On the topic of creating high dynamic range (HDR) photographs I was recently asked a question about the viability of taking a minimal number of bracketed exposures in a high contrast situation and later filling in the gaps in the exposure range with digitally generated exposures. While this isn’t a new concept in high dynamic range photography, it is one that comes up frequently and isn’t entirely intuitive. I felt that others might benefit from reading the Q&A or sharing their own thoughts on the topic.

The contrast range in this scene was too great to be contained in a single exposure.

The contrast range in this scene was too great to be contained in a single exposure.

QUESTION: I photograph with a Nikon D5000. One of the features I am less than satisfied with is that auto bracketing is limited to three exposures. Sometimes that is simply not enough. The conventional wisdom seems to indicate:

1. don’t spread out your brackets more than one or two exposure values (EV).
2. Lightroom/Adobe Camera Raw can safely push/pull an exposure ±2 EV

What if I exposed three captures -3 EV, 0 EV and +3 EV? Then I should be able to make virtual copies of these in Lightroom, spreading each ±1 EV and the ends also ±2 EV. I would then end up with effectively an 11 stop range which is large enough for all but the most extreme lighting conditions I have encountered.

Do you think this technique has a decent chance of producing a quality HDR image? Note that my goal is to produce fine art prints, so to be useful for me it would have to produce very clean digital negatives.

Bracketed exposures from -3 EV to +2 EV

Exposures bracketed in camera from -3 EV to +2 EV

ANSWER: The first consideration is how the high dynamic range image will be created. You can combine bracketed exposures using one of many HDR software applications or you can manually blend exposures using masking techniques. Exposure requirements will vary depending on the contrast of the scene and the blending technique. Having a full set of exposures at one stop increments can be more important for HDR software than manually blending exposures. HDR software usually does best (and often requires) being given a full range of exposures with narrow and consistent EV increments, as you point out. One stop is common, but I know people who go to extremes and bracket in 1/2 or 1/3 stops. I don’t use HDR software in my workflow. All the HDR software I have used tends to create image quality and adjustment control issues of various types, regardless of how the bracketed exposures were created.

Even though I usually take a full range of exposures, often only two of the exposures are required when manually blending using masking techniques. If I have one complete exposure for the sky and one complete exposure for the landscape, then I only blend these two exposures and keep my life simple. This video from my Extending Dynamic Range tutorials demonstrates a simple two exposure blend.

You may get lucky while using your camera’s three shot auto bracket feature and find that you captured the precise two exposures you need for such a blend, but since every scenario is different there isn’t a reliable way to reproduce this every time.

In more extreme contrast situations there may be brighter areas of the sky or very extreme areas of shadow and highlight in the landscape that require the use of additional exposures. When manual blending, one can make decisions about which exposures and how many exposures to use on an case by case basis. Having a full range of exposures to choose from is helpful in this respect. Blending exposures with masks is a challenging skill, but with practice I find that better control and better final image quality can be achieved. Some very fine printed images are being created with HDR software to be sure, but if the best quality is your goal you will eventually run into some downsides.

So, if you don’t capture a full range of exposures is it a good option to fill in the gaps digitally? As you describe, it is possible to create virtual copies in Lightroom with different EVs. Per the “conventional wisdom” Lightroom or Adobe Camera Raw certainly can be used to push or pull exposure. However, to what degree this can be done and what can be considered “safe” is subjective. It depends on whether you push or pull the exposure and what side affects you are willing to accept.

If the best image quality is your goal, having “properly” exposed captures can be just as important as the number and increment of those exposures. What constitutes a “proper” exposure isn’t necessarily intuitive. Digitally reducing  exposure (pushing) in Lightroom or Adobe Camera Raw is generally “safe”. In fact it can actually be preferable. One way to hide noise in an image is to digitally decrease exposure making the image darker. Intentionally over exposing (also known as Exposing To The Right or ETTR) and then digitally lowering the exposure produces better image quality. This is why ETTR is commonly advocated in digital photography. When it comes to bracketing, digitally creating +1 EV and +2 EV exposures from a +3 EV capture in Lightroom should give great results in the mid-tones and shadows. The highlights will be clipped but these should be contained in your darker exposures.  For a more complete run down on the virtues of ETTR and some examples check out this article by Jeff Schewe.

Digitally increasing exposure (pulling) is where you will run into problems. Dark regions of an image are very data poor so they have a low signal to noise ratio. The reason they are dark is because relatively few photons (signals) were collected by the sensor. Digitally increasing the exposure of underexposed images boosts the signal but also boosts the noise. With a low signal to noise ratio the noise will overpower the signal. An image captured at -3 EV and then digitally lightened two stops to -1 EV will show substantial quality issues. For best image quality it is preferable to avoid lightening dark areas as much as possible.

