Many of the most dramatic landscape photographs are taken in high-contrast light, with sunrise or sunset light stabbing through the clouds and spotlighting a distant peak or sandstone tower, leaving the equally important foreground in deep shadow. Capturing a full range of tones in such high-contrast situations and reproducing them in a way our visual system considers natural has challenged landscape photographers ever since photography was invented nearly 200 years ago.
High-contrast scenes are difficult to photograph because of differences between our complex visual system and a camera’s sensor. Our eyes can see a huge range of light intensities from brightest highlight to darkest shadow. However, the range of brightness levels in a print is limited by the amount of light reflected by even the brightest white paper or canvas and by the amount of light absorbed by the blackest ink or paint available. The range of light intensities in the real world, in one scene, can be 10 times greater than what can actually be reproduced in a print. As you’ll see, the problem is compounded because our visual system does not analyze contrast globally, but rather within regions. A straightforward linear compression of the tonal scale we see into one we can print looks quite unnatural because it doesn’t correspond to the way our visual system processes high-contrast scenes.
This problem is not new. In the preceding chapter I introduced the Rembrandt Solution, my term for a photographic technique that employs the same principles as countershading, the technique that Rembrandt used in many of his paintings to create the illusion of greater dynamic range than actually existed. This chapter will teach you the principles of perception that underlie this technique. I’ll discuss how our visual system processes high-contrast scenes, which will give you insight into how to craft photographs that look both believable and beautiful. Next, I’ll go over the best metering techniques to use in the field when you’re planning to employ the Rembrandt Solution exposure strategy, which requires making two separate exposures of a scene—one exposed for highlights, one exposed for shadows. Finally, I’ll describe how you can use Lightroom and Photoshop to create the countershading effect with much greater flexibility and precision than is possible with physical graduated neutral-density filters.
The principles of perception that underlie the Rembrandt Solution are still applicable even if you use the fourth basic exposure strategy: high-dynamic-range imaging (HDR). I’ll discuss metering strategies when you’re using the HDR approach, as well as a hybrid technique for high-contrast scenes that combines the Rembrandt Solution metering strategy with HDR merging of component images. Finally, I’ll offer some suggestions on giving your HDR images a realistic look. Before we get to HDR techniques, however, let’s first dig deeply into the Rembrandt Solution.
Understanding countershading is fundamental to understanding the Rembrandt Solution. Countershading is the technique of introducing a gradual change in the background illumination, let’s say from light to dark, so that light foreground elements placed against the dark part of the background look brighter than they actually are. Countershading relies on two principles. The first is that our visual system is much more sensitive to abrupt changes in luminance than gradual ones. In other words, our eyes are programmed to look for edges of objects. Our visual system evolved this way because it was much faster, for example, to identify the outline of a lion hiding in tawny brush than it was to distinguish the texture of fur from the texture of grass. The second principle is that surrounding a tone with something darker makes the original tone seem lighter; surrounding the original tone with something lighter makes the original tone seem darker. Vision scientists call this effect simultaneous brightness contrast.
Figure 8-3 shows a simple example of countershading. When a background gradient is present, as in the left-hand image, you should see a subtle gradient inside the inner rectangle, from darker on top to lighter on the bottom. When the background is pure white, as in the right-hand image, you can see that the inner rectangle is actually a completely even tone. Simply by creating a tonal gradient in the background, we’ve induced an apparent, opposite tonal gradient in the foreground.
Now let’s take it further, to an example that shows how the Rembrandt Solution can create the illusion of greater dynamic range. Examine figure 8-4, which shows the Cornsweet illusion. The left half of the figure shows two rectangles, one above the other. The top rectangle should appear lighter than the bottom. In the right half of the figure, the middle two quadrants of the diagram are hidden. Suddenly you see that the top quadrant of the strip is actually the exact same shade of gray as the bottom quadrant.
Here’s what’s going on: The top half of the illusion contains a gradient from midtone to lighter-than-midtone. The bottom half contains a gradient from darker-than-midtone back up to midtone. Our eyes are insensitive to the gradual change of density in the gradients, but very sensitive to the abrupt change of density in the middle. Here’s the crucial point: merely by introducing two simple gradients, you can create the illusion of a greater dynamic range than actually exists. The top and bottom quadrants of the illusion are the same density, but they look different. The Cornsweet illusion shows why you can use a split ND filter with a gradual transition from dark to clear in a situation where the actual transition from highlight to shadow is abrupt and not only hide the fact that you’ve used such a filter, but actually enhance the apparent dynamic range of your print. The digital equivalent of using a split ND filter works exactly the same way.
