With special reference to creating digital images of philatelic materials
By Peter G. Aitken x
Click here to send me e-mail. I would enjoy hearing any comments or suggestions related to this page.
Back to main stamp page
Selecting a Scanner
Choosing Scanner Settings
Which File Format?
Calibrating Your Scanner
More on Color
About Image Size and Resolution
Reading Difficult Postmarks
Lifting Postmarks and Overprints
Detecting Forgeries and Printing Varieties
Scanning and Uploading Images for On-line Auctions
What About Digital Cameras?
Many stamp collectors and dealers are interested in creating digital images of their stamps, covers, and other philatelic material. There are many uses for these images, including computerized databases, Web pages, and on-line auction listings. The procedure is not all that difficult, but it can be a bit confusing to the newcomer. I have gathered some basic information to help you get started creating digital images of your stamps and covers.
Scanners come in a variety of configurations. The type you should consider depends on both your planned scanning needs and your budget.
- Drum scanners. This type of scanner provides the highest level of image quality. They are typically found at professional printing businesses. In a drum scanner, the original is attached to a cylindrical drum and rotated past the sensing elements. These scanners are very expensive, with capabilities that go well beyond the needs of desktop scanning.
- Flatbed scanners. This type of scanner provides a flat glass surface onto which the original is placed. The illumination and sensing elements move under the glass to scan the image. Flatbed scanners are available in a wide range of sizes, prices, and capabilities. Some flatbeds offer a transparency scanning adapter as an option.
- Single sheet scanners. This type of scanner is designed for single sheets of paper. You insert one edge of the paper in a slot and the scanner grabs it, feeds it past the sensing array, and passes it out the other side. Some single sheet scanners are even integrated into keyboards. Such scanners were originally designed for digitizing documents and images for archiving, and many models are not suitable for creating high-quality images.
- Sheet-fed scanners. These scanners take a stack of pages and scan them in sequence while you get coffee. Some even do duplex (2 sided) scans. They are very useful in some situations but not much use to the philatelist.
- Photo scanners. This type of scanner is designed to scan snapshots up to approximately 4´6 inches in size. Some are separate desktop units, others install directly into a computer much like a diskette drive.
- Hand scanners. This type of scanner requires the user to manually scan an image. Hand scanners look something like an overgrown mouse. To scan, you manually drag the unit over the original document. Handheld scanners are suitable only for small originals that are no wider than the scanner itself. In theory, most hand scanners permit you to scan a wide original in two or more passes and "stitch" the scans together into a final image. This, however, never works as well as the manufacturers claim.
- Slide scanners. This type of scanner is designed for scanning slides (transparencies) rather than opaque originals, such as photographic prints. While rarely relevant for scanning philatelic material, a dedicated slide scanner is the best choice for scanning slides. Some flatbed scanners come with transparency adapters but they do not provide top quality results, particularly with small slides such as 35mm. Slide scanners have very high resolution, typically a minimum of 2400 dpi, required for getting all the details out of your slides. Many slide scanners also have the ability to scan color negatives and to convert the negative image to a positive image.
For philatelic purposes, a flatbed scanner is undoubtedly the most versatile. You can scan anything from a single stamp to an entire stockbook or album page. A hand scanner may be a viable alternative, particularly if your budget is tight, although the prices of flatbed scanners are so low that this is rarely a real consideration. The width limitation of hand scanners does not matter so much for stamps and covers. I have also seen sheet scanners and photo scanners used successfully for philatelic purposes, although they require that the item being scanned be sandwiched between clear plastic sheets for feeding into the scanner.
One additional factor to consider when selecting a scanner is how the scanner interfaces with your computer. The interface does not have an effect on image quality, but does influence scanning speed.
The most desirable and common type of interface is USB, which stands for Universal Serial Bus. This interface combines the advantages of being fast and easy to set up. Also, if you have more than one USB peripheral they can be daisy-chained off the same USB port on your computer. USB consists of a single connection similar to a telephone wire between the computer and scanner. USB has been around for long enough that it is rare to find a computer that does not support it, and if your system does not support USB then you have a real antique pn your hands! Also, your operating system must support USB, and this is universal with all versions of Windows since Win 98. The standard today is USB 3.0 which is a lot faster than the older USB 2.0, buyt this makes little practical difference when scanning because it is the mechanical scanning process that limits scanning speed, not the interface.
