Scanning 1, 2, 3
How To Get The Most Digital Image Quality From Film

sorcadmin's picture

Epson's TWAIN Pro for the Expression 1600 scanner is typical of flat-bed software with film scanning capabilities. It opens a preview scan and automatically adjusts for "exposure and contrast." You can revert to raw data simply by clicking the Reset button under the Adjust icons to the left of the preview window. The bottom image displays a scan made of unadjusted raw data. The Photoshop Histogram information window inset in the upper left of the image provides a graph of the information in the raw data file. You'll notice that this graph indicates the image information is contained in just a little over half the scan image space (gamut), with blank areas to the highlight and shadow sides of the histogram scale. In other words, there is no white or black in this image just a range of gray values.
Photos © David B. Brooks, 2000

The following step by step procedures and instructions are intended to assist you in obtaining the best possible film scans based on one basic principle: Every film image is unique. Therefore, the specific parameters of adjustment for every scan you make will be different. Although most contemporary scanners which support film scans offer various kinds of automated scan adjustment, all are based on the premise that every photographic subject image can be treated the same way. This limited approach is easy and convenient. It does not assure that the optimum adjustment for the highest fidelity of image reproduction based on the specific characteristics of the subject and the conditions under which it was photographed will be considered and incorporated into the scanner's image data processing. Even so, you do not have to give up all of the convenience of automation. You can use the following procedure steps to check what the automation has done and then make just the adjustments that are needed to accommodate the unique attributes of an image when necessary.

How A Scanner Works. The fundamental parts and functions of a scanner are really quite simple. The active imaging part is a triple line of CCD sensors. There is a light source on the opposite side of the film plane, as well as a lens between the film plane and the CCD linear array. The film passes by the CCD array on a carriage in a film scanner, or the light source/linear CCD and lens assembly move along the length of a flat-bed scanner, both transported by a finely threaded rod turned by a step motor. At each step the linear CCD array reads the value of the light passing through the film and focused upon it, recording one line of RGB pixel information. When the scan is completed its line-by-line gathering of information is compiled by the scanner firmware (code on a chip in the scanner) and transferred to the driver software application running on the host computer.

In the scanner driver interface the preview image of a scan is a low-resolution picture of what the scanner collects as raw RGB data from the CCD array. It is the automatic or manual software adjustments provided within the scanner driver software that interprets this raw data to result in a final scan that is transported to an application or saved as an image data file. So, essentially what the scanner does physically is always the same regardless of what kind of film and what characteristics are inherent to the film image. It is the driver software that defines how that raw RGB information should be interpreted. What follows is a step by step procedure that allows you to take control of how that interpretation is made.

Scan Adjustment Order. There is a very strict functional requirement to follow a specific order of progression making scan adjustments. The first thing to do is fit the image information portion of the raw scan data (gamut) into the output image space. Most contemporary scanners record data from the CCD array at 12-bit depth per RGB channel, or 36-bit color. This is a deeper bit depth than what is supported by standard personal computer applications and devices, which is 8-bits per RGB channel or 24-bit color. There is an advantage in this depth difference between raw scan data and the depth of a standard computer RGB color image file.

The Epson TWAIN Pro scan software provides four Adjust icons, the first of which actuates automatic adjustment. The second opens an Image Controls dialog providing access to the eyedroppers to set highlight and shadow levels. These are accompanied by sliders that can be adjusted after the use of the eyedropper to clip more or less into the image gamut. If after the eyedroppers are used and the gamut is adjusted the overall image appears too light or dark the second slider from the top of the window can be adjusted. This performs a simple overall elliptical reshaping of the image characteristic curve. If the lighter image tones are at a good level of brightness but the shadows are too dark, then click on the third icon to open the Tone Correction dialog. This provides a curve tool with five points spaced at the ends and at three intermediate points on the curve line; these point nodes may be moved with the screen cursor to shape the curve up or down. The curve created as illustrated moves the lower portion of the curve in an upward ellipse. This effect is reflected in the difference between the preview of the image at the top and a resulting final scan at the bottom. A Histogram is included in the upper left of the image indicating the gamut of image information fully utilizes the entire image space. In other words, the equalization using the eyedroppers and sliders provided an accurate pre-scan adjustment resulting in a final image that will reproduce in a print with a full range of tones.