Picture a typical split ND situation, with brightly lit mountains and deeply shadowed foreground flowers. You attach a split ND filter with a gradual transition zone from dark to clear and position the middle of the transition zone over the sharp dividing line between shadowed flowers and sunlit peaks. Let’s analyze the effect of the filter on the image from top to bottom. Figure 8-5 shows an example. The solid-gray part of the filter uniformly darkens the upper part of the peaks. As the filter’s transition from dark to clear begins, the sunlit peaks become brighter as the amount of light absorbed by the filter diminishes. At the shadow line, still beneath the transition zone of the filter, the shadow becomes darker than it otherwise would be because the filter’s transition zone hasn’t yet faded to clear. The bottom of the image is unaffected because it’s behind the clear portion of the filter. A print of the image will show the illusion of greater dynamic range than actually exists. As you’ll soon see, the digital version of the Rembrandt Solution achieves the same effect using slightly different means.
You might think that we analyze contrast globally; in other words, you might think that we look at the darkest part of the scene and call it black, then look at the brightest part of the scene and call it white—but that’s not actually how we see. In high-contrast situations, our visual system divides the scene into various zones and analyzes the local contrast in each zone independently. Shadows and highlights are the most obvious zones, but we also create zones more subtly. We then assign brightness values within zones, and don’t really pay much attention to brightness differences across zonal boundaries. For a scene to look natural, the local contrast must look right in each zone. When you use the Rembrandt Solution (either analog or digital), you can expose both the highlight and shadow regions somewhere close to midtone, which means both regions will have near-ideal contrast and color. Then, by using a Cornsweet-illusion-like pair of tonal gradients, you can marry the two regions in a way that our visual system finds believable.
For an illustration of how our visual system analyzes local contrast rather than global contrast, check out the optical illusion in figure 8-6. Examine the two thin rectangular regions I’ve outlined in red, one containing a strip of white paper in the shade, the other containing a portion of the second letter “l,” printed in black ink, in the sun. The white paper in the shade looks brighter than the black ink in the sun. How could it be otherwise? Isn’t white paper always brighter than black ink? Now look at the copies of the same two rectangles that I’ve placed to the right of the photograph of the newspaper. You’ll see that the white paper is actually much darker than the black ink. The two rectangles shown in isolation are exactly the same density as the two rectangles outlined in the photograph of the newspaper, but they look different because they are isolated against a plain white background rather than placed inside the context of an image with clearly defined highlight and shadow regions.
You might think it would be better to show all of the highlight region as a midtone and all of the shadow region as a midtone, without creating gradients that make the highlights lighter and the shadows darker in the region near the boundary between highlight and shadow. In fact, using such an approach leads to an image that looks patently fake, as you can see in figure 8-7. Creating a boundary between the two regions that has no gradients leads to an abrupt transition from cool shadow to warm highlight that has a color change but no density change. Your brain can be fooled, but it’s not stupid: the result looks awful. You must preserve a brightness difference across the boundary between highlight and shadow for the image to look natural. Use of a split ND filter with a gradual transition from dark to clear is one way to preserve that brightness difference while still placing most of the highlight and shadow regions close to midtone; use of the Photoshop procedure I’ll describe soon is another and better way.
Enough theory! Let’s take these ideas and put them to practical use. In chapter 7, I discussed how to use physical split NDs. Now let’s talk about creating the digital equivalent. To use the digital version of the Rembrandt Solution, you need to make two captures, one correctly exposed for the highlights, the other correctly exposed for the shadows.
For starters, be sure to lock your camera down on a solid tripod, and be careful not to move the tripod or camera when adjusting the exposure. When changing exposure, be sure to change the shutter speed, not the aperture. Changing the aperture will change the depth of field, which could prevent the images from aligning perfectly. Using the auto-bracketing feature on your camera makes the job much easier. In aperture-priority and manual-exposure modes, most if not all cameras will auto-bracket by holding the aperture constant and varying the shutter speed—exactly what you want. In shutter-priority or time-value mode, they will hold the shutter speed constant and vary the aperture—not what you want.
I strongly recommend setting the bracket order to metered, under, over. With this setting, the camera will first make the metered exposure (the one the camera thinks is correct), then make the darker frame, and finally the lighter frame. It may seem more logical to set the bracket order to under, metered, over, but that setting can lead to immense confusion. With many cameras, setting the bracket order to under, metered, over and pressing the shutter release halfway down to activate the meter will cause the exposure scale in the viewfinder to show you the exposure for the first frame in the bracket set (which will be the dark frame) rather than showing you the exposure for the metered frame. If you then readjust the exposure so the index mark on the exposure scale lines up with zero instead of some negative value, you’ll end up bracketing only in the direction of overexposure. If all this sounds confusing as you read it, believe me, it’s even more confusing in the field. Set the bracket order to metered, under, over, and save yourself a lot of head scratching.