Firewire, also known as IEEE 1394, is a connection technology originated by Apple for the Macintosh. From the end-user's perspective (ease of setup and speed) it is very similar to USB. Some scanners are available in Firewire versions, so if you have a Mac this may be a viable choice for you. I am not reallly familiar with Macs, but I believe they all come with USB as well.
Just for completeness I will mention two other ways to connect a scanner, SCSI (Small Computer System Interface) and the printer port. I haven't seen either of these used for a scanner in ages, so it's very unlikely you will encounter either one.
Once you have decided which type of scanner suits your needs, you then need to determine the scanner specifications, which are important in determining the quality of the scans. As you might expect, scanners with better specifications cost more. The good news is that you can get a scanner with very good specs for a very reasonable price. In fact, it's almost impossible to buy a concumer-level scanner that does not have great specs, but even so some background information may be useful. Scanner specs fall into two categories: color depth and resolution.
Color depth determines the number of different colors the scanner can record. As far as I know, every scanner sold today offers a color depth of at least 8 bits per color, which is sometimes specified as an overall color depth of 24 bits (8 bits for each of the three primary colors). This means that the scanner can differentiate 256 different levels of each primary color (red, blue, and green). Because digital image files also use 8 bits (one byte) per color for each pixel, this works out perfectly.
Almost all scanners on the market today offer color depths of 10, 12, or even more bits per color, meaning that 1,024, 4,096, or more different color levels can be detected for each primary color (red, green, and blue). For example, a scanner that is advertised as "48 bits" will have 16 bits of resolution in each of the three color channels. This does not mean that your final digital image file will contain all of this information because, as mentioned before, the files are limited to 8 bits per color. Rather, the scanner hardware and/or software will process the 10, 12, or 16 bits of color information in each channel to generate an 8-bit value that will be more accurate than if the scanner had only 8 bits of information to start with. Scanners with higher color depths still create 8-bit-per-color files, but they can produce superior results with originals that have a greater difference between light and dark areas (known as a large dynamic range).
For philatelic purposes the advantages of going higher than 12 bits per color are likely to be minimal at best. This is because stamps and covers do not have the wide range of brightness that is found in original photographs. You are most likely to notice a difference with modern stamps that have bright colors or use foil, holograms, and other nonstandard elements. There's no harm in using a scanner with a higher color depth, however.
The other scanner spec you need to be aware of is resolution (dots per inch, or dpi). Because scans have two dimensions - height and width - the resolution of a scanner is expressed as two values, pixels per inch horizontally and pixels per inch vertically. Strangely enough, these values are not always the same for a given scanner, so you will see resolution specs such as 600 1200.
Evaluating a scanner's resolution spec is easy - higher is always better. You can use a scanner at a lower resolution than its specification (this is a software setting), but you cannot go higher. Higher resolution is also more expensive, and it's pointless to pay for capabilities you won't use. However, scanner technology has advanced so that even inexpensive scanners offer all the resolution that a philatelist is likely to need. Typical resolution specs start at 600 x 600 and go up from there; it is very rare these days to see a scanner with lower resolution.
It is essential that you look at a scanner's optical resolution and not its interpolated resolution. Optical resolution refers to the actual capabilities of the scanner's optics and electronics. A scanner's optical resolution is the most important factor in determining the quality of your scans. Interpolated resolution is a software trick that generates additional pixels by averaging adjacent "real" pixels. For example, if the original image has a black pixel next to a white pixel, the interpolation process would generate a gray pixel between them. Interpolation does not add detail to an image, and instead tends to soften the image, making it appear less sharp. Who is to say that between the black pixel and the white pixel there shouldn't be a red pixel instead of the gray pixel that the interpolation process inserted there? Adding pixels by way of interpolation has its uses at times, but be sure that you take into account a scanner's true optical resolution when making a purchase decision.
For general purpose philatelic scanning an optical resolution of 600 x 600 is satisfactory. For special purpose work, such as enlarging small areas of stamps to examine printing varieties, 1200 x 1200 is preferable.
