Digitise Your Own Photographs

The picture to the right is scanned at a resolution of 300 pixels per inch. The original measures four by six inch, so the scan is about 1750 by 1150 pixels. On disk, the scan uses 6 MB. The picture is saved in TIFF format. In this format, every pixel uses 32 bits, 8 bits per colour channel (cyan, magenta, yellow and black; the black channel is used for printing purposes). On screen, pictures only use the three colour channels Red, Green and Blue (RGB) and not the black channel. Many other file formats, like bmp, use the same three channels, 8 bits per channel, 24 bits in total. Since 24 bits means 16.777.216 possible combinations, there are as many possible colours.

The scan has not the quality we want. The first step is retouching it, using the "clone tool": this copies one piece of the picture to another place. For instance, to retouch a piece of dust, I would set the clone tool to a width of about 8 pixels, select a piece next to the dust particle and copy it over this particle. You will never find the dust back. In the same way you can remove poles and wires from the landscape.

To cut away the edges of the photograph and give it its desired size, use the "crop tool". First, use the "mask tool" to draw a square of the desired size on the picture. Second, select "Image - Crop - To Mask".

The photograph is greyish. That's because the whites are not white and the blacks are not black. This can be improved by "stretching" the contrast. Select "Image - Adjust - Level Equalization". Drag the arrows above the histogram until the histogram exactly spans the space between pure black and white. You can vary this to adjust the contrast to your needs. The input value clipping I used to get the picture to the right is 22 and 225. As you can see, the contrast is much better now, compared to the picture above.

The colours still disappoint. As a rule, a photo leaves the lab with too much red. I have no idea why. This can be corrected by adding the opposite colour, blue. Select "Image - Adjust - Color Balance". To get the picture you see left, I set the Cyan - Red slider to -8 (adding Cyan) and the Yellow - Blue slider to 8 (adding Blue). A very nice feature is that you can confine the colour adjustment to shadows, midtones or highlights only, or to a combination. I checked the highlights and midtones, but left the shadows untouched to preserve warmth. Experimenting is important!

Time has come to correct the blur. The software automatically searches for edges and accentuates them. The result is striking: it looks as if the picture is more focussed. This can be done with different tools, but the tool I advise you is the "unsharp mask". Look for "Effects - Sharpen - Unsharp Mask" but use this tool with prudence. The percentage slider sets the intensity of the effect, the radius slider controls how many pixels are evaluated at once, and the threshold slider determines which pixels are affected. In the picture to your right, the percentage was 200, the radius 1 and the threshold 0. Avoid a too drastic effect. Later, when you compress your photograph to a jpeg file, too much sharpening will create a crumbly overall effect, as a visitor of this site pointed out to me.

Now, we will resize the picture to the size we want on our home page. Choose "Image - Resample". Check the "anti-alias" box. This smoothens the edges and prevents jagged edges and lines. Also, check the "Maintain aspect ratio" box. Then, fill in the width you wish (preferably in pixels) and click OK.

After resizing the picture, it is necessary to apply the "unsharp mask" again, but more prudently than the first time. The picture to the right has a percentage of 100, a radius of 1 and a threshold of 0.

Each of the small steps taken so far add to the quality. The overall effect, I think, is refreshing and this becomes clear when the original scan and the processed picture are compared: look below. But first I will explain you in which format digital Internet picture should be used.

You will see two types of pictures on the World Wide Web: "gif" pictures and "jpeg" pictures. The gif format is excellent for graphics with a limited number of colours. The "Home"-sign at the bottom of this page is a gif picture. The maximum number of colours saved in gif pictures is 256. By choosing a palette, the colours can be controlled. Since the number of colours is limited, this type of graphic is not useful for photographs. Nevertheless, using a technique called "error diffusion" subtle colour changes can be imitated. Have a look at the "Hiking Links" banner on my home page. This is a gif picture. A granular mix of a limited number of colours suggests a smooth change from yellow to green. To save a graphic in gif format, first convert it to a paletted picture using "Image - Convert to - Paletted (8-bit) ". Then, use "Save as" and choose "CompuServe Bitmap (GIF) " in the file type box.

When publishing photographs on the World Wide Web, however, it is better to save them in jpeg format. This so called "lossy" compression algorithm reduces the size of a picture drastically at the cost of a loss in quality. The advantage is that the full colour spectrum (24 bits) is used: ideal for photographs. Select "Save as" and choose "JPEG Bitmaps (JPG) " in the file type box. A dialogue window pops up. Set the "Smoothing" slider at 0 and choose a value for the "compression" slider. A high compression results in a smaller file size but also in a poorer quality. Try to keep the compression slider below 50. All the pictures on this page are jpeg's.