Here’s how I calculate the correct exposures for the good-highlight and good-shadow frames. Let’s use a field of wildflowers in the shade with a sunlit peak in the background as an example.
Start by putting your camera in manual exposure mode. Next, turn on Live View, and cycle through the various displays until the histogram is displayed. Set the aperture you need to achieve the depth of field you want. This aperture will remain constant.
Now start adjusting the shutter speed while watching the histogram. Ignore the left (shadow) side of the graph. Right now you’re only concerned with the highlights. Keep adjusting the shutter speed until the highlights are almost but not quite clipped, as shown in figure 8-9. You’ve just determined the correct exposure for the good-highlights image. Right now, you don’t care about the shadows, which will be very dark in the good-highlights exposure.
Now begin lengthening the shutter speed, one click of the adjustment wheel at a time, counting clicks as you go. Watch the shadow data on the histogram data move slowly to the right. Ignore the highlights, which will probably become clipped. Stop when the shadows have moved comfortably away from the left edge of the histogram, as shown in figure 8-11. They don’t need to be dead center. You’ve now set the correct exposure for the good-shadows image. Leave this exposure set on the camera.
By counting clicks you have determined the difference between the good-highlight exposure and the good-shadow exposure. On most cameras, one click equals one-third of a stop. If you changed the shutter speed by six clicks, the difference is two stops. Turn on auto-exposure bracketing. If possible, set the number of frames in the bracket set to two. Don’t worry if you can only shoot a three-frame bracket set. You’ll end up with one highly overexposed frame, which you can discard. Now set the bracket interval, the difference in exposure between frames, to the same value you calculated as the difference in exposure between the good-highlight and good-shadow frames. In this example, the difference is two stops. Finally, set the drive mode to continuous high.
Canon users should be aware of this pitfall: setting the number of frames in the bracket set to two causes the camera to disregard the bracket order set in the menus. The bracket order is instead set when you use the Quick Control button (or other means) to set the bracket interval. The first frame will still be the metered exposure, but the second frame can be set to be either lighter or darker depending on whether you move the pointer to the positive or negative side of the index mark. Test your camera to be sure you understand how it will behave when you set the number of frames in the bracket set to two.
I always make the good-shadows exposure (the one I just set on the camera) first because I usually have to wait for the wind to stop. The good-shadows exposure will be the one with correctly exposed flowers. I want to nail that exposure first while the flowers are completely still.
If all this sounds complex, consider this: in a Rembrandt Solution situation, the difference between the correct foreground and background exposure will almost always be between two and three stops. If you calculate a one-stop difference, your camera can almost certainly straddle the range of brightness levels present in the scene. If you calculate a four-stop difference, you probably won’t be able to merge the two images without the transition line being obvious. If the difference is four stops, you’re in an HDR situation.
At this point you may be wondering, “Why not just use the Universal Exposure Strategy? That’ll cover all the bases.” The problem with the Universal Exposure Strategy in this case is that the compromise exposure recommended by the meter for the scene overall will attempt to straddle the difference between the bright highlights and the dark shadows and do justice to neither. This unacceptable compromise exposure will be the first one the camera makes. By the time the camera gets around to making the correct exposure for the flowers, the wind may be blowing again and the flowers may be blurred.
In the directions that follow, I’ll assume you’re using a PC.
The first step is to load the two exposures as layers in a single file in Photoshop. The easiest way to do this is to start from Lightroom. In Lightroom, in the Library module, select both images, then choose Photo>Edit In>Open as Layers in Photoshop. If you don’t use Lightroom, start from Bridge (which ships with Photoshop). Select both images, then choose Tools>Photoshop>Load Files into Photoshop Layers. And if you don’t use Bridge, then start from Photoshop itself. Choose File>Scripts>Load Files into Stack and navigate to the appropriate files.
Whichever method you use, the next task is to drag the dark (good highlight) layer to the top of the layer stack if it’s not already there.
Target the highlight (top) layer and add a layer mask to it by clicking the Add Layer Mask icon at the bottom of the Layers panel. It looks like a square with a circle inside it (figure 8-14). Be sure your foreground color is white and background color is black. (Press D for default colors of white and black.) Be sure you’ve targeted the layer mask, not the image itself, by clicking on the mask. Select the gradient tool from the Toolbar. Click the gradient icon in the Options bar (figure 8-15) (not the drop-down arrow adjacent to the icon) to open the Gradient Editor and make sure you have Foreground to Background selected (figure 8-16). Also in the Options bar, check that Linear Gradient is chosen as the gradient type (figure 8-15). Click, hold, and drag out a gradient in the image window, starting where you want the transition from white to black to begin, and ending where you want the transition to end. Photoshop will fill in the mask with solid white above the starting point of your drag and fill it in with solid black below the ending point of your drag.
