Reproducing Photographs

A simple introduction

Peter Marshall

This really needs some illustrations and I haven't had time to prepare these yet.

Photos are continuous tone -they have a full range of tones from white to black; all current commercial printing processes can only printing black and white with no grey shades. As we know, the solution to this the half-tone, where the picture is made up of a pattern of small dots, normally too small to be seen by the eye, which blends them to a grey depending on the total amount of paper covered by the ink.

In this short note I want to examine a few of the problems involved, particularly for the producition of a magazine such as this. Images to be printed here are scanned to provide 200 dots per inch at the printed size. Each dot or pixel can have one of a range of 256 intensities from pure white to pure black.

Each pixel has to be represented in the half tone not by a single dot, but by cell containing a number of dots. To provide the same range of 256 shades, requires a 16x16 cell (16x16=256)

Scanner resolutions are normally expressed in dpi (dots per inch) and half-tone dot sizes as lpi (lines per inch), the half tone cells normally being in lines.

Each pixel on the image requires a cell in the printed image for all the information in the scanned file to be used, but in practice the scan resolution (at output size) is normally optimum at 1.5 to 2 times the final lpi. Newspapers typically use around 80 lpi, magazines perhaps 130 dpi and high quality books around 200 lpi.

Unfortunately a computer printer can only place perhaps 720 dots in an inch. For a simple 16x16 cell, his would give only 720/16 or 45 cells per inch - a 45 lpi screen. This gives a very obvious grid and very poor detail. Using a smaller cell, for example 5x5, would give roughly magazine quality sharpness, but would only allow 25 grey shades - not enough for most photographs. Software controlling the printer - the printer driver - is normally written to use a few tricks to allow rather better figures than the simple calculation would suggest, as well as dealing with other problems, giving a typical practical limit of around 80 lpi for most ink jets.

For good half tone plates (and photocopying) the dots in each cell need to be grouped - either in the centre of the cell, or, in the darker tones, around the edges of the cell leaving a white centre. (In pactice, more complex schemes may be used, and also the different dot shapes - such as elliptical - may give better resutls.) This is completely the opposite of what is needed for normal direct printing from an ink jet, where it produces smoother results by spreading the dots out fairly evenly. The drivers provided with most consumer ink jets simply do not allow you to produce suitable prints for making half-tone plates they are designed for printing pictures.


Serious Desk Top Publishing programs are generally designed to work with high resolution image-setters (often around 3000dpi) to produce good halftones. The best answer with relatively crude equipment like mine is to get the work output either to high quality `bromide' or direct to printing plate by a bureau using a such a high-end device. However to do so would considerably increase the cost - to perhaps double the current price.

Different methods are used for making the printing plate from the master image. At one time this was done photographically, but now for cheap offset printing a photocopier like device is used. This cannot cope with fine dot screens, limiting the lpi you can use for this method. More expensive work is processed digitally, so producing a paper or `bromide' intermediate is really just a wasted stage.

In the printing process itself there are particular problems at the darker end of the tonal range. When you print the ink will spread slightly on the paper, tending to fill in small white dots. This effect will depend on the paper in use. (Users of ink jet printers will have noticed this spread also as a dramatic difference in sharpness on different papers.) Cheap paper has a coarser surface and is more absorbent so you need to use larger dots (lower lpi) and also make a greater allowance for this spreading. There are also problems with reproducing very pale tones using a coarse screen. The combination of both these leads to the style of
reproduction typical of many cheap publications with everything more or less mid grey. Dull but safe - it seldom happens in LipService where I've tended to aim for higher contrast sometimes at the expense of shadow or highlight detail (or both). And sometimes I or the printer have got it very wrong. Michael Crawford Hick has kindly offered to try and keep us both in line!


Top of page    Contents Page     Next Page