The Theory Behind HDR

First, the basics. HDR stands for High Dynamic Range and encompasses any technique that increases the range of light values recorded in an image beyond the normal sensitivity of the sensor or film. For the purposes of this article I am only going to refer to the technique of utilising multiple frames of the same scene to create a final image that has more detail in the shadows and highlights than a single frame would be able to capture. It is possible to create an HDR ‘look’ from a single frame, but what I am talking about here is the theory behind HDR, and so I will restrict myself to the use of multiple frames. This includes processing a single Raw to give several images to work on.

Remarkably the first record of HDR pictures was way back in 1850. The French photographer Gustave le Grey took multiple exposures of seascapes, exposing for the sky and for the sea separately and then using the two negatives selectively to get a more lifelike print. Photographers have used similar techniques ever since but the skill required to produce natural results has meant that this was mainly something done by high end commercial photographers.

Modern HDR still does the same thing as those early photographers did but in a much more sophisticated way, allowing for images to be combined pretty much on a pixel by pixel level.

Do you need HDR?
So is it necessary? Well, if you take a picture and you can’t get the entire dynamic range to sit within the histogram then HDR is a great technique to avoid blown highlights or blocked up shadows. Your alternatives would be to put up with the clipping, fixing it in post-production, to add a graduated or polarising filter if appropriate, or to use lighting or reflectors to lift the shadows if possible. HDR however can give a more natural result than any of these alternatives because it is not altering the recorded scene, it is just recording more information and then interpreting it in a different way.

HDR does have drawbacks but most can be worked around. The main thing is that to do HDR well you need to plan. It is not something that you are likely to be able to fall back on in retrospect when you have discovered that the highlights in the pictures you shot the previous day have no detail in them. Where HDR really shines is in scenic photography: situations where you are able to set the camera up and take a series of frames with no movement, landscapes and cityscapes in particular and the interiors where light is pouring through the windows. If you plan carefully it is possible to include moving subjects, but make sure that the subject is correctly exposed in the frame you want to use.

The starting point for HDR is a source set of exposures which spans the entire tonal range in the scene, so using your camera histogram as a guide take a central exposure, and bracket in one or two EV increments either side, taking as many exposures as you need until the data is well clear of each end of the histogram – especially the left so that there is plenty of noise free data available for the HDR software to build the shadows. Some cameras make this particularly easy with autobacketing but you will still need to check the histograms to see that you have the range covered. The files need to be in a format that can be read by your HDR software but most will read various Raw formats as well as TIFF and JPEG. It is important that the images should be consistent: white-balance and aperture and focus should be the same with the speed varying. Ideally they should be shot quickly with the camera on a tripod.

Software pointers
With respect to software, there are many different options ranging from free utilities through to Photoshop CS5. The market leader is Photomatix Pro by HDRsoft. Turn to page 96 for reviews of three HDR softwares. Whichever you use read the manual first because it will tell you how many pictures you need, what format they should be in and how aligned they need to be in order for the software to work.

The software will do different things and it is helpful to have an idea of what the techniques do to the data when trying to choose which to use. There are two main branches of HDR processing: tone mapping and exposure fusion, and they differ fundamentally in what they do. Exposure fusion takes the data and creates a weighted average for each pixel so the shadows will have their data taken mostly from the over exposed frame and the highlights will have their data taken mostly from the underexposed frame. The result from this processing is a very natural looking but flat contrast. It does not substantially alter the rendering of the detail and the file it creates is immediately useable, since the data is compressed into a range that printers and screens can display.

Tone mapping refers to the way that the software displays an HDR image. The actual file is an HDR file, with each level of light recorded correctly, but you would not be able to see this on a print or on screen. The tone mapping process looks at the data in the file and renders it in a way that can be displayed, which then allows you to save the version for use. The file retains all the data and can be saved for alternative tone mapping in different versions. Tone mapping offers many additional options allowing you to alter local or global contrast within the scene and is the technique that produces the more stylised HDR images that have become so popular.

The elements that can be manipulated in exposure fusion relate mainly to the way that the averaging is weighted: in essence what the controls do is to adjust the curve of the image contrast.

In tone mapping the software interprets the image according to the distribution of tone in the picture. In areas where there is relatively low contrast it can boost the local contrast to emphasize detail. This means that taken as a whole the relative density of different areas in the image are somewhat elastic, so it is possible for darker areas to end up lighter than parts of the picture that started out lighter in the first place. This is, for me is often too much, it goes beyond what can reasonably be called photography and becomes digital art. It is essentially the work of the software rather than the skilled use of a tool.

In conclusion, HDR photography is very clever: used well it can produce superb results with amazing detail in highlights and shadows but still looking very natural. It is not an end in itself though, as first you need a great image with bracketed exposures that you can enhance. If your source picture is dull then the result will almost certainly be a dull HDR.

Taken from the February 2011 issue of Advanced Photographer magazine