White areas of the mask reveal the portion of the layer to which the mask is attached; black areas conceal that portion. Gray areas partially hide that portion of the layer, allowing the corresponding portion of the layer beneath to show through. The result is a blend of the two layers, with the balance between the two dependent on the shade of gray. The darker the shade of gray, the more the upper layer is concealed and the layer beneath is revealed. Figure 8-17 shows what the Layers panel should look like. Your image should now show the best parts of each layer, with an unobtrusive transition between the highlight and shadow regions. If it doesn’t, just redraw the gradient. There’s no need to hit control-Z to undo the first version. Photoshop will replace the first version with the new one. You can do this as many times as you like. Figure 8-8 shows the final image.
If necessary, refine the mask by painting on it with either white or black. Press B to get the Brush Tool. With the mask still targeted, paint with white to reveal more of the good-highlight region. Paint with black to reveal more of the good-shadow layer. Press the backslash key to see a translucent pink overlay of the mask. The pink overlay corresponds to black areas on the mask.
Let’s analyze the image from top to bottom just like I did when discussing split ND filters in chapter 7. Remember, white reveals, black conceals, the layer to which the mask is attached. In this example, the mask is attached to the good-highlight layer. At the top of the image, where the mask is solid white, all you see is the good-highlight image. As the mask begins its transition from white to black, still in the highlight region above the highlight/shadow line, you begin to see a blend of the correctly exposed highlights with the overexposed highlights from underneath. As a result, the highlights become a bit lighter, just as they did when using a split ND filter. Just below the highlight/shadow line, still under the transition part of the gradient, the dark shadows of the highlight layer begin to blend with the correctly exposed shadows of the layer beneath. As a result, the shadows become a bit darker, just as they did when using a physical filter. Finally, in the bottom portion of the image, where the mask is pure black, all you see is the correctly exposed shadows.
The image I chose for the example above works perfectly for this simple form of the Rembrandt Solution. The dividing line between highlight and shadow is nearly straight, and there are no large dark elements, like trees, that poke up into the highlight region. But what if the dividing line between highlight and shadow is irregular rather than nearly straight? The simple approach doesn’t work, as you can see in figure 8-18. Here’s how to handle even this extreme example.
Start by opening and stacking both images with the dark image on top, as before. Now, instead of adding a layer mask, press L for the Lasso tool. Draw a selection around the highlight region. You don’t need to be too fussy about making it perfect. In fact, you should draw the selection well down into the shadow region where the skyline is the boundary between highlight and shadow to avoid creating a halo (an unnatural bright region along the skyline) when you blur the mask in the final step. Add a layer mask by clicking the Add Layer Mask icon at the bottom of the Layers panel. Photoshop will automatically make the selected area white on the mask and black everywhere else. The image will look very odd, as shown in figure 8-19.
Now let’s make the image look natural. With the mask targeted, go to Filter>Blur>Gaussian Blur and blur the mask. You’ll need a high Radius, probably between 250 and 500 pixels, depending on the resolution of your image and how much you want to blur the mask. Click okay, and you’ll see that you’ve effectively blurred the mask, which in turn has softened the transition from the highlight to the shadow region. You can touch up the mask as needed by painting on it with either black or white. In effect, you’ve just created a custom digital split ND filter that follows the boundary between light and shadow. Figure 8-20 shows the result.
Before finishing this discussion of the Rembrandt Solution, it’s worth examining the Graduated Filter in Lightroom, which allows you to apply a technique I call the Rembrandt Solution Lite to a single frame. (If you only use Photoshop, you can find an identical tool in Adobe Camera Raw. The location of the tools is slightly different, but the result will be identical.)
The Rembrandt Solution Lite won’t work with images where the highlights are completely blown out or the shadows are solid black; in situations like that, you need to shoot two frames, one exposed for the highlights, the second for the shadows, as described earlier. The Rembrandt Solution Lite technique is appropriate for high-contrast scenes where a single frame contains adequate detail in both the highlights and shadows, but where the highlights are pale and the shadows are excessively dark. Using this technique can create the illusion of greater dynamic range than actually exists in the image. To put it another way, it allows you to place both the highlights and shadows close to midtone, so that they have rich color and detail, while marrying those regions in a way that our visual system finds believable.
