The range of light

In this page, I want to consider typical images that have a range of light that may be either “normal”, more than “normal”, or less than “normal”. This begs the question as to what might constitute a “normal” range of light. The answer is given by the full range that we might expect from an image whose elements are all diffuse surfaces; that is, we’ll see all Zones from 0 to the lower end of VIII, aka nearly 0% to 100%. Most of the image will comprise elements varying from Zone I to Zone VIII, and these elements will either print easily or render as digital images easily.

Full range subject

Here is a simple example of a studio shot with a mother-to-be posing with her oldest child. The image was shot using studio lighting setup for manual power values. The camera was set in aperture priority at f/6.3. The image here is a screen shot from Capture One Pro v7 with its highlight and shadow warnings turned on. These are showing blue areas for pixels less than “1” and red areas for pixels hotter than “253”. Clearly, the full span of pixel values are being used, even if the darkest and hottest pixels are a small proportion of the overall image. The child subject’s face falls between the upper end of Zone V and Zone VI. The whites in her dress are in Zone VII, with its highlights in Zone VIII. The mother’s dress is mainly in Zone II with some of the deeper shadows in Zone I. Some elements of the pattern in the child’s dress are at Zone 0.

Mother and child

Mother and child

After some minimal further enhancement, here is a subsequent screen shot without the read-outs everywhere. I brightened the child’s hair, removed one of the flash reflections in her eyes, and added some sharpening to her eyes, nose, and mouth. I also cropped in a little. Outside of the specular highlight of the flash in the girl’s eyes (really from the thin layer of water that covers the eyes), all the surfaces are diffuse. Blowing out that small specular spot in each eye is actually an enhancement to the overall visual appeal.

Mother and child

Mother and child

High contrast subject

Now, let’s turn our attention to an image capture taken in lighting that exceeded the “normal” range, and which therefore required compression in order to manage the range of light. The following image was captured at base ISO on my D700 at Castlewood Canyon State Park. What you are seeing here is a screen capture in Capture One with a set of read-outs placed over various elements in the image. I exposed this image for the bright clouds even though that meant suppressing true 18% diffuse Value V by about 2 stops. The image was captured at a location near Denver CO on May 20, 2012 at around 5:30pm looking almost due west, so the sun is low in the horizon and the clouds are extremely brightly lit. To deal with this image, I also used a Singh-Ray Vari-N-Duo combination polarizer and neutral density filter in order to achieve the capture at all. This lead to vignetting, as you may see.

Castlewood image capture

Castlewood image capture

To correct this image, I applied the following curve layer:

Curve

Curve

With this layer in place, and some judicious correction of the lens flair in the original, I achieved this result:

Corrected image

Corrected image

In effect, the higher Zones are compressed, relatively speaking, because the mid-tones have been expanded into them. The curve was designed to bring Zone IV to a Zone V exposure. This isn’t a complete correction for the displacement of Value V, but it yields, I think, an appropriate result. Following are some 100% views of the image:

100% close up

100% close up

100% close up

100% close up

This digital approach to tonal range compression is highly powerful. It is not the only tool available for tonal range compression. Many RAW processors, like Capture One, Lightroom, Adobe Camera Raw, Capture NX2, Aperture, DxO, and so on, have additional tools for shadow and highlight recovery that can also play a powerful role in such adjustment. The fundamental beauty of a curves layer is that you can clearly identify the center of one or another Zone value and push that onto the value of another Zone. Together with adjustments to levels, curves are the central tool for the expansion and compression of the range of light in an image capture.

Here is another full range image. It was shot with an 85mm manual focus lens at f/1.4, yielding strong bokeh in the highlights of the sun behind the foliage in the background. It was captured in early October 2010 looking due south at about 3:30pm here in Colorado. In this case, I exposed for the bright highlights in the autumn sky behind the subject. While her face should have been in Zone VI, it fell on Zone IV instead. Applying a similar curves layer as in the previous image still left her face in the lower part of Zone V. Rather that further push the image globally, I added a layer in Capture One with a mask of varying density over the subject. It is roughly 50% everywhere and close to 100% on her face. I then added about 1 stop extra of exposure to this layer and a touch of brightness. This brought her skin tone to about the right level without making, I hope, the effect too artificial in terms of the rest of the image.

