In Part II, we described how to scan a black-and- white print into Photoshop and obtain a screen image visually faithful to the original print. In this article, we’ll describe how to adjust the printer so the inkjet copies it produces are visually faithful to the screen image. This means, for example, that if no manipulation is done in Photoshop, the resulting print is a tonal clone (a virtual density-for-density match) of the scanned print.
 |
This match is illustrated in Figure 1, where we scanned a silver print original into Photoshop and saved the result as a .tif file for ease of publication. This file is reproduced on the left in Figure 1 (Original). We directed that .tif file to our inkjet printer to produce a hard copy of the original. We scanned this inkjet copy and saved the result as a new .tif file, which is reproduced on the right in Figure 1 (Copy Print). The two images are clearly very close tonal matches. Nothing was altered or manipulated. There were no rescans or reprints, but the workflow scrupulously followed what was presented in Part I and Part II of this series.
The ideal tone curve
Part II contained a graph (Figure 7, page 12) of the optimum relation between Screen Brightness Value and density (or Log ED) in the print original. (Your graph may be different as its shape depends on your monitor, workspace lighting, etc.) A fixed set of scanner settings forced this graph to be obeyed when scanning any black-and-white print, guaranteeing the visually faithful screen image sought. If we make the same curve regulate the conversion of a screen image to an inkjet print, the print will look just like the screen image. In Figure 2, the graph in Part II is recreated in a slightly modified form. We have switched scales from Screen Brightness Values (SBV, 0 to 255) to % Black (%B, 0 to 100). The following equation shows the conversion between screen brightness value and % Black.
%B = 100 – (100 • SBV/255)
The two scales are equivalent, but the %B scale makes it easier to work with printer settings.
The goal is to set up the printer so that %B values translate to densities, as closely as possible as Figure 2’s Ideal Curve instructs. Achieving this requires that we have something easily quantifiable to print. We used Photoshop to create the test target illustrated in Figure 3. It is a 1.5 × 2-inch array of half-inch square patches representing %B values of 100, 95, 90, 80, 70, 60, 50, 40, 30, 20, 10, and 5 (the paper surrounding the array represents a %B value of 0).
To use the test target, make an inkjet print of the target, read the densities (or Log ED values) of the patches, and compare them to the values Figure 2 indicates they should be. You’ll find you have either a perfect match, or—much more likely—a detailed measure of how much distortion the printer introduces to the workflow. The remainder of this article describes the steps to remove all the distortions with any printer, ink and paper you use.
As you make repeated prints of this test target in the following experiments, several of them will fit on a single 81⁄2 × 11-inch sheet of paper. Avoid placing any very near the paper edges, since printers can be non-uniform in these areas. It also is important to clean printer heads frequently, since even small amounts of “banding” or other defects can greatly upset subsequent measurements. With most papers, it is sufficient to dry a target image briefly under a hair dryer set to “cool” before reading densities and reinserting the sheet in the printer to add another image.
Roles of the various printer settings
By way of example, we will reference our Epson Stylus Photo 820 Printer. Figure 4 shows the adjustments accessible in this printer’s Advanced Settings window when configured for 2880 dpi and using the driver originally supplied with the printer. The purpose of each of the seven highlighted settings are described below.
Media Type (1): Most paper manufacturers suggest specific choices for Media Type with different printers. The selection generally has only a small effect on the tone curve (the shape of the %B vs. Density graph). In the absence of manufacturer recommendations, you may want to print test targets using several different Media Type settings, selecting the one that comes closest to matching your personalized version of Figure 2.
 |
 |
Print Quality (2): The tone curve is distinctly different when printing at different resolutions. If we try to make a quick lower resolution print of a work in progress, the result is misleading since the density of a particular tone can change by several tenths of a density unit when we switch to a higher resolution. See Figure 5, which shows the response to changing the resolution when printing on Epson Photo Paper. Similar results occur with all papers. To avoid these problems, we always print at the highest resolution setting, 2880 dpi.
Flip Horizontal (3) and Edge Smoothing (4): Neither of these on-off switches has any impact on how tones are reproduced.
Gamma (5): Gamma can be set to 1.5, 1.8 or 2.2. Figure 6
illustrates the consequences of each choice when printing on Epson Photo Quality Glossy paper. Both the 1.5 and 1.8 settings produce curves that sag far below the “Ideal” curve. Prints made with these settings will be much too light throughout the entire density range. The best choice for Gamma is clearly 2.2. This setting produces a curve that wanders a bit on either side of our “Ideal,” but it puts us well on the way to the perfect match.
Depending on your printer, you may find a different optimum Gamma setting. The best choice also depends on the paper being tested. With any paper, the three curves always have the same relative appearance (Figure 6), but they may be displaced upwards with some papers. For example, when we print on Ilford Printasia paper, the curve from Gamma 2.2 lies above the Ideal curve, 1.5 is below, with 1.8 closest to overlapping the Ideal. The Gamma control, or its equivalent on your printer, deserves to be evaluated when testing any paper. Its influence is substantial and useful. Contrast this with the next two settings, whose impact we classify as substantial but useless.
 |
Brightness (6) and Contrast (7): These settings have dramatic impact on the tone curve, but they are rather coarse and not very well designed. Photographers coming from the wet darkroom would expect changes to Brightness to shift a tone curve left and right while Contrast would cause it to swivel about some common density. Not so! Adding Brightness does prompt a shift to the right. But decreasing brightness simply causes a big boost in upper scale (shadow) contrast. A change in the contrast setting does very little to any density below about 1.0, but does shift (what we term) the brightness of higher densities. These effects are illustrated with small (10 unit) adjustments to both controls in Figures 7 and 8. (The “Ideal” curve has been omitted from the graphs to improve visibility.)