Before using the Graduated Filter, use the Highlights slider in Lightroom’s Basic Panel to restore some detail and color to the washed-out areas. I often use settings between -50 and -100. Use the Shadows slider to open up the shadows. I find that settings of +50 or less work well. Beyond +50, the shadows can start looking rather odd. Rather than pushing that slider to still higher values, turn to the Graduated Filter.
Both Lightroom and Adobe Camera Raw, which ships with Photoshop, contain a Graduated Filter tool. The principle is the same for both tools, and the procedures for using the tools differ only slightly. Raw captures work best, but Lightroom and Camera Raw can also apply their graduated filters to JPEGs, TIFFs, and PSDs.
Start by choosing an image with highlights that are lighter and less saturated than you would like, but which still has recoverable detail in the brightest regions. If you’re using Lightroom Classic CC, open the image in the Develop module. Choose the Graduated Filter tool (fourth icon from the left, just above the Basic panel, figure 8-21). Set the Exposure slider to a negative value such as -1. Now click-and-hold in the image window and drag downward. A single dot will appear, marking the middle of the gradient, along with three lines that mark the top, bottom, and middle of the transition zone from full-strength adjustment to no adjustment at all. In effect, you’ve just made a selection. Everything above the upper line is fully selected, meaning it will receive the full strength of the -1 exposure adjustment. Everything below the lower line is completely excluded from the selection and will receive no adjustment. The region between the two lines is the transition zone, where the exposure change fades from full strength to zero.
You can reposition the gradient by dragging the central dot, expand or compress the transition zone by dragging the top or bottom line, and rotate the transition zone by positioning your cursor on the central line either left or right of the dot. The cursor will become a double-headed arrow. Click-and-hold, then drag, to rotate the gradient around the central dot. You can modify the position of the Graduated Filter and the strength of the adjustment at any time. Click on the central dot to reactivate the filter and modify it. And you can apply any of the other parameters listed below the exposure slider as a Graduated Filter adjustment, either singly or in combination.
You can further refine your selection by enabling Range Mask, found at the bottom of the Graduated Filter dialog box (figure 8-21). You’ve already selected which areas of the image will be modified based on your placement of the Graduated Filter. Range Mask lets you reduce the area affected by the Graduated Filter; it will not increase the area modified by the filter.
You have a choice of refining your selection based on luminance or on color. (At this writing the third choice, Depth, only applies to images from certain iPhones.)
If you choose Luminance, you have two sliders to control the effect. The Range slider controls the range of tones that will receive the effects of the filter. The scale runs from zero (black) to 100 (white). If you’re trying to darken a bright sky, for example, you might move the left (shadow) slider on the Range slider to the right to limit the effects of the filter to the highlights. Dark parts of the image will no longer be affected even if they were initially fully selected by the graduated filter. The Smoothness slider controls how much tones that are close in brightness to the selected tones are affected by the filter. A setting of zero produces hard transitions between the included and excluded tones. A setting of 100 produces a soft transition. Click Show Luminance Mask (figure 8-21) to temporarily convert the image to grayscale and display a pink overlay over the selected areas.
Choosing Color as the type of Range Mask (figure 8-22) lets you modify the area affected by the filter on the basis of color. Click the Color Range Selector tool (circled), then click in the image to include colors you want to be affected by the filter. For example, let’s say you used the Graduated Filter to darken a blue sky, but the decrease in exposure has also affected mountains projecting upward into the sky. To confine the exposure decrease to just the blue sky and remove the effect from the brown land, click in the sky. Click-and-drag to select a range of sky colors. Hold down shift and click in different areas of the sky to select additional colors. Click once in the tool’s receptacle to put the tool away. The Amount slider controls how much colors that are close in hue to the colors you clicked on are affected. Setting Amount to zero means only the specific colors you clicked on will receive the effects of the filter. Setting Amount to 100 means a broader range of hues will receive the effects of the filter, creating a softer transition between affected and unaffected regions.
Both types of Range Mask can produce odd effects if mishandled. For example, if you’re trying to darken an overly bright sky, look closely at the land portion of the image to see if you’ve created strange changes in color or contrast. Be sure you zoom in to 100 percent to inspect for halos along the boundaries of areas that the Range Mask has excluded from the filter’s effects. With both types of Range Mask, you can see what areas of the image are being affected by the filter by checking Show Selected Mask Overlay (just below the image window, on the left) or by pressing the O key. Note that the pink overlay shows selected areas—those that will be affected by the filter. This is the opposite of the way layer masks work in Photoshop, where pressing the Backslash key displays a pink overlay over the non-selected areas of the image.
