Pixiq

Digital photographers are quick to adopt software that gives them more control over processing and different options for the look of a final image. One of the most popular examples right now is the use of High Dynamic Range (HDR) software to deal with situations in which the brightness range of a scene exceeds the recording latitude of the camera’s sensor. This approach also yields results that have very high levels of color saturation as well as extraordinary shadow and highlight detail that has really caught on with many photographers. The usual procedure is to take a series of bracketed “source” exposures that together capture a wide range of brightness. These images are then processed in HDR software into a single file containing all the captured brightness. But the dynamic range of this file is so large, monitors cannot display the full brightness nor can printers output all the data. 

Enter the need for a second stage of tone mapping in order to bring the image into a lower dynamic range usable by monitors and printers. To do this, HDR programs apply sets of algorithms to make the transition and this typically results in the extraordinary results that appeal to some but others find quite unreal. So is there an alternative? Specifically, is there a software approach that will capture a wider dynamic range but with realistic levels of color saturation and contrast as well as believable details in both shadows and highlights? In other words, it looks more like what our own vision system tells us about a scene. 

Software using the Exposure Fusion (EF) processing model is an alternative approach that characteristically produces such results from bracketed exposures. It is based on the published work of several software engineers over the years in the field of display graphics and given that name by Tom Mertens and colleagues in a summary paper presented at a computer graphics forum a few years ago (http://research.edm.uhasselt.be/~tmertens). 

 Unlike HDR processing, EF does not need a tone mapping stage since it does not produce an image with an excessive dynamic range. Instead, EF software builds a final image directly through sampling, evaluation and then selection of pixels from the source images. Or to put it in the words of Merten’s paper: Guided by our quality measures, we select the “good” pixels from the sequence and combine them into the final result. Furthermore, because the EF approach is based on pixel selection it is possible to introduce flash or reflector fill light into the weak shadow areas of a single source image. Consequently, these areas with better illumination will tend to be selected over those with lower pixel counts. This also tends to keep shadow noise to a minimum, which is not necessarily the case with tone mapping. So typically, the final EF processed file not only contains a realistic representation of a wider dynamic range but it can also be cleaner in terms of noise in shadow areas.

Right now, there are just a few choices in EF based software. Photomatix, the most popular HDR program offers an EF alternative selection to HDR. Then there is Enfuse, a utility built on Merten’s algorithms (http://enblend.sourceforge.net/index.htm). I prefer to use Brackeeter, available from Pange Software (http://www.pangeasoft.net/pano/bracketeer/index.html) that is a simple to use graphic interface for Enfuse. It is also inexpensive at $29.95 and has one advantage over Photomatrix’s fuse exposure approach. You can select and deselect any of the source images and see the results immediately in the dialogue box much like the “merge to HDR” arrangement in Photoshop CS5. 

The Brackeeter screen grab shows four source images on the left that were used to produce the result in the preview window using the software’s default settings listed on the right. The final image contains a pleasing balance between the interior and exterior lighting along with believable shadow and highlight details and all without any oversaturated colors. The illustration containing three versions of the same interior compares a single exposure (top frame) in which the bright areas outside the window are washed out because the exposure is based on the weaker interior lighting. Whereas, the results of Brackeeter EF processing (middle frame) and Photomatix HDR processing (bottom frame) both capture highlight details outside the window but with very different renderings of colors and shadow details. The single room image is another example of how EF processing can yield a very even lighting balance between window light and interior lighting by managing the wide dynamic range.  

Brackeeter is for Mac OS X 10.4 or later and Photomatix is available for both Mac and Windows platforms.  A search using the key phrase “Exposure Fusion” will turn up a few other sources. High Dynamic Range techniques and the Exposure Fusion approach are two of the topics covered in my latest Lark book: “Advanced Imaging” to be released in April 2011.