Image Scanning
Histograms
Let’s make the tones better, to improve the contrast of the image.
The controls named Contrast and Brightness are of very little value. They are even detrimental perhaps. They discard data indiscriminately, in an unintelligent manner. More about this later, but suffice it to say that the Histogram tool is very much more versatile and useful.
An example: The Microtek scanner histogram tool
The Umax VistaScan 2.45 histogram tool below, shown to make the point that it does exactly the same thing, and works the same way. VistaScan 3.5 has the same Highlight, Shadow and Midpoint slider controls that do the same thing, but no longer has the data graph as a guide. (VistaScan 2.4x did, and MagicScan does.
The HP scanners generally do not provide this standard tool (however PrecisionScan Pro has it). It is not that some scanners do not do this, but instead that they do this contrast adjustment automatically in software, and manual control is not available to the user. My opinion is that the HP philosophy ranks ease of use above control and versatility. And that is best for many users too, like one button cameras. Most scanner brands do have an AUTO contrast button that adjusts the tones automatically in the same way. But images do vary, and many users prefer to have the tools for full control themselves. That’s my vote too, and this section is about how it works.
Some scanners may not show the histogram bar-chart data, but may still provide the same Shadow and Highlight sliders, described here as the Black Point and White Point, and they work the same. The data chart is a good guide, but we set these by eye anyway. Use the Preview image to watch that, and this discussion will help you understand what is happening. Any Gamma slider is the Mid Point described here. You can set the AUTO button for good results, and boost the midpoint to brighten, if necessary.
The better photo editor programs also have histogram tools, so this can also be done after the scan. If you have a histogram tool, whatever it is called, it will work the same as described here. This procedure is generally applicable to most (but not all) scanner brands, and some photo editors too. This is a common standard tool.
Whatever your Histogram tool is named, it will work as described here.
What’s a histogram?
The Histogram shows the total tonal distribution in the image. It’s a barchart of the count of pixels of every tone of gray that occurs in the image. It helps us analyze, and more importantly, correct the contrast of the image.
Gray? Yes, photographers know that color prints must first have the exposure and contrast set right, exactly the same as for B&W, and only then do you worry about the color balance.
Technically, the histogram maps Luminance, which is defined from the way the human eye perceives the brightness of different colors. For example, our eyes are most sensitive to green, we see green as being brighter than we see blue. Luminance weighs the effect of this to indicate the actual perceived brightness of the image pixels due to the color components. The world won’t end if you simply think of luminance as brightness, that’s actually quite fine for our purpose (and it’s really fun to watch the purists have a fit anyway <grin>). But luminance is weighted by color, and there is more detail and explanation about luminance in histograms if you’re still curious. For now, luminance can be the “apparent brightness” of the RGB pixel tones in the image.
Scanning B&W photos uses exactly the same procedure as described here (for GrayScale, but not for Line art). The histogram shows the tone values as gray anyway.
Every pixel in the Color or Gray image computes to a Luminance value between 0 and 255. The Histogram graphs the pixel count of every possible value of Luminance, or brightness if it helps to think of it that way. Luminance is brightness the same way the human eye sees it, as opposed to absolute brightness. Anyway, the total tonal range of a pixel’s 8 bit tone value is 0..255, where 0 is the blackest black at the left end, and 255 is the whitest white at the right end. The height of each vertical bar in the histogram simply shows how many image pixels have luminance value of 0, and how many pixels have luminance value 1, and 2, and 3, etc, all the way to 255 at the right end.
The histogram barchart shows at a glance the relative image tone distribution over the entire range. What it is now, and what it needs. In this image, we have a very high count of pixels that are near, but not at, the white end. We also have many that are near, but not at, the black end. Our image does not totally fill the possible range from darkest to lightest tones. Our image could have more contrast. But we can fix that very easily.
OK, let’s begin with the Simple Way technique now…
Umax VistaScan automatically starts each preview at defaults, but Microtek ScanWizard has manual Reset button because it remembers previous settings. There are pros and cons for either way (a configurable choice would be great!). There may be a reason to retain settings for several scans, but ordinarily when starting a new image preview, we want to eliminate all of the adjustments made to the previous image. The Reset button is in every ScanWizard tool, and is also at the bottom of the extended Settings Window (I’d like to see it on the toolbar with the Preview button). It resets all previous adjustments to defaults, for a fresh start for this image, the proverbial known starting place so we know where we are.
