For DTP and Prepress personnel, threshold can be the name given to a control in scanning software or in an option window used for converting existing greyscale pixels to black or white 1Bit line image pixels.

It would then be a situation where software would have to decide that if the grey value of a pixel was equal to or more than the threshold value, then that pixel has to be given a new value of 0 (RGB black) and if the pixel grey value was less than the threshold value, that pixel would be given a value of 255 (RGB white).  And so it would go on through every pixel in the image.

Even when we scan a line copy and select the 'Scan to Line' or 'Scan to B/W' option, the light sensitive array of CCDs (using a flatbed scanner) will only "see" variations in light strength and send to the internal components, variations in electrical current that will be converted to one of the many grey colour values that the scanner can support (i.e. 256) - for each CCD created pixel. This never varies, a digital scanner always produces grey values.

Then either the scanners inbuilt software or, in the case of a programs conversion option, program algorithms will have to decide at what colour value (the threshold) pixels become either black or white. (user controls might present the range as 0% to 100%, or 8bit 0 to 255)

I have used the following example for a few years now and is from an exercise that students have reported as simple but very explicit in the way it explains some decision making processes that go on inside a scanner or conversion software. Students have to colour in boxes by making their own decisions, but here I have done it for you. (no crayons on the screen, please!)

You can see the line grids of  two virtual scanning resolutions over an image being scanned. At the top left is an example of one of the virtual pixel cells and the objective is to decide whether the average area covered by any portion of image in a pixel cell is more or less than half the total cell area.

Supposing a threshold value of 50% has been chosen, colour each pixel cell solid black if the image portion is over the threshold, and leave the pixel cell clear if the image portion is less than the threshold.

Note:The results are shown on the right hand side of each grid. Just remember that each individual CCD only sees a light variation or value, it does not see the actual shape as we do here, but the exercise should drive home the principle involved.

Detail capture

Once completed, another very important observation that comes out of this exercise is the increase in detail as the resolution is increased. In the 'Flatbed Scanning Hints' area on the 'Brightness and Contrast' page the importance of achieving maximum detail is covered with some simple graphic examples. When bitmap scanning fine detail from a line copy a judgement does have to be made about what fine detail in the original is most important to the result. It is often a case of 'keep a little and lose a little'. Prepress technicians are used to looking for apparently subtle differences, but once so trained it makes a world of difference to the results

Because CCD scanners always physically scan at the same size and resolution - with a few exceptions described in the 'High End & Flatbed Scanners' page - the quality (number of CCDs per inch) of the scanner can have a dramatic affect on the amount of fine detail achievable.

If your line image is to be finally output via a high resolution device like an imagesetter, NEVER use a laser printer to determine quality, use the high resolutions described on the 'Resolutions' page.