Making negatives for alternative processes requires for many of us to build a curve to fix tonal distribution. Many people are under the impression you need a lot of gear to make the correction curve. While a specialised piece of equipment like a densitometer is handy at times, it is by no means a necessity. Same goes for a scanner. While using a scanner is my preferred method for input I could also use a digital camera to photograph my correction curve step wedges and make the adjustments. If you've come this far I think it's safe to assume you've already chosen your standard exposure time and negative colour and are ready to proceed to the next step.
Scan in the Curve Building Step Wedge
|
The first thing is to digitise (photograph or preferrably scan) your processed 101-Step wedge (the one you have made with the proper blocking colour on the previous pages.) Turn off your scanning software's "Auto" settings. Notice the print on the left looks a little "muddy". You may end up with a slightly off-white for your white square and a not-so-black black square. Have no fear, this will be corrected momentarily. |
Setting the White and Black Points in Levels.
 |
Open Photoshop or your photo editing program's
"Levels" dialog. Notice how the black and white pointers are too
far off to the left and right -- both out of the image's histogram range.
This is why the scan looks too grey in the blacks and too off-white in the
whites.
Select the Black eye-dropper tool and click on the 101-Step wedge's 100 per cent square (the black one on the bottom left). Then select the White eye-dropper and do the same for the 0 per cent square (the white one in the upper left corner). |
 |
The results in the histogram should look similar to what you now
see on the left and below in Step 3. All the excess has been trimmed off
the histogram and it now shows zero (0) per cent as white and 100 per cent
as black.
Two things are accomplished with the step we just performed. The first is eliminating the "head room" that may be introduced by the scanner. Most modern scanners are decent machines that light up a print with a very bright light showing the full tonal range. Most prints just don't measure up to this scrutiny and you get a little bit of "play" on the ends. Moreover, the second thing that's happening is that the print may be using a certain emulsion that is not meant to achieve a deep-dark-black dMax (cyanotype, gum, VDB, platinum et al.) which is fine. But since our curve is fixing the "relative" tonal imbalances we want a scale that starts at zero and finishes at 100. Remember: if you don't change the blocking colour of the negative and the exposure time, the highlight and shadow in the print should remain constant. The curve is just going to help us fix the discrepancies between those two points. |
Measuring and Recording Tone Values
 |
For the purposes of building a curve the step
wedge on the left is what we need. White in the whites, and black in the
Blacks.
We are going to use Photoshop's "Info" palette to measure and record the tone in each square. This will allow us create a correction curve by entering the difference in values into the curve dialogue's "Input" and "Output" boxes. By entering the original value opposed against the printed values the curve dialogue will redistribute the tones in the print giving us the full range when applied to our original. This process is often referred to as "linearisation" or in America as linearization. ;^) The only problem is that you may have trouble reading the exact tones because of roughness in the paper, grain in the negative or anything that affects the emulsion otherwise. The solution is to average the samples. I find the best and fastest way to do this is to use one of Photoshop's blur filters. I usually prefer "Gaussian" Blur. Depending on the size of your image you may need to vary the pixel radius you set in this step. |
Reading Averaged Values.
 |
Now, finally, we can read the values and chart
our curve. Note: after applying the "Blur" effect you may need to
set the white point and black point in levels a second time if
they're not exactly on 0 and 100. e.g. Often they will read 1% and 99% Go
back to step two and repeat the procedure if necessary.
This is where the real grunt work begins in building by hand. You have to go back and forth between reading values in the step wedge (using Photoshop's "Info" greyscale palette) and recording them to your spreadsheet program (you can also just write these out on paper.) <--[n.b. This is how most of the world appears to me for the first hour the morning after my slo-pitch team plays on Wednesday nights. We are sponsored by a local English pub called the "The King's Head". Our team name "The Grateful Head" of course.] |
The sample values are written down in this order. The Original Value/Processed Printed Value...
(Actual curve values used are in white)
0/0; 1/1; 2/2; 3/3; 4/4; 5/6; 6/7; 7/10; 8/13; 9/16
10/22; 11/24; 12/25; 13/26; 14/27; 15/28; 16/30; 17/32; 18/34; 19/35
20/42; 21/42; 22/42; 23/45; 24/46; 25/48; 26/50*; 27/52; 28/53; 29/57
30/63; 31/64; 32/65; 33/66; 34/67; 35/69; 36/71; 37/72; 38/74; 39/75
40/79; 41/79; 42/81; 43/81; 44/81; 45/84; 46/84; 47/86; 48/86; 49/87
50/88; 51/88; 52/89; 53/91; 54/91; 55/91; 56/92; 57/92; 58/93; 59/93
60/93; 61/93; 62/94; 63/95; 64/95; 65/95; 66/95; 67/96; 68/96; 69/96
70/95; 71/95; 72/95; 73/96; 74/96; 75/96; 76/97; 77/97; 78/97; 79/97
80/97; 81/98;
90/99;
100/100
General Notes on the Data:
The curve built using the HSL RNP-Arrays with a standard print time should ideally start at 0/0 and 100/100. *At some point the "Original Value" will read 50% on the print out. I make a note of this using the asterisk (*). It's one of those "given" points and the first one I usually plot. After plotting the first point at 0/0 my data "straight-lined" until 4/4 which was my next anchor. No sense wasting anchor points on a straight line. Notice that once I hit 98 per cent in the printed value I stopped entering data. This is because Photoshop only allows 2 percentage units between anchors. So if I'm at 100/100 in the top-right my next nearest point can only be at 98 per cent. I have highlighted the actual values used in constructing the curve in white ink to make them easier to read. Notice most (13 of 16) of the values fall below the 50 per cent block. This is typically where most of the ink is laid down and hence needs the most control anchors to be shipped off to the shadows.
Entering the Data into a Curve.
Open a New File in Photoshop. It can be an existing image or a blank screen, you just need to have something open so you can access the Curves dialogue.
Open the Curves Dialogue.
Once you have opened the dialogue click the mouse somewhere near the middle of the curve adjustment line (also known as the "spline"). |
No we can see the "Input" and "Output" boxes. Let's go to our raw data and find the printed value that read 50 per cent or very near to it. The box was 26 per cent on the original 101-Stepwedge. So we set the Input to 50 per cent and the output to 26 per cent. |
Behold. Your curve begins to take form... |
Using a little finessing I enter the 16 data points for the curve (Photoshop only allows 16 anchors to be set). Remember to Save your curve as an ACV file. You can take a look at mine if you're interested. Download the sample ACV file here. Sample ACV File. |
Sometimes you get a nice smooth Bezier curve with no weird little dips and rises -- sometimes, but not very often in my experience. In practice the printer's driver and the interaction with substrates, ink choice and light can do some strange things on the step wedge of life. Have faith in your numbers though and you'll always be within reach of a decent if not perfect curve.
ChartThrob Generated Curve Using Sample Values |
Curve Generated by manually entering the sample values. |
The differences between ChartThrob and the handmade curve are subtle but fairly close overall.
There are advantages and disadvantages to creating the curve by hand as opposed to using an automated method like ChartThrob. ChartThrob is definitely much faster and if you have properly prepared data in your Chart it gives decent results. But there's an old computer guru saying "garbage in = garbage out". Doing the job by hand is much slower and gives you and appreciation of what the numbers actually mean and how they change the body of the curve. Hand curving also allows you to pick and choose your control anchors in case you need to work around an anomaly in the printer driver. ChartThrob might not always be as forgiving. Nothing wrong with getting a second opinion either -- I used ChartThrob to give me one on my work here.
-- Michael Koch-Schulte
26 June 2007.