You should not shop for a scanner on the basis of price alone, particularly if you are doing critical work and image quality is important to you. One "600 dpi" scanner is not necessarily the equal of another. If you have done any shopping you have seen that scanners with the same DPI specs can have widely varying prices, sometimes by more than a factor of 2. Are the more expensive scanners a rip-off? Not necessarily. Component quality can vary widely and better components cost more. Among the things that your extra $$ may buy are:
Unfortunately it is impossible to quantify these factors, at least with the information that the scanner manufacturers provide. Among consumer-level scanners, a higher price does not always mean better performance, nor does a brand name necessarily mean anything. Advances in technology mean that you can get an excellent scanner for a very reasonable price - around $100 for 600dpi and under $200 for 1200dpi. I strongly recommend relying on reviews that test the performance of specific models. If you Google "scanner reviews" you'll find lots of information..
The scanning program will be provided with your scanner. Most of these work as stand-alone programs--in other words, you start the program, make the scan, and save the scanned image as a disk file, which you can then open in your graphics program for further manipulation. Some integrate with graphics programs, which means you scan directly into the graphics program and save the file from there.Since you'll probably be using the scanning program that came with your scanner, there is no choice to be made here. When it comes to the graphics program, however, you have plenty to choose from. Almost any graphics program will handle the basic image manipulation tasks that are required for creating images of philatelic materials, so there's no need to spend hundreds of dollars on a high-end program like PhotoShop or Corel Paint. If you do not have an image manipulation program, I recommend either Paint Shop Pro or Adobe PhotoShop Elements, powerful programs at quite reasonable costs.
When you scan a stamp or cover, there are two settings you will have to make in the scanning program: resolution and color depth. Here are some tips to help you make the best settings for your purposes.
There are three color depth settings that will be useful when scanning philatelic materials. The one you'll probably use most often goes under different names in different manufacturers' scanner programs: True Color, 24 Bit Color, and Millions of Colors are some of the terms I have seen. Images scanned with this color depth can display over 16 million different colors permitting accurate rendition of all the color variations in the item being scanned.
Scanner programs usually offer a 256 Color mode. Images scanned with this color depth setting display a maximum of 256 different colors. 256 Color images can be very realistic, particularly for originals that have a limited range of colors to begin with. The advantage of 256 Color images is that the files are significantly smaller than true color image files. This color depth may be suitable when scanning stamps that are printed in only one or at most a few colors.
Gray Scale or 256 Grays mode discards all color information and creates an image that can display black, white, and 254 shades of gray in between. Gray Scale images can be useful for specialized purposes, such as when you are creating an image for evaluation of perf condition or centering.
When in doubt, I suggest that you always scan in True Color Mode. You can always use your graphics program to convert a True Color image to 256 Color or Gray Scale mode.
Selecting the resolution of your scan is perhaps the most important choice you'll make. In selecting a resolution, you need to take into account the intended use of the scanned image. Here's why.
Just like the scan you make has a dots per inch resolution, what the image is displayed on does too. Whether the image is displayed on a screen or is printed on paper, there is a resolution associated with the output device. A computer monitor might have a resolution of 72 or 96 dpi, while a laser printer might be 600 dpi and an ink jet printer can be anywhere from 300 to 720 dpi. The optimum results are always obtained if the scanned image's resolution is selected based on the resolution of the output device. There's more to consider, however – specifically, the desired output size. Let's consider the situation for images that are to be displayed on-screen.
The default display of images on the screen uses one screen pixel for each pixel in the image. This gives the ideal viewing quality. Of course you can force the image to display at a different size, larger or smaller than its "natural" size. Let's see what happens when an image is displayed on-screen at a size other than its natural size:
When scanning images for screen display, you can calculate the ideal resolution as follows. First, determine the relative size at which the image will be displayed. If the image will be displayed at half size, this factor will be 0.5; if it will be displayed at twice its actual size, the factor will be 2.0. Then, multiply this factor by 96 (the most common screen resolution). The result is the scanning resolution you should use. If your scanning software does not offer the precise resolution you calculated, select he nearest value. For example, if the ideal resolution is 192 dpi, use 200 dpi.
