Before the advent of Windows and TWAIN, scanner users and developers faced the difficult dilemma of drivers. For every hardware device owned, users required a driver for every software program owned. When there were only three different printers and three different programs, only nine drivers were needed. But as the number of printers, programs and other devices increased, so did the number of drivers.There are currently over 200 different printer types, over 20 image-editing packages and over 20 different scanners. If every image-editing program was to work with every scanner and every printer, there would be 80,000 different drivers to make everything work with everything. Obviously the market couldn't survive like this, either some items would go unsupported or the industry had to find a better way. Enter Windows. Windows changed the way people looked at peripherals. All printers are essentially the same, varying only in detail and implementation. What Windows did was to develop generic print services, so any program that wanted to print would just print to the generic device and let Windows handle the specifics. TWAIN is the standard that does the same for scanners. All scanners basically produce the same type of information, varying only in detail and implementation. The TWAIN specification was written by a group of eight hardware and software companies. TWAIN driver, TWAIN interface and TWAIN module are all synonyms for a software module that allows image editing, optical character recognition, desktop publishing and fax software to deal with the scanner in a generic way. Users don't need to determine if "XYZ" program supports scanner "ABC" before they buy it. Instead, they just look to see if both support TWAIN. Through TWAIN and Windows, the number of required drivers drops from 80,000 to a mere 220. Popular Windows programs that support TWAIN include Adobe PhotoShop, Corel Photo Paint, WinFax Pro, PageMaker, Picture Publisher, iPhoto Plus, Fractal Design Painter, Microsoft Publisher, Wordlinx OCR and TextBridge OCR. Scanner mechanics ( how they workFor simplicity, we will discuss flatbed scanners in this document. However, the basic principles also apply to hand scanners and sheetfed scanners. You begin by placing a document face down on the glass of your scanner and closing the lid. When you press the scan button, the scanhead inside your scanner's chassis begins to move. A fluorescent light on the scanhead shines light upward onto your document or picture. This light is reflected downward and hits a mirror in the bottom of the scanhead, which reflects it towards a Charge Coupled Device (CCD), inside the scanner's chassis. Scanhead of a flatbed scanner The CCD contains an array of photosensitive cells that read the intensity and/or colour of the light that hits them. The CCD works with an analogue to digital converter to convert the light to a level between 0 and 255 (0 being darkest and 255 brightest). The CCD is what determines the optical resolution of your scanner. The most common scanners have a 300 dot per inch (dpi) CCD. This means that for every inch of width in the image, the CCD can sample 300 individual dots of colour. The scanner reads one line of data from your document at a time. If you are scanning a 6ins. wide photograph at 300 dpi, the scanner will send 6 x 300 or 1800 individual dots to the computer for the first line of data. The motor will then move the scanhead to the next line. If your photograph is 4ins. tall, the scanner will produce 4 x 300, or 1200 individual lines when scanning at 300 dpi. Each of these dots may be a different colour and they are actually more like tiny squares than round dots. If you were to take a large number of tiny ceramic tiles made in a wide variety of colours, you could arrange them in such a way that they would show a picture. When viewed from a distance, your tile mosaic would not look jagged at all. But if you came up for a closer look, you would see jagged edges between various colours. The image produced by your scanner is basically a tile mosaic made of many very small square dots. Using the example above of a 4 x 6ins. photograph, the scanned image would consist of 2,160,000 dots. This is the product of 1200 x 1800 which were calculated in the above paragraph. Your scanned image does not appear jagged because the dots are very small. If you scanned in an image of a postage stamp and magnified it five times, you would see the mosaic dot pattern detected by the scanner. The stamp will appear on your screen actual size. This is because computer monitors have a resolution of 72 dpi. (Hint: Scan at 72 dpi for web pages. If you scan higher than about 100 dpi, your images may not fit on your web page unless, of course, they're as small as postage stamps.) These tiny squares of colour are the data that a scanner sends to the computer. Once you get the dots into your computer, different software applications handle them in different ways. Scan modesComputers represent pictures in a variety of ways. The four methods that are most common are line art, halftone, greyscale and colour. Line art: Line art is the smallest of all the image formats. Since only black and white information is stored, the computer represents black with a 1 and white with a 0. It only takes 1-bit of data to store each dot of a black and white scanned image. Line art is most useful when scanning text or line drawing. Pictures do not scan well in line art mode. Halftone: While computers can store and show greyscale images, most printers are unable to print different shades of grey. They use a trick called halftoning. Halftones use patterns of dots to fool the eye into believing it is seeing greyscale information. Greyscale: Greyscale images are the simplest of images for the computer to store. Humans can perceive about 255 different shades of grey. Computers represent greyscale information by storing a number from 0 to 255 in a single byte. When you view a greyscale