Full-range image, II

Full-range image, II

This raises a couple of practical points. I am fully aware that it is not always possible to go around metering off of gray cards. However, it is relatively straightforward to capture an image with matrix metering and then apply sufficient exposure compensation to prevent highlights from blowing out; or some similar combination of two captures. It is this difference in exposures that is of value, in terms of stops. If you take the first exposure as a reference or proxy for metering from an 18% diffuse gray card, then the second, practical exposure will give you useful information about how many Zones (stops) Value V was suppressed in your image captures. Likewise, if you have a model, getting a well-exposed image or two at the start of a session will give an effective Value reference for their skin tones in every subsequent capture. They may not be an 18% diffuse surface, perhaps they are a 36% diffuse surface, or whatever. Almost any suitable surface can be used as a tonal reference marker, especially if you have taken at least one shot of that surface with proper exposure during a session. Failing this, you can at least adopt a general guideline of what tonal values can be expected for skin tones, foliage, wood, stone, or whatever happens to be in your image.

Short-scale subject

Let’s now take an image that has a less than “normal” range, sometimes called a short-scale subject. In an image of this class, the range of light will be less than 8 stops. This gives the photographer a good deal of latitude in making the exposure because there is more freedom to place Value V on Zones up or down from Zone V without concern about blowing highlights or blocking shadows in a destructive manner. There are many good examples an extremely short-scale subject in the previous page on Zone exposures. That white wall had no more than a single Zone of variation in luminance. Of course, it was also not the most interesting visual subject either; but it provided several stops of latitude in exposure that could be adjusted in post processing to achieve a variety of effects.

Here is an example of a short-scale subject. I took this image because the few leaves remaining had interesting fall color. As a B&W, with a simple conversion as shown here, it is a dull and muddy image. The histogram shows only a range from about Zone III to Zone VI.

Short-scale image

Short-scale image

One thing to do with this image as a B&W conversion is to throw it away. Here is the same image with its tonal range expanded to occupy all 8 Zones. This was done with a simple levels adjustment. Besides bringing in the end points, the mid-point (Zone V) was also adjusted. A filter to emphasize the reds and de-emphasize the greens was applied in order to improve the tonal contrast between the red and green leaves. I added clarity and structure enhancements globally, and I selectively sharpened a twisted pattern around the dark branches and the brighter leaves. I added some contrast as well under the same mask.

Expanded short-scale subject

Expanded short-scale subject

Now, I’m not running off to print this and hang a copy on my wall; but I trust that the difference before and after is apparent. The most significant adjustment is with levels:

Levels change

Levels change

The end points have been brought inwards to drive the intensity of shadows and highlights up. Also, the mid-point (Zone V) has been adjusted downwards (to Zone IV), generally darkening the mid-tones. This eliminates some of the “muddy” impression of the original. You can think of this 3-point levels move as a mapping from the original Zones to new Zones. It both expands and shifts the Zones of the original exposure to those you see in the resulting image. The extreme shift of the shadow end-point, into the histogram, blocked up some of the blacks in the branches. I am OK with that, since I am less interested in any texture in their surfaces than I am in the detail of their overall shapes.

Another change, to drive up visual contrast, was achieved with selective color filtering:

Color filter

Color filter

This filter darkened the green leaves and brightened the red leaves, separating their tones in the resulting image. This also provides some interesting enhancement of the detail and texture within the leaves themselves. This was further enhanced with selective sharpening, clarity, and structure (mid-tone contrast) filters under a mask to select just the patterns of the main branches and leaves.