There's a simple formula you can use to determine the ideal scanning resolution. First, let's define some terms:
SR = ideal scanning resolution in dots per inch
DR = resolution of final display device in dots per inch (96)
OW = width of the original being scanned in inches
DW = width at which the image will be printed or displayed in inches
SR = DR ´ DW / OW
With this formula, you can easily determine the ideal resolution at which you should scan. Unfortunately, things usually aren't that simple. There are a number of factors that can, and usually do, prevent you from using that "ideal" resolution when scanning:
- You are not sure of the final use of the image - how it will be reproduced and at what size.
- There are multiple uses intended for the image - for example, you want to make printed copies as well as display it on a Web page.
- The calculated ideal resolution is an intermediate value, such as 117 dpi, that is not supported by your scanner.
In these and other cases, the general rule is to "move up." In other words, you should always move to a higher resolution rather than to a lower one. Thus, if you cannot use the ideal resolution of, say, 117 dpi, you should scan at 150 dpi rather than at 100 dpi. Likewise, if you plan to use the image for printing at 300 dpi as well as for screen display at 72 dpi, use 300 dpi when scanning if possible. Scanning at a higher resolution captures the maximum amount of detail, from the original. You can always throw away some of that information by reducing the image's resolution after scanning (in your graphics program), but you cannot regain information if the image was scanned at too low a resolution. The process of changing an image's resolution is called resampling. (Resampling is covered below) If, for example, you scanned an image at 300 dpi for printing, you could resample it at 72 dpi for screen display.
Another advantage of scanning at high resolution is that the resulting images are easier to edit. Certain editing operations, such as rotating an image, give better results with high resolution images. Rotating an image can be useful if the item was not perfectly aligned in the scanner, and is usually easier than rescanning.
The following figures show the effects of scanning at different resolutions. The first set shows fours scans of a stamp done at different scanner resolutions. The images are displayed at their "native" sizes, one screen pixel per image pixel.
And here are the same 4 scans displayed at the same size:
It's clear which resolution gives the best image. There's a tradeoff, however (isn't there always?). The 25 dpi file is approximately 2KB in size and the 200 dpi file is 45KB, with the others being proportional. For use in databases and publishing the file size may not be much of a consideration, but for Web publishing it certainly is. The most beautiful image in the world is not much use if visitors to your page won't wait for it to download! I feel that for Web publishing of stamp images 100 dpi offers the best compromise between image quality and file size. On most user's screens, the image will be slightly larger than life-size, ideal for viewing.
Be forewarned that performing a scan at high resolution can be a slow process. A lot depends on the speed of your scanner and the scanner-computer interface, but the scan itself and the transfer of data to the computer can take several minutes or longer.
What about printing? Printers have a "dpi" setting too, but because printer "dots" are not the same as pixels on a monitor the situation is different. The minimum scan resolution for best quality printed output is one third of the printer resolution. For example, when scanning for my 720 dpi inkjet printer I use a resolution of 240 dpi.
Most graphics programs let you change the resolution of an image, a process called resampling. If you scanned a stamp at 300 dpi and need a 90 dpi copy for your web page, this is how you do it. Note that resampling to a lower resolution works very well, but resampling to a higher resolution often gives less satisfactory results. This is because the resampling program must generate new information. Suppose you are resampling a 100dpi image to 300 dpi. The new image will have 3 pixels for each single pixel in the original image, so these pixels need to be "guessed at" in a manner similar to the process of resolution interpolation discussed above under Scanner Capabilities.
When resampling an image remember to always save the resampled image under a new name. This ensures that the original image file will remain available.
For saving local copies of your scanned images you have many choices of file format. The Windows Bitmap (BMP) format is widely used, particularly by people using the Windows operating system. Perhaps the most widely recognized file format is Tagged Image File Format (TIF). Unless you have a specific reason to use another format I doubt you can go wrong using BMP or TIF. The resulting files are relatively large but unless you are scanning a large number of stamps this should not be a problem, particularly with today's huge hard disks. You can also use your graphics program's proprietary format, if you prefer, but this may limit the extent to which you can share your image files with others without converting to a different format.
For Web publishing your choices are much more restricted. At present there are only three graphics file formats with wide support on the Internet: Graphical Interchange Format (GIF), Joint Photographic Expert Group (JPG) format, and Portable Network Graphics (PNG) format.