There is one additional way to modify the effect of the Graduated Filter. With the Graduated Filter dialog box open, choose the Brush tool (circled in figure 8-21). I generally leave Feather, Flow and Density set to 100, and turn off Auto Mask. Now you can paint on the effect you’ve chosen (in this case, a decrease in exposure of one stop) over areas of the image that don’t already have the effect applied. You can also press-and-hold the Alt key, or choose the Erase mode (also circled), to subtract the effect from the selection. Note that Luminance and Color range mask settings apply to areas you’ve added to the selection by painting with the brush.
In situations that lend themselves to this technique, the digital Rembrandt Solution can quickly produce results that are both realistic and beautiful. While subject movement can be a problem, it’s often easier to manage movement using this approach than it is when using HDR software. Let’s say the wind never stops completely, so that a few blossoms are blurred, even in your very best frame. If you use the Rembrandt Solution, only one layer (the good-shadow layer) is visible in the image region containing the flowers. You will see slight motion blur in the blossoms that moved during the exposure, but you won’t see totally unnatural ghosting, with two versions of the same blossom faintly visible through each other, as can happen with HDR software. The only region where motion could be a problem is under the transition zone between the two images. Usually that can be placed on the midground or background, where it’s unlikely that subject movement will be noticeable.
My first choice when shooting high-contrast scenes is to try to get all the detail I need in a single capture. If the dynamic range of the scene exceeds the range of my sensor and subject movement between frames is likely to be a problem, I turn to the Rembrandt Solution. If the scene is high contrast but the subject is stationary, I resort to high-dynamic-range software.
When I wrote the first edition of this book, I was an HDR skeptic. Today I am a convert. HDR imaging has come a long way toward solving one of the great challenges of photography—taking the broad range of tones we see in the real world and compressing it into the much narrower range of tones we can see on a monitor or reproduce on paper.
Before tackling HDR techniques in detail, let’s dive a bit deeper into the concept behind HDR. The number of brightness levels a digital image can contain is determined by its bit depth. JPEGs, for example, are 8-bit images, which means they can contain 256 brightness levels (shades of gray, in black-and-white terms), from solid black to pure white. Most DSLRs capture RAW files in 12 or 14 bits, which is then interpolated up to 16 bits. A 16-bit Photoshop file can contain 32,769 brightness levels. (The mathematically inclined among my readers will have noticed that 16-bit Photoshop files actually only have 215 brightness levels for technical reasons we don’t need to get into here.) While greater bit depth doesn’t directly translate to greater dynamic range, it does help preserve smooth tonal gradations in high-contrast images. A 16-bit file can cover a broad brightness range, but it’s still not enough to cover the full dynamic range of a high-contrast scene.
Conventional images use an integer data type (don’t ask). By contrast, HDR images are floating-point images, which can contain far more brightness levels than integer-data type images—enough to cover the full dynamic range of the highest-contrast scenes we encounter. Unfortunately, HDR images contain such an enormous range of brightness levels that no conventional monitor can display them, and we’ll probably never be able to print them using ink on paper. Before an HDR image is usable, it must be tone-mapped. In other words, the dynamic range of the image must be intelligently compressed into a range we can display on a monitor and then print. HDR programs differ both in the quality of the initial floating-point file they produce and in the quality of their tone-mapping algorithms. As I write this, the best way to achieve a natural-looking HDR image is to use the HDR utility built into Lightroom Classic (the desktop version).
Here’s how to use the HDR approach in the field. As with the Rembrandt Solution, be sure to lock your camera down on a solid tripod and be careful not to move the tripod or camera when adjusting the exposure. When changing exposure, be sure to change the shutter speed, not the aperture. Changing the aperture will change the depth of field, which could prevent the images from aligning perfectly.
At a minimum, you should bracket high-contrast scenes in a set of three, with a bracket interval of two stops. In other words, shoot a frame at the metered exposure, then at -2 stops, then at +2 stops. In most situations you can let the camera pick the starting-point exposure, then bracket around that setting. With certain subjects, however, that approach will fail. Let’s assume you’re working on a shot where 80 percent of the frame is shadowed rock and 20 percent is ultra-bright sky (figure 8-26). Your meter will recommend an exposure that will render the dominant subject—the shadowed rock—as a midtone. That’s all the detail you need in those dark rocks; you really don’t need any lighter exposure for the shadows. However, if you’re bracketing around the zero exposure-compensation mark, your camera will also make an exposure two stops brighter and an exposure two stops darker. The +2 exposure is unnecessary (not a big problem) but the -2 exposure may not be dark enough to bring in all the detail you want in that bright sky (a very big problem). The solution? Set your exposure compensation to -1, and bracket around that starting point. That gives you -3, -1, and +1 as your three exposures. You’ll still have plenty of shadow detail, and you’ve brought in the highlights as well. In extreme situations you may need to set exposure compensation to -2, which will give you the exposures -4, -2, and zero.