Expansion and compression

As shown in the previous examples, it is possible to use curves and levels, among other global adjustments, to either expand or contract the range of Zones in a capture. Expansion is perhaps a more intellectually easy concept to grasp: one can simply shift the end points of an image with a levels adjustment to put, say, Zone II onto Zone I or Zone VI onto Zone VIII. One can also shift Zone V up or down as desired my moving the mid-point of the levels adjustment.

Tonal compression may be somewhat more difficult to grok. It begins at the time the exposure is made. The exposure is set to place something like Value X or XI onto Zone VIII; for example. This makes Value V fall onto Zone III or even II, depending on the scale of the offset in stops. Later, in post processing, we raise Zone III (or II) back to Zone V globally or selectively (or both) with a curves layer, exposure enhancements, dodging, or whatever other tools we might use to achieve our goals. The net effect of this treatment is the same as if we had been able to place Value V on Zone V originally and somehow push Value X or XI onto Zone VIII. Either way we do it, we wind up with Value V from the original scene on Zone V in our image and Value X on Zone VIII as well. We have clearly compressed the range of light in the scene into something suitable for the printer or the monitor.

In one way, I am not offering advice that people haven’t been applying already with the typical “expose to the right” concept, which is to say, don’t blow out your highlights. What I am pointing out though is that this exposure technique will invariably suppress true Value V in the capture, and this will likely require some correction in post processing to recover the proper position of mid-tones. The Zone System model tells exactly what Values in the original capture need to be shifted to what other Values in post.

Knowledge of your own camera’s behavior, taken together with that of your RAW processor, provides you with all the information that you need to make these tonal compression or expansion adjustments and still get your primary reference Value right (18% gray from the scene mapped to 18% gray in the print).

A personal beef

I am getting on a soap-box here. I would argue that Ansel Adams was not “afraid of the dark”. His images, go look at them again, are characterized by strong elements at Zones I and II. At present, there is a view that prints should retain “detail” in shadows and in highlights. This is often interpreted, naively I think, that prints should retain “texture” in shadows and highlights. The distinction is worth two stops in complete tonal range in the final print. If you strive to retain texture at the end-points of your print, you can only occupy the range from Zones II to Zone VII. If you are not afraid of losing texture while retaining detail, your print can span the full range from Zones I to VIII.

My observation is that many photographers are “afraid of the dark”. While they will print out to Zone VIII (say out to about 250 or so on an 8-bit binary scale), they avoid Zone I like the plague. Zone 0 is considered like the “Black Death”, as it were. This is in spite of the fact that Adams’ prints frequently show large expanses of Zones I and II. As well, consider the depth of data at Zone 0 in a modern 16-bit image file. Zone 0 is 5 stops down from Zone V, by definition. As a range in binary number terms, this is equivalent to dropping 5 bits of resolution. In a 14-bit RAW file, the maximum value is 16,383. Value V occurs at 18% of this; namely, 2,949 decimal. The binary representation of this value requires 12 bits. Let us now go down to Zone 0; we still have 7 bits of resolution in the RAW file at this point. We are nowhere close to the noise floor. In fact, we would have to be getting down to Zone -5 (“-V” seems odd) before that is a problem at base ISO. In a DSLR like the D700, with base ISO at 200, this is equivalent to shifting the ISO up to 6400: then you see noise in the shadows in an 8-bit rendering because the noise has been expanded by 5 stops.

In fact, with 7-bit resolution, we have almost as much data down in Zone 0 as a JPG has in its entire 8-bit span. Think about this. In a well-exposed digital capture with a pro-grade DSLR, you have as much information in Zone I and below as a JPG shooter has in their entire image. We do not access this information under normal circumstances because we suppress it when we render the image as a JPG for delivery on the Internet (as on this page) or to go to an 8-bit print.

My point is, do not be afraid of the dark. Zone I is your friend. Touch it, pet it. Take it for coffee. Get to know it. It’s a gateway to Zone 0.

Previous Page … ••• Next Page …