GIF format can be used only with 256 color and gray scale images. Since many stamps, particularly older ones, were printed in a limited range of colors, the GIF format is perfectly adequate. When the purpose of an image is to show the stamp's centering and perforations, GIF will always serve perfectly well. Its advantage over the JPG and PNG formats (discussed next) is that the file size is often smaller. GIF files are compressed to save space using a loss-free compression scheme, which means that the image you get out is exactly the same as the image you put in.
How can a GIF file, with only 256 colors, give decent reproduction? Here's how it works. When you save a file in GIF format, the software analyzes the image and selects the 256 colors, from some 16 million that are available, that the image will use (this is called the palette). For example, with a stamp that uses shades of red on a white background, the palette will consist entirely of shades of white and red - enough different shades to give an excellent reproduction. Look at the following example (Japan Scott #195) to see what I mean.
|Saved as GIF|
|Saved as JPG|
For True Color images you must use JPG or PNG. A True Color image can display over 16 million different colors. For accurate reproduction of stamps with many colors, such as those that use photographs, one of these two formats is required.
The JPG format introduces one additional complication. Like GIF, JPG is a compressed format. Unlike GIF, JPG uses a lossy compression algorithm which means that some information is lost during compression - in other words, the image you get out is not as good as the image you put in. When you save an image as a JPG file you must specify the level of compression to use. More compression results in a smaller file and lower image quality. At lower compression (higher quality) levels, JPG files give you perfectly good images as you can see below.
The images below give you an idea of the effects of different levels of JPG compression. I used a program that offers compression levels from 1 to 99. Be aware that other graphics program divide the JPG compression range differently, 1-10 for example or 0-255.
Compression 15 - 15KB
Compression 40 - 8.5KB
Compression 90 - 3.1KB
You must decide what level of compression suits your needs and file size restrictions. A level of 90 is clearly unsuitable, while a level of 40 would probably be okay for non-critical uses. Compressing at a level of 15 certainly gives the most attractive and detailed image, but at a significant cost in file size. Be warned that different programs express JPEG compression differently--with some, higher numbers mean higher quality (and hence) larger files, while others use higher numbers to mean more compression (less quality and smaller files).
Another aspect of JPG files I will mention is called generational loss. If you open, edit, and save a JPG file multiple times, a little bit of quality is lost each time due to the compression procedure. This is rarely of any real concern but, if you will be doing a lot of opening and saving with an image, you might want to keep it in a different format, such as TIF or BMP, and then save it as JPG only once it is finished.
A third relatively new Web file format, PNG (for Portable Network Graphics), is now widely supported. PNG combines GIF's loss-free compression with JPEG's 24 bit color support. For many uses, however, JPG remains superior to PNG because of file size issues. Using a low level of JPG compression usually gives perfectly acceptable image quality with a significantly smaller file size than PNG.
I have noticed a lot of confusion about size and resolution of digital images. I'll try to provide some clarification here. Note that this discussion applies only to bitmapped digital images, which are the kind you are dealing with when you use a scanner or digital camera. Vector images are another matter but they are not relevant in this context.
First of all you need to realize that a digital image really does not have a physical size. To say that a digital image is, say, 4x6 inches is meaningless. After all, the image is just a collection of pixels stored in memory or on your hard disk. A digital image does, however, have a pixel size - the number of pixels horizontally and vertically. This is the only "size" that is inherent in any bitmapped digital image.
It can be useful, however, to assign a physical size to a digital image. Suppose that you scan a 4x6 inch postcard. If the resulting image is assigned the size "4x6 inches" then people viewing the image will know the actual size of the postcard. It also means that when you print the image at "normal" size it will print at this size (at least with all the software I have seen). All digital image files include this size information with an image. Scanned images are assigned a physical size that corresponds to the actual size of the area that is scanned - which makes perfect sense.
Then there's the matter of resolution, which is usually expressed in terms of pixels per inch (ppi, sometimes referred to as dots per inch or DPI). Perhaps you have already realized that once a digital image is assigned a size it automatically has a resolution. Here's the formula for horizontal resolution:
horizontal pixels per inch = (number of horizontal pixels) / (horizontal size in inches)
Note that the vertical resolution can be different from the horizontal resolution, but usually they are the same.