Similarly, if the shot involves 80 percent bright sky and 20 percent dark land (figure 8-27), the meter will recommend an exposure that will render the sky as a rich, saturated midtone. If you’re bracketing around the zero exposure mark, however, you’ll get a frame that’s two stops darker (unnecessary, but not a big problem) and a frame that’s two stops lighter, which may not be light enough to give you the detail you want in the land (a very big problem). The solution is to set your exposure compensation to +1 and bracket around that starting point, giving you -1, +1, and +3 as your bracket set. You’ll still have plenty of highlight detail, and you’ll have adequate shadow detail as well. Again, in extreme situations, you may need to set your exposure compensation to +2, which would give you the exposures 0, +2, +4.
Bottom line: if the scene is predominantly dark, bias your starting-point exposure toward dark (-1 or -2 exposure compensation); if the scene is predominantly light, bias your starting-point exposure toward light (+1 or +2 exposure compensation). Then check the histogram for each frame in the bracket set to make sure you have adequate highlight detail in your darkest frame and adequate shadow detail in your lightest frame.
For the highest-contrast scenes, such as ultra-wide panoramas with the sun in the frame, you may need to shoot five frames at a 2-stop bracket interval. This covers a range from -4 stops to +4 stops rather than -2 to +2. This disadvantage of this approach is that the +4 exposure, if not actually necessary, may be so long that it wastes precious seconds of perfect light. My preference is to use exposure compensation to change the starting point for a three-frame bracket set whenever possible.
HDR software is always improving. If you capture a wide range of exposures today, you may be able to improve your rendition of the image in the future as the software becomes more capable. Regardless of how many frames you shoot, be sure to check your histogram after your first bracketed sequence to be sure you have at least one frame with excellent shadow detail and one frame with excellent highlight detail.
The drawback of using a two-stop bracket interval is that you may leapfrog the perfect exposure for a certain part of the scene. In some situations, such as those where parts of the subject never stop moving (a stream or waterfall, for example), you may want to use the Universal Exposure Strategy I described in the previous chapter: shoot a five-frame bracket set with a one-stop bracket interval. Here’s an example of how that can help you handle a very difficult exposure situation.
Let’s say you’re shooting a stream backlit by the rising sun with the sun in the frame, as shown in figure 8-29. HDR software works well for the stationary parts of the scene, but the flowing water exhibits ghosting. On close examination, you can see two translucent versions of the same waves and ripples roughly stacked on top of each other but not perfectly aligned. HDR software offers tools to correct ghosting, but they’re imperfect at best. The solution is to create an HDR file from all five frames in the bracket set, then edit the HDR file for best detail in areas surrounding the flowing water. Now identify the single frame that best shows the flowing water, ignoring other areas of the frame that may be too dark or blown out. Edit the single frame so the flowing water looks perfect. Bring the HDR version and the single best-water frame into Photoshop as layers in a single file. Drag the HDR layer to the top of the layer stack. Add a layer mask to the HDR layer. Choose the brush tool, set the foreground color to black, and be sure the mask is targeted, not the image itself. Now paint over the ghosted water in the HDR layer to reveal the good water from the single-image layer beneath.
Once you’ve captured the component images, the next step is to combine them into a floating-point file. Fortunately, Lightroom Classic now offers a simple way to do just that. Select all the images you want to combine, then choose Photo>Photo Merge>HDR. Alternatively, you can right-click on any selected image and choose Photo Merge>HDR. Lightroom will produce a 16-bit floating-point RAW file using Adobe’s open-source DNG file format. That means Lightroom’s HDR files retain all the editing flexibility you expect from a RAW file.
The next step is to open the image in Lightroom’s Develop module. The first thing you’ll notice is that the Exposure slider in the Basic Panel now provides a range from -10 to +10 stops. For an ordinary integer-based RAW file, the range is just -5 to +5 stops. That gives you an idea of just how much data you get to play with. Don’t expect to do all the work with the Exposure slider, however. You’ll probably need to make some large adjustments of the Shadows, Blacks, Highlights, and Whites sliders to bring out all the detail in the file. Many HDR images benefit from adding contrast in the Basic or Tone Curve panel, and you may need to further adjust shadow and highlight density with the Graduated Filter and Adjustment Brush. You’ll find that you can recover more clean, usable detail, even in the darkest shadows and brightest highlights, from an HDR image than you can from a standard RAW file. And it’s all non-destructive. You can return to the image at any time to make further refinements. If your goal is to create natural-looking images of high-contrast scenes, you’re going to love this approach.
