Simply put, dynamic range is the range of brightness values, from shadow to highlight, in a given scene. Cameras don’t have nearly the dynamic range of the human eye. When a scene is high in contrast, the camera can’t record all the disparate brightness levels. Either the shadows will be recorded as pure black or the highlights will be recorded as pure white, or both.
A solution is to take a series of bracketed exposures at different exposure values capturing the extremes of shadows and highlights.
Shown below is the set of source exposures that were combined to generate the final image shown at the start of this post.
The simplest way to combine the differently exposed frames is to stack the exposures in Photoshop and use layer masking to hide/reveal certain parts of each image. This is the manual approach.
Two automated approaches also exist. The first automated approach combines exposures to create an intermediate 32-bit HDR image which is then “tone-mapped” into a low dynamic range image. Tools such as Photomatix use this kind of HDR processing. These tools are quite popular for creating other-worldly, surrealistic images. But for commercial work, the results can be unrealistic or in some other way unacceptable.
A second automated approach is known as exposure fusion. The fusion software employs algorithms to analyze and select which pixels to use from each of the source images to create a fused result. The product I use for exposure fusion is called Enfuse.
I usually use a combination of exposure fusion with the manual approach. That is, first I’ll create a fused image from the series of differently exposed frames. Then I’ll select the most-correctly-exposed individual frame as a basis for the final image. Starting with this single exposure, I’ll then blend in (with layer masking) the fused image to reveal details in highlights and/or shadows that the single exposure lacks.
This usually involves trial and error, and patience.