It’s possible these responses are unique to our Epson 820 printer; if your printer is different, its comparable controls may be worth evaluating. For our part, we do not use these settings. Instead, we use a Photoshop feature—Transfer Function—to fine-tune the tone curve.
 |
Fine-tuning the tone curve
The Transfer Function is accessed from Photoshop’s Page Setup screen (File: Page: Setup: Transfer). This screen is seen in Figure 9. To illustrate its utility, reference Figure 10, which shows a portion of the tone control curve for Ilford Printasia paper. We are cheating a bit here. This is Printasia’s tone curve when printed at Gamma 2.2. As noted earlier, Gamma 1.8 is a better fit with this paper. We used 2.2 to illustrate the power of the Transfer Function to correct even huge mismatches between actual and ideal tone curves. Note this curve lies well above the “Ideal” curve, producing prints that are much too dark. This is what the Transfer Function repairs.
In Figure 10, note that a region of the screen image with a %B of 50 should produce a density of 0.75, according to the Ideal curve. However, when printed on Printasia paper, this %B produces a density of 1.02. If you follow the density 0.75 to the left, you will see it intersects the Printasia curve at a %B of 37. To make a %B of 50 print correctly, it must be “converted” to a %B of 37 wherever it appears in a screen image. This is what the Transfer Function does.
Go to the Transfer Function screen (Figure 9), enter an output value of 37 opposite the preprinted input of 50. Then, make similar measurements on the Printasia tone curve for all other input values on the Transfer Function screen and enter the values as output. The resulting curve looks like the one in Figure 9. If you make a second print of the test target with the Transfer Function on, it will be a near perfect match to the “Ideal” curve. It will be tonally faithful to the screen image, and, in the absence of any intermediate manipulation, a near perfect tonal match to the scanned print. Figure 11 compares the “Ideal” curve we want to match to the test target print made on Printasia paper with Figure 9’s Transfer Function.
 |
 |
In truth, the software algorithm that “smoothes” the transfer curve seems to be an approximation. Your first print may not be an exact match to the Ideal, and you may want to tweak one or two points to make it more nearly perfect. Focus on the lower densities when fine- tuning. The lower the overall density, the more visually obvious small discrepancies are.
You can name the Transfer Function, save it, and use it whenever you print an image with this particular printer, paper and ink. Barring any unit-to-unit variability in the ink or paper, and provided you maintain the rest of your system calibration as described in Parts I and II, you can expect future prints of any original image to print correctly first time and every time.
Will the inkjet dupe and the original be visually identical? They will certainly be a near perfect density match at all points. They may have a slightly different “look”—especially as a function of ambient lighting—but that will be a function of the inkjet paper’s surface and tint, how neutral the black ink is, etc.—things you can adjust with your choice of materials.
Are there other ways to fine-tune the tone curve? Yes and no. There may be occasions where one of the (previously discouraged) Brightness or Contrast adjustments will bring the raw tone curve closer to what you’re seeking. It’s tempting to use these adjustments, then add the Transfer Function to fine-tune. However, we found this worked poorly on our 820 printer. Apparently, the 820 applies the Transfer Function first and the Brightness and Contrast corrections second, a sequence that leads to unexpected results. The only way to arrive at a Transfer Function that works in this circumstance is by trial-and-error. A second alternative to a simple Transfer Function is a Curves Adjustment Layer in Photoshop built from the same numbers as the Transfer Function. However, such a layer distorts the appearance of the image on the monitor, whereas Transfer Functions remain invisible until the image is printed.
Other papers, other printers
We have made prints on 16 different inkjet papers, using the paper manufacturers’ recommended settings. Otherwise, we let the printer run at its default settings. Results range from not-so-bad to grotesque. By printing the test target (Figure 3), we determined the best choices of Media Type and Gamma (while holding resolution constant). This brought the previously grotesque into the realm of the not so bad.
Examples at this stage for three papers are illustrated in Figure 12. These charts also contain curves for a second printer—an ancient but still useful Epson Stylus Color 600 unit—to emphasize that the printer is also a huge tonal determinant. Some printer-paper combinations are a reasonably good fit to our aim Ideal curve, but most combinations print some densities too light or too dark, sometimes both within a single print. The important point is that all these combinations can be made to print spot-on with a well-designed Transfer function.
Conclusion
While arriving at these optimum conditions—Resolution, Media Type, Gamma, Transfer Function—takes perhaps half a day of experimenting, it is something that only needs to be done once for any combination of printer, paper and ink. Coupled with the suggestions in Parts I and II, digital black-and-white begins to look pretty good. Scan a print and what you see on the screen looks like the original. Print it and the print looks like what was on the screen. First scan. First print. Every time. Maybe there’s hope for this new technology after all!
Dick Dickerson and Silvia Zawadzki are retired Kodak black-and-white product builders. Their PT articles include “The Genesis of Xtol” (Sept/Oct ’99), and “Testing Kodak’s (Remanufactured) Black-and-White Films” (March/April ’03). Through their company, Dick & Silvia’s B/W Toolbox, Inc., they offer consulting and teaching services in black-and-white photography.