Reset, then do a Preview scan. The Preview scan makes the histogram data available. The histogram represents the area of the image that is marked in the Preview area to be scanned. The scanner software stores the histogram data for the entire preview area, but shows in the histogram only the data to represent the currently marked Preview area specified for the scan.
This is the histogram for this photo seen after the Preview scan. We can just about explain the three peaks in the histogram. There is the highest peak due to the light background towards the 255 end, meaning that many pixels are nearly white. Yellow is way up there too. Then the midrange red and green values, and the very dark green leaves near 0. The peaks mean that the pixel count with those tone values were high. The photo’s upper left corner is more white than the other corners, meaning those values are closest to 255. Place the mouse over a part of your ScanWizard graph, and it will show the histogram Gray value there and shows the total count of pixels with that value.
The tone values at the ends with zero pixel count are of the most interest. Note that for this image, the original histogram values don’t extend full range 0..255. 0 is the blackest black, and 255 is the whitest white, and we don’t have any of either, at either end. But we can expand what tones we do have for best effect, by “Setting the Points”. This is a very standard scanner technique, really THE standard technique, and is the reason that any better scanner software provides the Histogram tool. You can ignore it, but you and your scans would be missing out on a really great thing.
Histogram
- Moving the Black point right (inward) emphasizes shadows and makes the image darker. But you may not always want the shadows to be darker. The Black Point determines level 0. Therefore values to the left of the Black point are black too, and can contain no detail.
- Moving the White point left (inward) emphasizes highlights and makes the image lighter. Normally you want to move the White point. The White Point determines level 255. Therefore values to the right of the White point are white too, and can contain no detail.
- And you can do both in a balanced way to enhance the total range. This does not imply that you must always move both points, one may already be OK. The idea is that you see it all here, and then you know. You can experiment if you don’t see the best choice immediately.
The biggest mistake is to NOT experiment. Be more daring and extreme than you think right. You may not like it, you may put it back, but at least you will have seen it (and you may like it!).
Go to source: Summary – Histogram
Unsharp Mask
Radius controls how wide the edge rims become, and Radius = 1.0 is about the right ballpark, with 0.6 to 2.0 often being useful. Higher Radius values can cause halos at the edges, a detectable faint light rim around objects. Radius units are not the same as pixels, the units step in tenths, but the Radius width is usually at least 4 pixels overall, you will see various effects. Radius is a very important parameter, and the easiest way to ruin a good scan is with too much Radius. Inanimate objects can use the most radius, human faces can tolerate the least, and landscapes fall in between. But it really depends on the size of the details. Fine detail needs a smaller Radius, or else you may obliterate tiny detail of the same size as the Radius width. Large images have larger detail (more pixels involved) and can use more Radius, so therefore printing at higher resolution can support the larger radius. Radius and Amount interact, reducing one allows more of the other.
Threshold specifies how far apart adjacent tonal values have to be (values of 0..255) before the filter does anything to the edges, before it is judged to be an edge at all. This lack of action is important to prevent smooth areas from becoming speckled. Low values should sharpen more because fewer areas are excluded. Higher threshold values exclude areas of lower contrast. Human faces want values greater than 1 or 2, like perhaps 5 or more. For inanimate objects, perhaps 0 or 1 is useful. General work, try 3 or 4. This control has little effect at high values, but has more effect changing between low values of 0 to 5. This Threshold is not to be confused with Line art Threshold.
Amount is like a volume control, exaggerating the edge differences (how much darker and how much lighter the edge borders become). Amount interacts with Radius as to degree of sharpening, but it does not affect the width of the edge rims. Amount has a large effect, and values of 80 to 120 are normally usable if the Radius isn’t too large.