When you are scanning, here's how it works. There are two things under your control. One is the size of the area being scanned - a single stamp, a postcard, whatever. The other is the resolution that will be used for the scan. With some scanner software you set the resolution directly - 96 dpi or 150 dpi, for example. With other software you select the type of document being scanned (color document, color photograph, etc.) and the use you will make of it (printing, web page) and the software selects the best resolution for you. In any case, when you make the scan the resulting file has a pixel size that is determined by these two settings:
number of horizontal pixels = (horizontal pixels per inch) x (horizontal size in inches)
Thus, if you scan a 4x6 inch area at 150 ppi the resulting image will be 600 x 900 pixels in size (4 x 150 is 600, 6 x 150 is 900).
What about digital photographs, which unlike scanned images do not have an inherent physical size? These images are assigned an arbitrary resolution either by the camera or your software. For example, when I open an image from my digital camera in Photoshop it is assigned a resolution of 72dpi. Since the image is 2560x1920 pixels, this results in the image having a "physical size" of about 36 x 26 inches. This "size" is essentially meaningless, of course.
In your graphics program you can change any of these three image size parameters: physical size, pixel size, or resolution. Because they are all linked to one another, changing any one of them means that one or both of the other parameters must change as well. Let's look at an example. Start with the following image:
Size: 4 x 6 inches
Pixel size: 400 x 600
Resolution: 100 ppi (pixels per inch)
Suppose you want to change the size to 8x12 inches. In order to do so you must also do one of the following:
Your graphics program will do this automatically, but you must tell it which one to do - change the pixel size or change the resolution. Note that the process of changing an image's pixel size is called resampling. Please see the section Resampling an Image earlier in this document for more information. The specific instructions for this differ from one graphics program to another, but the concept is the same. Here is the Image Size dialog box in PhotoShop:
Resampling of the image is controlled by the Resample Image option. If this option is on:
If the Resample Image option is off, you cannot change the pixel size. Changing the width or height also changes the resolution, and vice versa.
In some areas of philately, watermarks are very important and can make the difference between a $2 stamp and a $2000 stamp. The ability to create digital scans of stamps that show the watermark will be useful to many philatelists. I have devised a technique that can be used to create such scans. It can be used only with a flat-bed scanner, and may not work well with the less easily seen watermarks.
You will need a flat-bottomed container with a clear glass bottom, preferably with a lid. Do not use plastic as watermark fluid may dissolve some plastics. The size of the container does not matter, although a smaller one will be more convenient and will require the use of less watermark fluid. As long as you can lay the stamp flat on the bottom, that is all that is required. I use a Petri dish, a round flat container used in biological laboratories (see photo). You also need a small piece of black material - I cut one from the black backing on a stockcard. Finally you need some watermark fluid. I use SuperSafe Philatelic Watermark Fluid. The photo shows, from left to right, the lid and the base of the Petri dish, the piece of black material, and a 5-1/2 inch long pair of tongs to provide scale.
Here's how the technique works:
1. Place the stamp, face up, on the bottom of the container.
2. Add enough watermark fluid to cover the stamp.
3. Place the piece of black material over the stamp, pressing down to be sure that there is no air between the material and the stamp.
4. Put the lid on the container and place the container on the bed of your scanner.
5. Scan as usual.
Here's a scan of a Germany Scott #95 made using this technique. This is the raw image which was scanned at 200 dpi in 24 bit color.
After a bit of manipulation - converting the image to grayscale and increasing the contrast - the pattern of the watermark is clearly visible, as shown here.
If you cannot find a suitable container for scanning your watermarks, I have heard from some other collectors that you can do without one. Simply put the stamp directly on the glass of the scanner, put a drop or two of fluid on the stamp, and then press the black background over the stamp, using a small weight to hold it in place. Of course you would want to do this away from the edges of the scanner glass, to prevent the possibility of fluid dripping down into the scanner mechanism.
Another approach to scanning watermarks was told to me by Mr. Jim Sorenson of Madison, WI. His steps are:
1. Cover the scanner bed with a sheet of clear plastic to protect against possible dripping watermark fluid. You want a fairly stiff plastic that will not wrinkle and degrade the image.
2. Dip the stamp in watermark fluid and remove any excess.
3. Put the stamp in a black plastic stock card with a clear plastic front flap. Use care to avoid air bubbles as much as possible. Try redipping the stamp if bubbles are present.
4. Place the stock card face-down on the scanner bed.
5. Place another sheet of black plastic over the entire assembly to suppress reflected and scattered light, then close the scanner lid.
