative to positive again, so you see the image correctly when it comes off the press.

When an image is scanned, the tonal changes are converted into a gradient grid much like the picture to the left.

 

 

When it is translated into film for printing, the pixels become corresponding halftone dots like the picture to the left.

 

Each printer dot is made into a sort of round shape outlined within a grid like the picture to the left.

 

Halftone dots grow in size as they are pressed against the paper, this is known as dot gain. Round halftone dots tend to create shadow areas that are too dark, due to the filling in that occurs as the edges of the dots connect. Elliptical halftone dots are widely used because they tend to minimize midtone jump.You will need to adjust for this in the process of creating the perfect scanned image. Higher line screens and more porous papers require more compensation than do coarser screens and coated papers. This adjustment can be made in the densitometer settings of the scanner interface or in Photoshop after the scan has been completed. The goal is to have a balanced image that takes up only as much space is required to keep quality high.

A screened image is made up of countless tiny dots with the bright (highlighted) areas consisting of a few small black dots totally surrounded by white space; the middle gray areas (midtones) consisting of medium-sized dots; and the darkest areas (shadows) consisting of dots that are so large they overlap, with very small areas between them.

Bit depth
In order for the gradient to be very refined and capable of printing millions of colors, each pixel can have multiple layers of bits assigned to it to define a different color.The bit depth ranges from 1 bit which can only define black or white, to 4 bits, which can define 16 shades of grey, to 8 bits which can define 256 shades of color or grey, to 24 bits which can define millions of colors. The bit depth quantity requires a corresponding amount of information to be defined and since the image is defined by a grid, the physical size of the image also adds to pixel information; the more information the image contains, the larger the file size will be.

Basic guidelines for scanning an image:
Determine the paper type and printing method your image will printed on.
Determine the dpi which will be needed to gather the maximum pixel information needed to create a clean, crisp image. (2xls=dpi)

Start with a high quality, glossy photo or transparency with good contrast of highlights, midtones and shadows. “Garbage in is garbage out.”

Making Adjustments to Greyscale Scanned Images

Once the image is scanned, use Photoshop to adjust for the best possible output.

Unsharp Mask
Apply the Unsharp Mask filter in Photoshop to increase the contrast between neighboring pixels. these filter combines a slightly blurry version of the image with the original. It can sharpen only the edges in an image or it can sharpen any portion of an image accordingly. This can result in sharp details in high-contrast areas (the areas where adjacent light and dark samples are markedly different) without accenting tonal shifts in low-contrast areas (smooth gradation w/subtle transitions).

If too strong, can create a halo effect and look artificial.
With lower resolution scans can cause the jaggies by aliasing diagonal edges.
Speckling and mottling accentuates irregularities of the image. Run a despeckle filter before unsharp mask to resmooth gradations.

In the amount dialog box, enter a value between 1 and 500 percent. Higher values produce more pronounced effects. As the image gets larger, detail becomes more apparent so try decreasing the amount on larger images. Likewise, images with crisper edges will require less sharpening.

The radius indicates how large of an area of pixel values will be changed. In the Radius dialog box, enter a value=resolution divided by 200 (85 line=.425, 200 line=1.0, 300 line=1.5)

Threshold describes how much difference there has to be between the pixels before the sharpening happens. A higher threshold means that only more marked differences are considered edges, a lower one means that less tonal variation is required. 0 sharpens everything, 255 won’t sharpen anything.

Levels
Adjust the output sliders to values such as:

Printing stock	Output levels
Newsprint	12/88
Uncoated stock	10/90
Coated Stock	12/43

Adjust the Input Gamma levels:
Use Image/Adjust/Levels to pull up a histogram of the image. Touch the black eyedropper on the left on the darkest shadow pixels in the image and the white eyedropper on the whitest highlighted area of the image. Slide the left and right triangles toward the center to encompass the bulk of the histogram. Make sure the preview box is on to check the image visually on the screen. Move the midtone triangle a little to left of center until the image looks right on the screen (1.5) then up a little more to adjust for monitors backlight. Light to the eye on the monitor is better, too many desktop scans are too dark when output. Watch out for posterization in the shadows and blowout in the highlights.

Screen Angles
The black and white halftone dot grid is usually printed at a 45 degree angle to make it less obvious.
In the case of a duo tone, the colored film is set at 45 degrees and the black film at 0. A tri-tone shares three screen angles set at 30 degrees each.

In the case of color, CMYK screen angles would be C=15 degrees, M=75 degrees, Y=0 degrees, K=45 degrees. This arrangement of screen angles creates a rosette pattern. An open-centered rosette is more robust and resists color shifts, even when slightly misregistered. A closed-centered rosette lets less paper show through so color will be more brilliant, creates less pattern and is good for lower screen frequency, but registration is critical.

Moiré patterns
If screens are misaligned or you need to rescan something that has already be printed (thus containing a grid of dots) the result will be an unsightly woven effect of visual interference.

Solutions
In the case of misalignment of screens, try swapping the magenta and cyan angles or up yellow’s linescreen by 1.08 per cent.

When scanning a previously printed piece, try rotating the image at several different angles when you scan it, then rotate it to 90 degrees with the rotation feature of the crop tool.

Try scanning the image at twice the resolution you need (4xls), then apply a Filter/Gaussian blur at, say .9, Filter/Despeckle, cut the resolution in half, blur again as needed, touch up with Filter/Unsharp mask.