Many film images when scanned with only an adjustment to map the raw data image information to the full gamut of the 8-bit per RGB channel depth, reproduce the overall image brightness too light or dark. There may be a large part of the image information concentrated in just the shadows or highlights. Adjustments to the placement of the levels of tone to effect lightening or darkening of the whole or part of the image thins the information depth in one part of the image and compresses it another. But, when the 12-bit data is translated as part of the final scan to 8-bit data there is very limited or no loss of image information.

This is why it is so extremely important to make all of the image adjustments possible with the scanner driver based on the preview so the final scan will not require any post-scan adjustment, which would be at the expense of lost image data.

Scan Adjustment Step No. 1: Fitting the image gamut of a raw scan into the 8-bit per RGB channel space is sometimes referred to as "equalization," a function that is referred to in Photoshop's Image/Adjust tools as Levels. The tools provided to accomplish equalization in scan driver software vary, but often include an eyedropper to identify and select the highlight and shadow level in an image. The procedure is to click on the eyedropper for highlights and then move the screen cursor to the preview image to carefully place it in the brightest highlight. Then repeat the procedure selecting the shadow eyedropper moving the cursor to the darkest shadow in the image and click on it. With some scanner drivers a histogram graph will be provided accompanying the use of eyedroppers, and the selection points will be indicated by small triangles or arrows under the graph. Others, like Epson TWAIN Pro, set the level of a slider that also has a number value readout. In either case you can adjust the setting to include more of the raw data gamut if the eyedropper selection clips into it too far so you would lose shadow or highlight information in your final scan. On the other hand, a few images, like a portrait with a white background, may require clipping further into the highlights to render the white as brightly as desired.

Scan Adjustment Step No. 2: Once the image is optimized to fit into the 8-bit output space used by a computer application, the second adjustment that should be done before a final scan is initiated, is to adjust either the overall brightness if necessary, or adjust the relative brightness of one part (shadows) of the gamut relative to the whole. This is most effectively accomplished by a tool that reshapes the characteristic curve of the image. This is the function of the Curve image adjustment tool in Photoshop and is referred to in some scanner software as tone correction, tone adjustment, tone mapping, etc. Usually this tool provides a curve chart which characterizes the preview beginning with a straight line. An overall ellipse adjustment upward provides overall brightening, and downward darkening. Limiting redrawing the line to a curve along just part of the curve line corresponds to adjusting the brightness of just one part of the image, like just curving the bottom up to open the shadows while leaving the lighter image tones unchanged.

The use of an eyedropper, to identify and select the highlight and shadow levels to equalize input to output as a first step in scan adjustment, has a second function scanning color. Many color film images may have a color caste, an overall coloration not a part of the image information but due to tints in the film base, residual effects of processing, or the result of age deterioration. When the highlight is sampled and set as the highest density level for output it may have an rgb density value that is different for each channel. Using the eyedropper to set the highlight changes these values so they are all the same for each RGB channel, neutralizing the color caste. The same function is achieved setting the black with the eyedropper tool. For instance, if the preview read R-2, G-5, B-7, it will adjust them to R-0, G-0, B-0, rendering the black as neutral removing the coloration. With LaserSoft SilverFast Ai scanner drivers used here as an example, the eyedropper tools are accessed by clicking on a white and black triangle from the icon bar at the top of the main control dialog window, which I have outlined in red. The blue color caste in this slide preview scan at the top was the result of the photograph being exposed by very cold light in open shade from a clear blue sky. Just by using the triangle eyedropper cursor to set the highlight and then the shadow the effect of removing the blue color temperature caste is seen in the preview window after both highlight and shadow setting are made. In addition, using SilverFast, if the image has a tone value in the mid range that should be neutral gray, you can click on the eyedropper symbol in the center of the icon and use it to remove a mid-tone color caste.

Scan Adjustment Step No. 3. Some film images require for one reason or another an overall or global shift in value. This kind of global color correction can be used to correct for photographs made in lighting that varies from the standard color temperature balance of the film used. Other film based causes can include a processing abnormality or a severe color shift due to age and film dye deterioration. The global color adjustment tool is also essential when there is no easily identified highlight in an image that should be neutral, precluding the use of an eyedropper to remove a color caste. It should only be used after both equalization and any curve adjustment needed has been applied.