6. Scan at 200 dpi using "color photo" mode. This resolution would not be adequate for examining most stamp details but is acceptable for identifying watermarks.
Now you have your image and can apply various image processing techniques to clarify the watermark. For example, edge sharpening and smoothing may help make the watermark stand out better. Some sofftware (e.g., HP Image Zone) allows color filtering, which may be helpful for suppressing the stamp design and permitting clearer presentation of the watermark. You can convert the image to black and white if desired. The figure shows a scan of watermarks made with this technique.
It is necessary to calibrate your scanner if you want it to reproduce colors accurately. Calibration is not available with some low-end scanners, but if your scanner supports calibration I highly recommend that you take the time to perform it. It's a fairly simple process requiring very little time, and needs to be done every month or two. It involves scanning a target that contains known colors, then adjusting the scanner output so the information in the image accurately matches the target colors. You'll need to refer to your scanner's manual for detailed information.
As an alternative to calibrating your scanner you can include a reference patch, a small area of known color, in every scan. Anyone viewing your image can use a graphics program or their monitor adjustments to get the reference patch to display at its proper color, or RGB value. Then, the stamp colors will display accurately. A common reference color is 18% gray; you can buy a "Kodak 18% Gray Reference Card" at most photography shops. Of course, the person viewing your image must know what the reference patch is so that they can make the proper adjustments. You can label it, as shown in this figure, so there is no room for doubt.
In addition to including the reference patch in your images, you can use it to adjust your scanned images without actually leaving it in the image. Here's how:
1. Scan the stamp and the reference patch together, using true color mode.
2. In your graphics program, examine the color value of the pixels in the patch. Typically there is an "eyedropper" tool that lets you do this. The pixels in a 18% gray patch should have RGB values very close to 128, 128, 128. If they do, no adjustments are needed and you can go to step 5. If the RGB values are the same but not close to 128 (for example, 110, 110, 110) you can go directly to step 4.
3. Use your graphics program's color adjustment tool to change the color of the entire image until the RGB values of the gray patch are the same. Make changes to move the values toward 128. For example, if RGB was 115, 98, 102 you would adjust the color until the RGB values in the patch are 115, 115, 115. If on the other hand your values were 145, 135, 152 you would change them to 135, 135, 135.
4. Use your graphics program's brightness adjustment tool to change the brightness so that the pixels in the patch have RGB values of 128, 128, 128.
5. Crop the image to include only the stamp, and save it.
When a scanner bulb turns on, it warms up from room temperature to its operating temperature. When it turns off, it slowly cools down. The color balance of the light is dependent to some extent on the temperature of the bulb, and changes in color balance are more pronounced with less expensive scanners. What this means to you is that the color balance of your scans can change as result of changes in bulb temperature. The first image you scan during a long session may look a lot different from the 100th! depending on your scanner you may not find this a problem, but here are some tips.
Scanner bulbs also change slowly with age. You can compensate for age changes by calibrating your scanner on a regular basis, as described elsewhere in this article.
Accurate color in stamp scans is always difficult. In other words, the image of the stamp on-screen will have colors that are not quite the same as looking at the actual stamp. In most cases this is not a big deal--it is obvious what the stamp is despite the color being a bit off. In other cases, particularly with older issues, a minor color variation may be the difference between a rare and valuable variety and its common cousin. This is due primarily to the fundamental difference between reflected and transmitted light. When looking at an actual stamp you are viewing ink pigments laid down on paper by reflected light. On a monitor, you are viewing the colors of phosphors via transmitted light. It's also true that monitors, even specialized expensive ones, are not able to display all the colors found in the real world. This problem bedevils digital photographers who sometimes spend thousands of dollars on high-end monitors and calibration systems in order to get a reasonably close match. So, I do not have a solution to the color issue but I wanted to mention this so that you don't think you are doing something wrong when screen and stamp colors do not match.
Have you ever been faced with a used stamp where you cannot quite make out the details of the postmark? Postal history collectors, who are interested in place names and dates, sometimes encounter this problem. If the postmark is smudged there's not much you can do, but if the problem is that the postmark "blends in" with the stamp design the following two techniques may enable you to make out more details. It can be used with stamps off-paper as well as stamps still on cover.
