The blending of multiple exposures, with offset exposure values, of the same scene to create a single image with a wider tonal range than is possible with a single exposure.
The modern thinking is that HDR was initially pioneered and developed in recent years for use with digital images and computer-generated graphics. While techniques and technology have developed greatly in this age, the pioneering of high dynamic range images – using more than one exposure to increase the tonal range of a print – goes way-way back.
Jean-Baptiste Gustave Le Gray became famous for his seascape photographs 160 years ago. Le Gray used two negatives, one for the sky and a second longer exposure for the sea, to produce one positive print with a much higher luminosity, or dynamic range, than was possible with only one exposure – back in the 1850′s!
The technique further developed from the 1930′s into layering films of different exposure values to create the higher dynamic range, dogging and burning techniques were introduced to the process in the 1950′s. Now here in this digital age there is truly nothing new under the sun.
My personal preference for HDR images is realism, (not a fan of the overdone grunge look) where HDR technique is another tool at the photographer’s disposal. When executed carefully, HDR makes for some striking images – photos with the “wow” factor. In point of fact, HDR technique is more widely used by professional photographers than most people realise.
There are many software solutions now-a-days to aid in the combining of multiple exposures to create one HDR image. I use Photomatix Pro, which allows exporting the files directly from Lightroom 3.6 and reimporting the tone mapped combination image back into LR. I have tried the Photoshop CS5 merge to HDR, maybe I need to spend more time learning this particular tool, but not a fan at this stage.
The common HDR technique is to bracket two or three images with 2EV differential between each sequential exposure, then tone map these images in the software of choice. Of course the number of images and exposure offset value is subjective and will vary by photographer, equipment and scene. As an example, the images that follow of Mount Ruapehu on the Desert Road are unedited JPEG conversions of the original raw NEF files. The first image is camera exposure setting in Aperture Priority mode, the second image is -2EV from this setting while the third image is +2EV from the initial setting, each exposing for detail in a different area of the image:
These were exported into Photomatix Pro for tone mapping, then imported back into Lightroom 3.6 as a .TIFF format for final sharpening and colour adjustments. The resultant tone mapped image received a levels layer in CS5 to produce the final image, which worked very well in this case because the evening was still with no breeze and therefore no foliage movement:
This approach of layering several different images has obvious draw-backs for landscape photography. If there is a light breeze for example ghosting appears around leaves and grass and any moveable object due to the different position of the object(s) in each exposure, which is particularly nasty against a sky or a highlight area. There is also a chance of camera movement and misalignment between images. Bear in mind also, I am referring to a final image of quality to print to 24″+ wide (on quality print paper) where ghosting and fringing are all the more obvious, rather than a screen presentation image where a lesser quality image can still be made to look reasonably good.
When referring to image quality, a screen image of smaller proportions that looks “wow” what a great image may not be that great an image. Every image I publish, I know intimately. Scrutinize at 100% or even 200%, which is, or should be, standard practice I believe. Look for quality of detail, sharpness, fringing, moire, chromatic abberation, noise etc. You will often find the smaller screen image does not present that well under close examination, ergo printing to large sizes is out of the question.
Shooting RAW is also standard practice for me, allowing the maximum chance of maximum quality. Adjust the images before exporting to HDR software, eg. black clipping, white balance etc. Highlight recovery is ok for a small amount but it tends to leave a halo around leaves, branches etc that are against the high lit sky or background, which is accentuated more during the HDR process.
If it is possible to balance an image exposure such that the camera histogram is not clipped or only slightly clipped, indicating that all the detail in the image is retrievable, there is the opportunity to create a finely detailed, noiseless HDR image from the one exposure.
There are a number of ways to create a second, exposure adjusted image from an original. I have found that exposure adjusted copies of the one image, saved in Capture NX 2 do not align perfectly when layered for HDR processing – not sure about the latest update though. Lightroom 3 can create a virtual copy of any image, which allows editing separately and apart from the originating image. Both the original image and the virtual copy can then be exported to Photomatix Pro for tone mapping, returning one HDR from the one image, and the layer alignment is always perfect.
Because both images are essentially the same image, detail is exactly the same for both images so when they are layered for tone mapping they are an exact match. Details remain crisp and noise is not introduced from the usual underexposed bracketed images used for HDR tone mapping.
In the following images of the Wanganui River, once again the first photo is a JPEG conversion of the original raw file with some black clipping applied. The second image is the exposure adjusted virtual copy of the original image by -2EV, and the final image is the returned tone mapped final HDR. The -2EV exposure adjusted image is much the same image that would be produced if I bracketed the shots. The important difference is that if this were a bracketed separate exposure, it would be very noisy in the dark areas which is the majority of the image, but because the image I am using is an exposure adjusted copy of a correctly exposed image there is very little or no noise and this obviously means finer detail in the resultant HDR.
A virtual copy of the above image, exposure adjusted by -2EV:
Both the original and virtual copies are exported from Lightroom to Photomatix Pro, to produce the following resultant HDR:
Finally a few more examples of (in my opinion) effective HDR both herein and elsewhere. Each image links back to the original: