BEYOND CALIBRATION 2.0
The Missing Link
By Christopher Schneiter
In my last article, Beyond Calibration, I talked about the problem of having an image look fine on a monitor, but then printing very dark. This is due to a lack of shadow contrast. My suggestion then was to put a lightening curve onto the image to lighten and increase contrast in the shadows.
If you remember this image:
My problem was that most of the tones in the image were very low key, causing the whole image to print dark. By applying this curve,
the mid-tones and shadows were lightened, allowing the image to print properly.
THE MISSING LINK
As usually happens, shortly after the last article was published, I ran across information that caused me to rethink this technique. While teaching a class on printing technique, the class completed a project creating threshold curves. As always, as I looked at the results from the threshold curves we were doing, I started making more connections.
The images that were created had a familiar look. They had air. You could see detail deep into the shadows, and highlights were right where they needed to be. They reminded me of the Zone System images that I grew up with. They had a "relaxed" look. It occurred to me that this might be what is missing in the conversation regarding "Digital Zone System."
(How it works with film)
With negative film to shoot and print precisely, we must first find the THRESHOLD values of both Paper and Film. This made me think back to conventional, film Zone System practice. For those who have worked with Zone System, you will remember that the FIRST task in the series of Zone System testing is to find the black exposure threshold, to determine a Standard Print Time. This is the minimum amount of exposure of a print that will give you a maximum black on the print. This is the black print threshold, where tonal gradation begins. Any more exposure of the print only serves to submerge shadow detail. Any less exposure, and there isn't a solid black. From setting this standard print time, you then adjust your film speed so that your darkest exposure gives you just a very slight difference from maximum black when printed at your standard print time. This will be the "threshold", or the beginning of tonality in your shadows, taking into account both your film and paper data. From there, you need to adjust your development time so that at your standard print time, and at your custom film speed, your highlights would print as they should, without manipulation. Each paper and film combination needs to be tested for Standard print time (Shadow Threshold), ISO, and development time.
The upshot of all this testing is that once you had determined your personal data for a specific film and paper combination, your images would print much more easily, with less manipulation, giving you a solid starting point in making your print. This would help you achieve precisely open shadows and controlled highlights, giving your prints an "airy" feeling, with your highlights where they should be and giving prints a "relaxed" look. This is why Ansel Adam's prints were so beautiful. They were planned in advance for specific film/paper combinations, so that he indeed was able to make a "performance" of his prints.
So, how does this apply to our digital imaging? On a monitor, we view our images in RGB mode, which involves light emanating from the viewing device. Prints are created in CMYK, which uses reflected light to view images. These are two very different animals, which is why images don't usually print exactly as you view them on your monitor. A monitor has a much longer tonal range, and not only is a print tonal range shorter, but it can also be affected by the print surface, and even the print profile and rendering intent. We must make a translation between RGB and CMYK. The first step in that translation is to precisely pinpoint our threshold densities.
WHAT IS THE PAPER THRESHOLD?
Threshold density is the precise point at which tonal gradation begins. In the case of shadow threshold, anything darker than this point will be pure black with no tonal differences. Anything lighter will show gradation, giving the print what I call "air", keeping the lower "Zones" from blocking up and running together. With highlight threshold, this is also the point where highlight tonality begins. Above this point, is lack of density or pure white. Below this point lies gradation. When carefully controlled, a print will have very delicate highlight detail, or what I call a "relaxed" look. All densities will fit into the given tonal range of the paper.
Just as with conventional prints from film, all papers respond differently to density, and have different threshold densities. The surface of the paper causes light to reflect differently. A gloss surface (Printed on an Epson 8 color pigment printer) will have a much lower threshold (about 20 RGB), giving more easily achieved shadow detail (higher contrast), while matte Papers, such as Velvet Fine Art diffuse the surface, causing it to have a much higher shadow threshold (about 30 RGB). This means that any density below these numbers will not show tonal gradation on the print. From this point downward, the paper will not print a blacker black.
Likewise, highlight density is very important. I have found that this threshold is less variable, with most papers having a threshold of about 253 RGB. Without controlled highlights, prints can have a "staccato" look, where highlights are just a little too hot, attracting too much attention, and detracting from the image.
For these reasons, it is important that a file needs to be "set up" for each specific printer/paper combination. Threshold densities need to be precisely set so that densities will fall where you want them. I routinely print on Epson Ultra Premium Matte paper with 7 and 8 color Epson printers. Through testing, I have found that with this paper, my shadow threshold is about 23 RGB, and my highlight is about 253. I'll go over how I arrived at these numbers below.
Let's take a minute and see what I'm talking about. The following image is a snapshot, actually shot with my IPhone one night while skiing. it's not an exciting image, but works very well to explain how important shadow threshold is.
This image is very typical of what your results might be with no correction. It's ok, but the info palette tells me that the density at point one (center right) has a density of about 12 RGB. This has been determined with the threshold chart (below) as too dark to reproduce, and will render as pure black. The density at point two is averaging 253 RGB, so needs no modification. I am getting a fully gradated highlight in that area.
Given that I know that there is more to be had in those dark tones, I then added this curve. The high point of the curve is where point one is located, and serves to raise that tone from 12 to 20 RGB. In making this curve, the entire curve was locked down with anchor points, and special care was paid to bringing the toe of the curve down to zero, to assure a good black. If the curve is raised too generally, it will affect areas that aren't desired to be changed.
Note that the curve is locked down, and only a very small area of the tonal range is affected.
The results in this case are spectacular. The delicate grays of the misty snow are raised to reproducible levels, giving the image an open airy look. Here is the result.
The only difference between these two images is that in the second version,
point one was lightened from 12-20 RGB, allowing reproduction of subtle tonal gradation in the shadows.
CROSS THE THRESHOLD
While it may be assumed that the threshold for gloss paper will be around 20 RGB, and soft matte papers around 30 RGB (and others in between), it will always be more accurate to test your materials. To do this, you have to print the actual densities to pinpoint your thresholds
Professor Ike Lea of Lansing Community College, in Lansing Michigan has devised a rather brilliant test chart, comprised of 70 density patches, numbered with their densities. In this sample, I have marked the patches that my tests on Epson Ultra Premium Matte showed as thresholds. In this case, the shadow threshold is 23 RGB, and the highlight threshold is 253 RGB. This test chart is printed exactly as you would print any other print, using the profile that you normally would use.
Printer Threshold Test Chart, Courtesy Professor Ike Lea.
©Ike Lea 2012
This chart was made of individual density patches and was fairly time consuming to make. Each patch was individually made to fit an 8x10 canvas. This required that 70 patches be made. To make your own patches, simply make a 1x1" canvas. Put a curve layer on it, and in the Input window, type in the desired density. This curve is set to give the canvas a density of 240 RGB. When this is done, simply flatten and drag it onto a waiting 8x10 canvas. Don't forget to label the patch with the density, so they don't get mixed up. After this is done, go back to the patch canvas, back up in history to the curve layer, and repeat 70 times.
This could be a tedious project, but remember, once you have your own chart built, it can be used to test every paper and printer made. it even occurs to me that it might be used for any digital output device, even in producing digital negatives for non-silver processes. The point is, that you only have to do this once, and once you have it, it has unlimited uses. Take your time and make it as accurate as you can. That being said, there's nothing that says that it has to be an 8x10 patch. It could be made in a more compact size. I have never seen a shadow threshold higher than about 30 RGB, so it is possible that your shadow patches could only go from about 15-35 RGB. Likewise, most papers have a highlight threshold of about 253 RGB, so it makes sense that if you gave a liberal range of down to about 240, your highlight threshold will be covered. After putting your chart together and labeled, you can flatten it and save it for future use.
To use the test chart, simply print it as you would any other print. Choose the paper and the proper profile. Since it is B&W, rendering intent could be that which you use normally. When it is printed, you will see that there will be a patch where beyond which, there will be no further change in density. In the case of Ultra Premium Matte paper with my printers, I was able to see that my shadow threshold was about 23 RGB, and my highlight threshold was 253 RGB. Once I had this knowledge, I could apply the threshold curve to accurately set those densities for my important highlights and shadows. It is important to note that these threshold densities are at the point where important tonal gradation begins. This is not necessarily the darkest or lightest point in the image. To have a full tonal range, you still need a good black, and in most cases a good white to keep the image from looking flat or muddy. It is up to you to decide where that point is. Determining this is subjective, and takes practice. If you are familiar with film based Zone System, you will have a huge leg up. Remember, the threshold density is not the darkest point in the image or the beginning of detail, it is the emergence of gradation only. This is the equivalent to finding an adjusted film speed with the Zone System, and in the digital arena, is the missing link.
WORKFLOW, THE FINAL FRONTIER
The issue that I've had with this concept in my classes is that students think of it as an exercise, as something optional. Unless the issue of print threshold is addressed, you will be forever plagued by prints that aren't what they could be. You will be subject to blocked shadows and highlights, and won't know why. The truth is, your paper and printer are only capable of a certain tonal range. Blacks won't print or show detail any darker, and whites will fall apart if you don't help them along. This is an issue of workflow, and is a part of making a good print. When you create a threshold curve, you are setting an image up to print with a specific printer on a specific paper.
ONE LAST, AND VERY IMPORTANT TWEAK
But it doesn't stop here. In addition to the importance of the limits of your ink and paper combination, there is another item we have to consider. As I've said throughout this article, what we see in an image on our monitor, we don't always see in the resulting print. Setting your print thresholds for an image, paper and ink combination is only part of the job.
We see images on a monitor as RGB . This means that it is a rear illuminated image made up of Red, Green and Blue. In this case, black is made by an absence of light. A print is viewed by reflected light and is made up of Cyan, Magenta, Yellow and Black ink. As we can see, a CMYK image can be affected by conditions such as print surface. In addition, when an RGB image becomes a CMYK image, the tonal range is compromised.
To make this conversion, we have to translate this image from RGB to CMYK. This makes the colors appear more correct on the print and approximates a proper tonal range. To do this, the translator we use is a profile. Many so- called "canned" profiles are pretty accurate, but sometimes even custom profiles need a little help.
A last tweak that seems to live in urban legends is the "Soft Proof." It is often misunderstood, and all too often seen as an extra step that a printer could try, or is only used when images are being printed remotely. The truth is that even custom profiles can only do so much, and soft proofing corrects minor density and saturation issues specific to an image/profile/ink/paper combination. Like observing threshold densities, soft proofing must be seen as a part of workflow. Unless you make it a part of your printing practice, you will be letting quality go.
FINISH YOUR IMAGE
In printing, the print profile causes visual changes in the image. If you barge along and depend on the output to tell you how it will "turn out", you are wasting time and money. Soft proofing is a tool that we have to see what those changes will be, and correct them before printing.
Often, those changes will darken areas of an image, and lower contrast, along with desaturating it. Every image is slightly different, which is why you can't rely on profile alone.
This image is a luminous scene. On a monitor, it is beautiful with glowing highlights and rich colors.
But when I soft proofed it, my heart sank. Soft proofing shows you the effects that the profile and the paper surface will have. When the soft proof was applied, this is what happened. This may look familiar to you. Not only was my color gamut limited, but contrast and saturation were muted. This is how it would have printed, without soft proofing.
As I said before a print profile compensates for the difference between RGB and CMYK, but every image is a little different. In this case, the profile and print surface cause the image to go muddy, so that needed to be remedied.
The first step to doing this is to identify the print profile. In Photoshop, go to VIEW > PROOF SETUP > CUSTOM. This will open a dialogue box called CUSTOMIZE PROOF CONDITION. In this box is a drop down menu that allows you to choose your desired print profile. In this case it is SP2200 Enhanced Matte_MK. Note that SIMULATE PAPER COLOR must be checked. This is what will help you see the affects of profile and paper clearly. Once you click "Ok", the soft proof of the profile will be applied to your image. It can be toggled on and off by using the keystroke "Command Y or Control Y".
We are now ready to do a soft proof. First, with the soft proof toggled off (Command/Control Y) duplicate your image. (Image >Duplicate) . The duplicate will be used to compare the original image to the proofed one. Click on the original, and apply the proof to it (Command/control Y). This will show you the effects of the profile/paper. Sometimes pretty scary! Note, the soft proof is not making any actual changes to the image, only simulating it.
We then need to evaluate what is actually happening to the image. In this case, the image went dark, especially in the highlights. So I put a curve layer on and used the targeting tool (the little hand in the curves palette) to show me where in the curve the significant highlight was (the puffy grass on the upper right). I put a point on that area (3/4tone) and lightened that to match the original (the copy) as closely as possible. However, when I set the highlight point, the contrast in the image dropped, so I added a low end point (1/4 tone) and darkened up the low end, giving it a classic "S" curve, increasing the overall contrast.
Since this image was largely comprised of out of gamut colors (unprintable) attention to rendering intent was important. Both Relative Colormetric and Perceptual rendering intents bring the gamut (color range) to within printable limits. Just as with tonal range, there are colors that cannot be printed. Relative Colormetric does this by chopping off the brighter colors to bring the gamut to the closest printable color. It does this at the expense of color subtleties that can show as texture, and in an image like this, left much of the image without subtle color contrast. Perceptual rendering intent compresses the entire gamut, leaving color variation relationships intact. In extreme situations, this can actually shift even mid range colors. However, in this image, it allowed the subtle color contrast relationships remain, giving more of a feeling of detail in the extreme ends of the gamut. It pays to try both rendering intents to see their effect. For me, in this case it paid off.
Another problem that happens when an image is printed, especially with matte papers, is that saturation drops. I have also begun adding a Hue and Saturation layer to my soft proofing curve. I label both the curve and the Hue/ Saturation layers and save them for future use. With just a touch of saturation, it is possible to almost entirely match the original on screen rendering of an image.
This is the final correction of this image. If you compare it to the original, you can see just how close it is.
© Christopher Schneiter June 2013
Although information intense, these two items (Yes! Only two!) are essential to your print quality, and absolutely should be steps in your workflow. For truly fine printing, you can't rely on automation or trust objective organs like profiles. Every image is just a little bit different, and you need to embrace those differences. All too often, I hear people say things like "It makes my images look worse!" If it does, that means that you don't quite have your workflow finessed yet. This stuff takes practice, and I've found that in writing my curves, sometimes even a correction of one or two points can make all the difference in the world. Work on it until it's right.
As the old saw goes, when the jazz musician was asked how to get to Carnegie Hall, his reply was "Practice man, Practice!"
Christopher Schneiter Adjunct Associate Professor Lansing Community College