A Case for Exposing to the Right
In the quest for optimum image quality, photographers often debate the merits of how to best capture the information present in a scene. Back in the film days, we routinely altered the exposure of our slide and color negative films to get just a bit more saturation and ‘pop’ in the colors, or density in the shadows. For black and white shooters we had Ansel Adam’s Zone System to rely on for manipulating tonal values through exposure, processing, and subsequent printing. Many times this involved under or overexposing the scene to obtain the right placement of shadows, then developing for the highlights. Digital capture and post production methods provide us with the opportunity for much greater control over the value distribution in a given scene and because of this we cannot simply use the same way of thinking about exposure with digital as we did with film. If highest quality images are the goal, then new methods of working with digital capture must be established, and with that a deeper understanding of how modern cameras record information must be obtained.
Many photographers wrongly assume that the information captured in their images is distributed evenly along the gradient ramp from black to white. If we take a step back and begin to understand that digital sensors do not capture information this way, we can then begin to alter our exposure methods so that the optimum quality is obtained. Using a 12-bit capture as an example, which has 4096 levels of value per channel, and ignoring color for this example, we can see that more than half of all the information for a digital capture is recorded in the top 2 stops of an image.
Using this knowledge we can begin to understand why shadow values that are underexposed in camera and corrected in post begin to exhibit high levels of noise and banding, and that these areas often feel ‘chunky’ as well as lacking in subtleness that is the hallmark of many fine quality landscape photographs.
So, how does one go about making sure they capture the best information in a given scene, while making sure that the highlights don’t get blown out? There are a few pieces to this puzzle that must be in place so that you can work the system, so to speak.
- Always shoot in RAW. This will ensure that you have 12-14 bits of data per channel to work with, and not a compressed 8-bit JPEG.
- Understand that the histogram on the LCD of your camera is generated from an internal jpeg, and based on the Picture Style selected within the camera settings. It is not a totally accurate representation of the actual data in the RAW capture. If you have the zebra pattern or highlight warning turned on in your camera, very often the RAW image has recoverable information in the highlights. But this is where it gets very tricky when the contrast in the scene is very high. You do have to watch for complete overexposure of highlights.
- This is not the holy grail of photographic capture. It can help you create better, cleaner, and more noise free images in many situations, but cannot be applied carte blanche across every image you capture.
In the next section we will take a look at a few case studies where this technique helps to create higher quality captures that provide opportunity for better results in post production. Very often photographers will immediately begin dramatically overexposing all their images, with little success in recovering information. This is because a lack of awareness of how far one can push exposure in the highlights without losing needed detail. Apply this information with a subtle hand and your images will begin to have cleaner shadows, less noise, and fewer situations where banding is a concern.
In the following section we will take a look at a few images where the ETTR technique was very useful in creating lower shadow noise, more nuanced highlight detail, less banding, and posterization, and an overall cleaner image.
On a bright day in Michigan, I went out to record the patterns of blown sand and snow. This image is an example where I used the Expose to the Right or ETTR technique. In Figure 1 you can see that it is clearly overexposed from where the final tone distribution will be. In this case the main goal was to maximize higher value smoothness and fluid tonal transitions throughout the snow. Shadow detail in this image was not a consideration. In Figure 2 the original histogram shows the majority of the tonal distribution to be in the top two stops of dynamic range.
Figure 3 shows the comparison between what appeared to be near highlight loss on the right and the final adjusted image on the left. All the adjustments needed to the image were done globally through the Basic adjustment panel in Lightroom CC.
In order for this technique to work effectively you have to pay close attention to clipping highlights, but also be aware that the histogram is very conservative, based on an internal jpeg and not the RAW file itself. I tend to also be conservative and not let the highlights lose detail in the histogram. However, the knowledge that there is additional information there to recover in case I make an error is comforting. Figure 5 shows the final result.
Case Study 2
The next example has a much wider dynamic range, which is more typical of a scene where this technique is useful. An image of a sunrise in West Virginia created an opportunity to make sure that the shadow detail was open enough and noise free. In this image the sky was nearly pure white, with virtually no definition, as the histogram on right shows.I wanted to keep the shadows more open than normal, just to keep a sense of lightness in the image. Because of this there is no true black point in the final image, as indicated by the adjusted histogram in Figure 3. Figure 4 shows the very open, and virtually noiseless shadows. This is something that would not occur in a situation where the exposure was made for the sky. For this exposure I pushed the highlights as far as I could, and the histogram indicated near pure white with no detail in the lightest portions of the sky.
I wanted to keep the shadows more open than normal, just to keep a sense of lightness in the image. Because of this there is no true black point in the final image, as indicated by the adjusted histogram in Figure 3. Figure 4 shows the very open, and virtually noiseless shadows. This is something that would not occur in a situation where the exposure was made for the sky. For this exposure I pushed the highlights as far as I could, and the histogram indicated near pure white with no detail in the lightest portions of the sky.
As you can see from this example, RAW files have significantly more headroom than one might expect. This information stored towards the right side of the histogram is easily recoverable, and in many cases provides the image characteristics that photographers are seeking in their work.
Case Study 3
In this last case we will take a careful look at the actual noise difference in a slightly underexposed capture, and one that is optimized towards the ETTR technique. A image from Jökulsárlón in southeast Iceland provided the images for this example. In the case above the image on the left is a classic underexposed image, which can be fixed in post, but at the expense of noise free shadows, and smoothness throughout the tone scale. For this example I will be converting both images to BW so that we can concentrate on noise levels. Figure 2 shows the exposure histograms. The image on the left had a neutral density filter on the lens, and was not adequately corrected in the exposure, hence the drastically different exposure statistics. What is more interesting is that the underexposed 100 ISO image will show significantly more noise than the overexposed 500 ISO image, which we will see in the detail images below.
At first glance both images as seen below in their final state exhibit very similar tone structures and would be ‘acceptable’ to many photographers. There are some differences in highlight and shadow densities, but generally the two exposures are similar. Now lets take a look at the detail of the shadows in both images and compare noise levels. Keep in mind that the the ETTR image was exposed at a different ISO than the original, so this comparison would have been even more dramatic at the same ISO.
Figure 4 shows a detail of the noise structure of the underexposed image. While not excessively noisy, the darkest shadows do exhibit a lack of smoothness, especially when compared to the lighter gray tones towards the bottom of the frame.
Given the fact that the images were shot at different ISO settings and one image exhibits blurring from the moving ice chunks, one could argue that these aren’t ideal examples, but in Figure 5 we can see a smoothness that is not present in Figure 4. Figure 6 is a side by side comparison of the two images at high magnification.
Whether or not to use this technique is entirely up to the goals each of you have as photographers. From my perspective the technique is not a panacea, but a tool to have in your kit, for situations where you want the smoothest tonal transitions throughout an image. At the least it is beneficial to understand the basics of how digital sensors capture information, and to maximize how you leverage this knowledge throughout the capture and editing process.
In summary, I often use this method for high key images that have little shadow information, so as to maximize the smoothness and transitions in subjects like snow, sky, and reflections on water. For higher contrast situations, I apply this technique more sparingly, and more often resort to combining multiple captures into an HDR file. However, for those cases when a single capture works, I tend to push the histogram towards the right side as far as I can, without losing important information. You will notice an improvement in your images, as long as you are willing to put in just a bit more time in post production, and follow good capture methodology. My advice now is to get out there and give it a try on your favorite subjects and see how it works!