Go to source: Sharpening with Unsharp Mask
Best file types for these general purposes: Photographic Images Graphics, including Logos or Line art Properties Continuous tones, 24 bit color or 8 bit Gray, no text, few lines and edges Solid colors, up to 256 colors, with text or lines and sharp edges Best Quality for Archived Master TIF or PNG (no JPG artifacts) PNG or GIF or TIF (no JPG artifacts) Smallest File Size JPG with a higher Quality factor can be decent (JPG is questionable quality for archiving master copies) TIF LZW or GIF or PNG (graphics/logos usually permit reducing to 2 to 16 colors for smallest file size) Maximum Compatibility (PC, Mac, Unix) TIF or JPG (the simplest programs may not read TIF LZW) TIF without LZW or GIF Worst Choice 256 color GIF is very limited color, and is a larger file than 24 bit JPG JPG compression adds artifacts, smears text and lines and edges Go to source: Image file formats – TIF, JPG, PNG, GIF
A Few Printing Resolution Guidelines
For Printing Grayscale or Color Images
(photo images and graphics)
Useful scaled printing resolutions are shown below
for common printing destinations, and they indicate
the final size at the paper. This number is also the
scan resolution if printing at 100% or original size.
See Chapter 6 about scaling to print other sizes.
For photo-quality inkjet printers
240 to 300 ppi for color or grayscale, at printers
good setting on good photo paper.
300 ppi sometimes may be a little better than
240 ppi, but the difference is often hard to see,
and convenience usually makes the decision.
For reports on inkjet Premium paper
Use a decent printer quality setting for the text,
and photos up around 240 ppi.
For inkjet printer using plain copy paper
150 ppi is plenty for images on plain copy paper
For digital lab printing services
Wal-Mart, Costco, Ofoto, Shutterfly, Snapfish, etc.
Up to 300 ppi, for up through 8×10 inch size.
Inquire about pixel dimensions for poster size
prints, which are likely much less than 300 ppi.
For dye-sublimation printers
Ppi up near the dye-sub printers dpi rating,
which will be around 300 dpi
For laser printers, B&W or Color
300×300 dpi printer (54 lpi) 100 ppi
600×600 dpi printer (85 lpi) 150 ppi
1200×1200 dpi printer (100 lpi) 200 ppi
For commercial offset printing
(including publishing in books and magazines)
Image ppi between 1.5 and 2.0 (minimum and
maximum) multiplied times the lpi specification
for the screen process. Assuming 150 lpi, then
most editors will routinely ask for 300 ppi, scaled
to final size (300 ppi is 150 lpi x 2.0).
For printing in newspapers and newsprint
150 to 200 ppi (85 or 100 lpi) at final size.
For Printing Line Art Mode Images
(text pages, line drawings, sheet music)
For laser printers
300×300 dpi printer 300 ppi
600×600 dpi printer 600 ppi
1200×1200 dpi printer 1200 ppi
300 ppi line art is fine for casual purposes.
600 ppi is best for more critical work.
For inkjet printers
600 ppi for best but at least 300 ppi.
For Fax
200 ppi, see Chapter 11
For commercial offset printing
(including publishing in books or magazines)
800 to 1200 ppi commercially for line art
For printing in newspapers
600 ppi is enough, cartoons for example
For video screens and web pages
Forget about 72 or 96 ppi, that mythical concept
doesnt exist. However, scanning at 75 or 100 ppi
does create roughly original size on many screens,
if that is the image size you want. You can always
resample a large image smaller later for the screen.
For computer screens, including web pages
Regardless of mode, use the necessary scanning
resolution to create the desired image size in pixels
from the photo size, see Chapter 5. For example:
6×4 inches at 100 ppi gives 600×400 pixel size
For TV screens see page 54.
For PowerPoint screens see page 54.
Graphics for web images (for example logos or
screen shots of dialog boxes, as opposed to continuous
tone photo images) are often much better images
as GIF files instead of JPG files (higher quality and
smaller file than JPG), if reduced to 16 colors index
mode using an Adaptive palette with Nearest Color
(instead of allowing dithering). See Chapter 14.
Use the Unsharp Mask filter once with the following settings:
Image base/16 base/4 base base*4 base*16 Resolution 128×192 256×384 512×768 1024×1536 2kx3k Amount 100 100 100 100 100 Radius 0.25 0.5 1.0 2.0 4.0 Threshold 2 2 2 2 2Be prepared to undo and change the settings if the result isn’t what you want. Unsharp Mask finds edges between areas of color and increases the contrast along the edge. The Amount setting controls the strength of the filter. I usually stick to numbers between 60 percent and 140 percent. Radius controls how many pixels in from the color edge are affected (so you need higher numbers for bigger images). Threshold controls how different the colors on opposing sides of an edge have to be before the filter goes to work. You need a higher threshold for noisy images or ones with subtle color shifts that you want left unsharpened.
| | |