Scan Adjustment Step No. 4. Selective color correction was introduced to a broad audience with Photoshop Version 5, and is incorporated in the Image/Adjust/Hue & Saturation tool dialog. It has also been a part of professional-level scanning software for some time, and is now included in an increasing number of consumer film scanning packages. It is an extension of global color adjustment that supports selecting and limiting to just one segment of the color spectrum that can then be adjusted on the Hue, Saturation, and Lightness scale. The flexibility of HSL adjustment combined with being able to isolate the changes just to one segment of the color range in an image makes it more practically useful than even global color adjustment. The great majority of images that are scanned from film are usually of good quality, and of subjects with a more or less typical distribution of tones and colors. More often what is needed to make the scans of these majority images ideal digitally are minor adjustments of color once the basic parameters of equalization and curve tone adjustment are set. For instance a picture of a scene with a person in it may be well-balanced with a good overall color fidelity to the scene, but the person's complexion may look too pink and ruddy. Selective color adjustment allows first setting the dialog to only affect the warm reds, and then the hue can be shifted a little toward the yellow/green side, and the saturation can be reduced to make the complexion less flushed looking.

Scan Adjustment For Color Negatives. Scanning color negative films can be an advantage because the density range of the image is lower, fitting into the dynamic range of the scanner readily, but also a disadvantage because the software processing by the scan driver software must eliminate the "orange" dye base color of the film. Most contemporary scanner software offers a selection of "film terms" or reference values for the interpretation of different brands of color negative film. However, I've yet to use a scanner which included film terms for but a few of the different color negative films I've used and attempted to scan. However, most scanner software includes a film term that is generic, but one which usually requires significant levels of adjustment to provide a neutral interpretation of the image information free of the influence of the color of the film's dye base. The solution is to use whatever facilities of adjustment the scanner software offers to compensate for the dye base as independently of the image contained as possible. A good test image to accomplish this would be a subject that contains only blacks, whites, and grays, preferably a gray scale. Then once the color only adjustments are determined which neutralize the dye base, they should be saved so they can be called up as needed as a starting point for adjusting scans of color images on that particular brand of film.


The LaserSoft SilverFast Ai scanner driver software Global Colour Correction tool dialog window contains the near universal color wheel function. This is a circle representing the visible spectrum with the six primary/secondary colors arranged to be complementary across the wheel. A small cursor node is set in the center when the dialog is opened. Move the node with the mouse in a direction away from the hue of the shift you want to balance until the shift coloration disappears. With SilverFast this color wheel tool is augmented by sliders underneath, as well as an ability to apply a color balance shift applied to just one or two bands of the total image gamut which is divided into three segments: highlights, mid tones, and shadows.

 


This image was made on a Kodak ISO 1000 color negative film for which a specific or even a close film term was not available. SilverFast Ai provides a tricolor Histogram dialog as an extension of the Negative tool dialog, that allows setting the gamut for each of the RGB or CMY channels to perform an individually adjusted interpretation (which can be saved and reused readily). This approach, made easier by the software, is not usually an exact correction however, in part because the dye base varies relative to the image color and density content. So, it is most imperative that if possible the eyedropper highlight and shadow selection method is used as the first step in image adjustment to assure that all of the color caste created by the dye base is removed. Then go on to finish the scan adjustment as you would a transparency film scan.


This picture of a moss festooned oak in front of a meadow of lupin was photographed at dusk, well after sunset. The odd color caste from this light required a major color balance shift to remove the mixture of excess blue and cyan. The point at which this balance shift became satisfactory was identified visually in the preview scan image window when an unobscured section of the oak's trunk was turned to a neutral gray.

 


The selective color correction dialog in LaserSoft's most recent Version 5 of SilverFast Ai provides a six and a 12 level breakdown of the color spectrum so both a wide and a narrow range of color can be selected for adjustment. In addition, being a Photoshop plug-in this scanner driver has added the ability to mask just a part of the image in which adjustments are made using a Lasso selection tool. In this example the advantage was to select just the weathered, gray wood rail fence, which looked very blue because it was illuminated just from the sky immediately overhead. This allowed eliminating the blue in the fence without altering the blue in the distant forest hills in the background of the image.

Share | |