image, it is equivalent to seeing a black and white photograph. Colour: Colour images are the largest and most complex images to store. TVs and computer monitors mix the colours red, green and blue to display all the colours visible to the human eye. If you were to look at your computer screen right now through a high powered magnifying glass, you would see that the white background of this page is actually made up of high intensity red, green and blue dots. A pixel is a group of three dots, one of each colour. Because the dots are very tiny, your eyes blend them together and you see white. The monitor's internal electronics can vary the intensity of each colour dot to 256 different levels of intensity. At the 0 intensity level, the dots are completely off and the screen appears black. If the red and green intensity is 0 and the blue intensity is 255, you will see a rich blue colour like the one above. By varying the intensity of each colour dot between 0 and 255, there are 16.77 million different combinations. Each combination appears as a different colour. If the intensity of each dot is set to an equal value, say 128, the monitor will appear as a grey shade. 128 would be the level of 50 per cent grey. This is why there are only 256 greyshades when scanning in greyscale mode. Computers use 8-bits (1 byte) to represent each of the colour components (red, green and blue). With 8-bits for each of the three colours, there are a total of 24-bits to represent the entire colour spectrum. How to determine file size If you're curious how to determine how large a scanned image: File Size = (Resolution x Horizontal Size) x (Resolution x Vertical Size) x Scan Mode where Scan Mode = 1/8 for line art and halftone, 1 for greyscale and 3 for colour Here are the file sizes for a 4ins. x 4ins. photo at various scan modes and resolutions. Notice that the greyscale files are 8 times as large as the line art files. The colour files are 24 times as large as the line art files and 3 times as large as the greyscale files. InterpolationThe resolution of a scanner is determined by the optical resolution of the CCD and the Stepping Speed of the scanner's motor. A 300 x 600 dpi scanner has a 300 dpi CCD and a motor that goes slow enough to scan 600 lines per inch as it travels the length of the bed. If you scan at 300 dpi on such a scanner, the motor runs twice as fast as it does when scanning at 600 dpi. If you scan at 600 dpi on such a model, the motor runs slower and the scanner's hardware interpolates the horizontal data from 300 up to 600 dpi. Basically, an integrated circuit chip in the scanner generates new data where there is none through an algorithm by averaging the colour of adjacent dots and creating a new dot between them of the average colour. This is Hardware Interpolation and it allows a 300 x 600 dpi scanner to produce a 600 x 600 dpi image. Software Interpolation can increase the resolution even more than Hardware Interpolation. Software Interpolation is performed by the TWAIN driver within the Computer's CPU. This type of interpolation is quite misleading. It does not create sharper images. Image quality and sharpness is always limited by optical resolution. Software Interpolation merely increases the amount of data in a scanned image. It is roughly equivalent to scaling an image to make it larger. The only good reason to scan at very high resolutions, such as 4800 dpi, is when you need to enlarge an image dramatically. If the postage stamp used above were scanned at 4800 dpi, the file size would be 19.1 Mb. This would increase to 76.4 Mb at 9600 dpi. If you were even able to scan an 8.5ins. x 11ins. page in colour at 9600 dpi, you would need a 12 Gigabyte hard drive to store the resulting file. (Obviously, your computer would crash if you tried this.) All graphic image file formats fall into two categories, either Bitmap formats or Vector formats. Vector formats use mathematical formulas to recreate graphics. If the graphic is a circle, the file format will have the location of the circle's centre, its radius and other pertinent information, such as what colour to draw the circle. Vector formats are easily scalable without creating jagged edges, but they are only suitable for simple drawings with a few colours. Scanners do not use vector file formats. Bitmap formats store the actual colour of each dot in the image. They are more suitable for scanned images due to the large number of colours and complex tonal variations in a photograph. A disadvantage to Bitmap formats is that they create large files that take lots of drive space to store. This problem is overcome by several bitmap file formats which compress the data prior to storage on disk. Most scanning programs, including iPhoto Plus and Picture Publisher, use the TIFF file format as their default. TIFF files use compression but does not shrink the file sizes significantly unless the image is black and white (Scanned in Line Art Mode). Another popular file format is the BMP file. This is the format Microsoft uses for Windows Wallpaper. So if you want to use your own photos as Wallpaper, save them in the Windows directory in the BMP format. The two most popular bitmap file formats today are GIFF and JPEG. Both of these formats are popular on the Internet because of their high compression ratios. GIFF compression uses an algorithm similar to ZIP files. It also reduces the number of colours in the image. A scanned image can have up to 16.7 million colours. A GIFF file reduces this down to a maximum of 256 colours. With appropriate utilities, you can reduce this even further to 16 colours. Obviously, you wouldn't want to do that with a photograph though. JPEG does not reduce colour depth. It merely uses compression algorithms similar to ZIP files. This is the most popular format for scanned images on the Internet and for sending pictures as attachments to email. Compiled by Rachel Hodgkins (c) 1999 Mustek Inc.
This was first published in September 1999