The first procedure involves scanning the postmarked stamp and an unpostmarked copy of the same stamp, then subtracting the two images. In theory the stamp design will cancel out leaving only the image of the postmark. In actuality the technique is difficult to use and requires a steady hand, some experimentation, and lots of practice. Here's what to do:
This technique is based on the two stamps being the exactly the same, except for the postmark. This is often not the case, unfortunately. For example, a used stamp may have changed size slightly with respect to a mint stamp as a result of being glued to an envelope and then soaked off. Also, colors may fade which prevents the subtraction from completely canceling the two stamp images. This technique does not always succeed, and you should not expect perfect results the first time you try it.
The second technique was told to me by Mr. Robert W. Hisey. It works best used when the postmark is black and the stamp design has little or no black in it. It involves using your graphics program to separate the image into its 4 color components: cyan (blue), magenta (red), yellow, and black. It is described in detail in the following section.
Sometimes there is not a problem reading a postmark, but rather you want to "lift" the postmark to make a clean copy of it - in other words, an image of the postmark without the stamp. This is sometimes needed for overprints, surcharges, and the like as well as for postmarks. The technique described here can also be used in some situations for reading a difficult postmark. This technique works best if the postmark or overprint is black and the rest of the stamp is colored with no black elements in the design.
1. Scan the stamp in true color mode. Here's an example of a Canada Scott #51 with a great CDS.
2. Use your graphics program's channel splitting command (or equivalent) to create a CMYK split. This creates four new grayscale images, each containing the color information for one of the four colors of ink (Cyan, Magenta, Yellow, and blacK) that would be used to print your original image. You need only the black channel, which contains a negative image of the black parts of the original scan:
3. Invert the image colors, using the Negative Image or Invert Colors command to replace black with white and vice versa:
4. Crop and rotate the image as desired for the final product: In this example I have also done some retouching to "fill in" missing parts of the outer ring.
Scanning techniques can be useful in some situations to detect forgeries and printing varieties. The principal behind this technique is that the printed design on the forgery or variety will differ from the design on a normal stamp. By scanning both stamps and subtracting one image from the other, any differences should stand out. The technique is essentially the same as described for reading difficult postmarks.
A common need for scanned imaged of stamps is for on-line auction listings. In my experience, selling many hundreds of lots at on-line auctions, a good image is an important factor in whether or not a lot sells. In addition to the general principles of scanning that have already been discussed, here are a few pointers for creating images for auctions:
Here are a few things you probably know already, more in the line of ethical guidelines.
Some people have expressed confusion about the process of uploading an image as part of the auction listing. At Ebay, you have the choice of uploading the image to the auction site's computer or hosting it on your own site. The former technique is handy but there is a small fee if you want to use more than one image. If you do not want to upload your image to Ebay, you must upload the image to your own web site, and then include the address, or URL, of the picture in the auction listing. For example, if you created a scan called stamplot1.jpg you would first upload it to your web site (www.joeblow.com, for example). Then the URL you would enter in the auction listing is http://www.joeblow.com/stamplot1.jpg.
Some people may be wondering whether a digital camera can serve the same purpose as a scanner. For the most part, the answer to that question is "no." Consumer-oriented digital cameras are designed to take pictures of people, scenery, buildings, and the like - large subjects that are relatively far away from the camera. With older digital cameras the closest you can get to your subject is 2-3 feet, and the area in the photograph will be roughly the size of a newspaper page. Here's an example of a photograph taken with an Olympus DL-300 digital camera showing some stockbook pages I was putting up for auction. This photo was made at the camera's closest focusing distance, and clearly, individual stamps in this image are too small to be extracted as individual images.
Images such as this can be useful, however. The lot in this photo was one of several large lots that I had been trying to sell on the on-line auction services with no success. As soon as I included an image with each auction listing the lots sold quickly. It just goes to show that people like to see what they're buying even though images like this are essentially useless for evaluating individual stamps
Newer cameras are a different story. Most of them offer close focusing, very close in some cases (a few inches) and can be use to zoom in on individual stamps. Here's an example taken with a Sony F707 camera and then cropped:
It's more difficult to get good closeup images with a camera, so I recommend using a scanner if at all possible. Given the low price of many scanners this should be possible for most collectors.
Return to top
Main stamp page