Recently I’ve begun experimenting with a hybrid Rembrandt Solution/HDR technique for shooting grand landscapes with wildflowers in the foreground (figure 8-30). Here’s the problem: wildflowers look best in soft light, which is usually dim, but the mountains in the background look best in strong, bright light. That means I often end up shooting grand landscapes containing wildflowers in high-contrast light. The Rembrandt Solution can work well if there’s a clean dividing line between the shadows and highlights, but not so well if a dark shadowed tree projects up against the bright sky or if the wildflowers themselves are silhouetted against a bright background. The standard HDR approach of shooting a three-frame bracket set, two-stop bracket interval, works fine if the day is completely windless—a rare occurrence in the mountains. Most often there’s a breeze, which means waiting for the wind to stop so you can capture razor-sharp flowers. Often the lulls are very brief. If you’re using the standard metered-under-over bracketing order, the camera won’t record the flowers correctly until the third frame. By that time the wind may already be blowing again.
Here’s my solution. I use the Rembrandt Solution metering strategy described earlier in this chapter, which ensures that the very first frame I shoot will give me perfectly exposed flowers. That way I can take advantage of even the shortest lull in the breeze. The second frame will record proper detail in the bright background. I then use Lightroom’s HDR utility to merge the two frames. Although this approach could certainly produce ghosting with some pairs of images, I’ve found it works surprisingly often. When it does, you can achieve a natural look more easily than you can using the standard Rembrandt Solution, and retain all the editing flexibility of working with a RAW file.
Despite all the advances in HDR software in recent years, achieving a natural-looking result can still be a challenge. As I’ve discussed, we’ll never be able to create an image on paper that displays the full dynamic range we can see in the real world. In that sense, it’s impossible for a print to ever look completely real. But it is possible to get close—close enough that a print evokes in the viewer many of the same emotions that the real scene evoked in you.
My guide when preparing prints has always been what I saw, rather than what my film or sensor captured. I was never satisfied with the limited dynamic range of my 4×5 film, for example, and I carried eight graduated neutral-density filters to try to capture shadow and highlight detail the way I saw it. When affordable film scanning became available, I began using Photoshop to further adjust shadow and highlight density. Digital capture was another step forward, but even the high-end DSLR I’m using today still has less dynamic range than my eyes, which is why I use a variety of digital techniques, including HDR, to capture what I saw.
But how do you really know what you saw hours, days, or weeks later? According to Rochester Institute of Technology professor Mark Fairchild, people are notoriously bad at remembering colors. We can distinguish thousands of different colors if they are placed side by side, but we can accurately remember less than 100. We can easily remember if the flowers were blue or red, for example, but we quickly forget what shade of pink we saw in sunset clouds. To further muddy the waters, according to Fairchild, people tend to remember colors as more saturated than they actually were. In addition, we tend to substitute certain memory colors for common objects. For example, we remember yellow-green grass as greener than it actually was, and we tend to remember sky as pure blue when it actually wasn’t. Your best guide to what you saw is your original bracketed set of images—with one qualification.
Let’s assume you’ve created an HDR file in Lightroom. As you begin to adjust the file, compare the region you’re working on to the frame from your bracketed set that is properly exposed for that region. If you’re adjusting the portion of the image that contains flowers, for example, “properly exposed” may mean the frame in which the green foliage surrounding the flowers was rendered as a midtone. If you’re adjusting a sunset sky, however, don’t pick the frame in which the glowing clouds are midtone and everything else is black. As pretty as those clouds may be, they’re underexposed. Glowing clouds are a highlight; they should be brighter than midtone. Midtone clouds will look unnaturally dark if placed in a landscape where the flowers beneath are also midtone.
Here are some other key principles for making your HDR photos look more realistic:
In the days before digital, a film aesthetic ruled. Straight photographs were generally deemed “realistic” even though our eyes could see more detail in the highlights and shadows of the actual scene than the film could record. For a high-contrast scene in the film era, bright highlights and inky black shadows were considered realistic because that is what we were used to seeing. The invention of digital photography gave photographers greater control over highlight and shadow density, and the film aesthetic began to seem less realistic because it didn’t capture the real-world scene as accurately as a digital sensor could.
Today’s HDR software goes even further: it gives you total control over the density of every part of your frame, from the deepest shadows to the brightest highlights. This unprecedented power has opened the door for a new aesthetic, but this new aesthetic has not yet become widely accepted. Some people love HDR and the way it can render every part of the scene in rich detail; others hate it. There is no consensus among landscape photographers about the best use of the incredible new tools at our disposal. Opinions on HDR vary, and the technology continues to evolve; in this period of flux, your best guide to producing a realistic image—one that represents what you saw during capture—will be your bracketed set of images, your knowledge of how our visual system processes high-contrast scenes, and your own good judgment.