Setting the black and white points

Introduction

Setting the black and white points of an image is probably the single most effective way of improving scanned images. This page aims to demonstrate how to set the black and white points using a practical example. A far more rigorous description of this and other techniques for improving your scans can be found on Wayne Fulton's site www.scantips.com.
This article descibes how to set the black and white points using Adobe Photoshop. An article describing how to do this using Paint Shop Pro can be found here.

The example image

This picture was taken on 35mm slide film and scanned on to Kodak PhotoCD (click on the image to see a larger version):

The image is rather dark and is lacking in contrast. In case you are interested this picture was taken at Alum Bay on the Isle of Wight, England in about 1976.

The histogram or "levels" display

Most scanner software and image manipulation programs have some sort of histogram display. This is the "Levels" dialog from Abobe Photoshop 4.0 LE:

The histogram is the the graph on the left with three little triangles underneath it. The histogram displays the distribution of bright and dark pixels. The left side of the histogram represents black pixels and as you move towards the right pixels get brighter and brighter until you reach white pixels on the right side of the histogram.

Adjusting the black and white points

If we look at the histogram display we can see that most of the pixels are in the left half of the graph. This is what we would expect with a dark image like this.
The darkest pixels are a little to the right of the edge of the histogram. This means that the darkest portions of the picture are a dark grey rather than black. We can make these dark greys black by dragging the black triangle so that it is underneath the start of the graph - this is known as setting the black point.
The lightest pixels are near the middle of the graph. This means that the brightest portions of the picture are a dull mid-grey. Most photographs have bright highlights which should be white and not the dull grey we have here. We can set the white point (i.e. make the brightest pixels in the picture white) by dragging the hollow triangle on the right so that it is underneath the right end of the graph.
The "levels" dialog now looks like this:

If you look at the "Input Levels:" line above the histogram you can see that we have set the black point to 13 and the white point to 146. In the original image the darkest pixel had a value of 13 (on a scale of 0 to 255 where 0 represents black and 255 represents white) and the lightest pixel had a value of 146. By setting the black and white points we have ensured that the darkest pixels are black and the lightest ones are white. The original and adjusted images are shown below:

This is done in a slightly different way in Paint Shop Pro and is described here.

When to set the black and white points

Most scanner software has a histogram display and allows you to set the black and white points when the image is scanned. In the example above the black and white points were set after the image was scanned. So when is the best time to do it?
The short answer is that it is best to set them when the image is scanned. The reason is that we are aiming to extract as much information as possible from the scan. The image loaded in our image editor is made up of 8-bits (which represent 256 levels) of red, green and blue giving a total of 24-bits. Most scanners are capable of scanning 30-bits or 36-bits but they output an image with 24-bits. This means that some of the information read by the scanner has to be thrown away. By setting the black and white points when we scan the image we are sure that we are extracting the maximum amount of information from the scan and that the parts that are thrown away contain no information.

Unsharp mask

You may feel that the adjusted image doesn't look very sharp. This is often the case with scanned images and can be corrected by applying an unsharp mask filter. How to do this is described on a separate page.

chris.breeze@iname.com