If you are using HDR software the following is important to note. As far as I know, no HDR software is capable of reading the exposure adjustments you make to raw or DNG files in Lightroom or Camera Raw. Raw adjustments are only parameters contained in the metada and HDR software does not access these parameters. This means that in order for HDR software to use your digitally bracketed exposures you would first need to open the virtual copies and save them as tif files with different EVs, and then load the tif files into the HDR software.

So, your proposed technique can work to an extent if you take the right steps. In many cases it may even produce very good results. However, to minimize noise, it would be best to create your digital exposures by always lowering the exposure of the brighter frames and not the other way around. For example, generate +1 and +2 EV virtual copies from a +3 EV capture, -1 and -2 EV copies from a 0 EV capture and -4 and -5 EV copies from a -3 EV capture.

There are occasions in which I will create exposures digitally for the purpose of blending. Sometimes I fail to take a complete range of bracketed exposures or I’m not able to use one or more of my exposures due to things moving in the frame between shots. At a minimum I make sure I have one exposure for highlights, one exposure for mid-tones and one exposure for shadows. Usually I try to capture a full range of exposures if I can. This ensures the most flexibility. A properly exposed series of images with one stop increments will give you a more complete gradient of tonal information to work with compared to capturing just three exposures taken at three stop increments.

Taking all of the above into account, let’s address the central motivation behind your question, the limited auto bracketing capability of your camera. If your camera would auto-bracket more than three exposures it is unlikely that you would be considering using digitally created exposures in the first place. I recommend making it a non-issue by not using the auto bracket feature at all, regardless of how many shots your camera will bracket. My camera will auto bracket up to seven EVs but I never use auto bracketing unless I have a specific reason to (such as hand-held bracketing for example). Nearly every scene requires a different number of exposures and the range of exposures is rarely centered perfectly around the meter’s 0 EV. Using an arbitrary number of auto bracketed exposures means regularly capturing too many or too few exposures (in your case with a max of three, usually too few). And if the actual exposure is shifted from what the camera meter picks as 0 EV, then you might have the right number of exposures but going in the wrong direction. Manually bracketing exposures solves these issues. I take my first frame at 0 EV and then I check the histogram. Sometimes one exposure is all I need. If the histogram shows clipping of highlights or shadows or both, then I compensate my exposure up and/or down one stop at a time until I have one frame with a histogram that properly exposes for the shadows, one frame with a histogram that properly exposes for the highlights and some number of one stop increments in between. Sometimes the scene dictates two exposures, or three or five or eight or eleven. Whatever the contrast range requires, this technique ensures that I have what I need, but not more.  This video from the Extending Dynamic Range series illustrates this concept further.

Another manual bracketing technique I use is tonal region bracketing using my camera’s live view feature. In aperture priority mode I move the exposure target box around on the screen to various tonal zones, such as an area of brightest sky, mid-tone land and dark shadows, and take a shot in each zone. The camera automatically adjusts the exposure for each zone as I shoot. The result is a series of exposures that should contain the proper exposure for each tonal zone. They may not be one stop increments, but as long as I have optimal exposures for highlights, mid-tones and shadows I don’t care. Of course, this technique doesn’t work well when using HDR software which expects consistent EV increments.

Finally, not every scene we encounter has high contrast light so bracketing is frequently not necessary. When photographing in balanced, low and medium contrast light situations all the tonal information readily fits within a single exposure. I regularly come across people auto bracketing nine exposures in low and moderate contrast light, just as a matter of habit. In such situations it isn’t necessary to take two exposures, let alone nine. I start with one exposure and look at the histogram. If I see that all the tonal values have been captured, I’m done. Taking valuable time and filling up memory cards with additional exposures is something I do only if I absolutely have to. Many of my images are captured in a single exposure. With a single exposure I attempt to get a proper exposure (ETTR) without clipping the highlights. In post this enables me to decrease the EV and do as little shadow recovery as possible. This gives better results than starting with an underexposed image and trying to lighten it to bring out shadow detail. Again, see the Jeff Schewe article for more insight into why this is the case.

Photo Cascadia Logo

Keep in touch with

Photo Cascadia

Join the mailing list to receive the latest news, articles, events and workshop updates from our team. We publish one newsletter every two months.

Thanks for subscribing!

Pin It on Pinterest