naviquan

High Dynamic Range imaging (HDRI or simply HDR) has become recently a very popular technique of post-processing photos to enhance them and achieve a desired artistic effect. It attempts to deal with the fact that photographic pictures are not as powerful as human eye in displaying both dark and light objects in photos since photo cameras have a limited dynamic range (see "What is Dynamic Range" text box). I guess one of the reasons for the proliferation of HRD is the easiness of handling digital images from the moment you use your digital camera to take a picture, its consequent post-processing with software tools and printing on an ink-printer. This article is the first in a series dedicated to HDR photography describing general HDR concepts and HDR processing using Photoshop.

Doing HRD with
Photoshop

Image exposure EV -2.64

Image exposure EV +2.64

Image exposure EV 0

Immediate result of the
HDR merge

HDR merged image after
additional adjustments
(contrast enhancement and
warming to match the
original EV 0 image)

Although there had been a number of attempts to achieve HDR, nowadays, one specific technique became predominant with digital photography (and the term HDR is used almost exclusively to refer to this technique). It consists of using several photographs taken with various exposures and produce a single high dynamic range image by combining them into one.

For example, let's say you have three photos: one taken with regular exposure, and the other two with exposure compensation of -2 and +2. Obviously, the "+2" picture will have highlights blown up, but it will show more details in shadows. On the contrary, the "-2" picture will have no shadow details, but highlights will have some color data. The idea of combining the three pictures is to "borrow" highlights from the underexposed photo and shadow details from the overexposed photo. So, by adding those pieces to respective areas of the regular photo, we can hopefully enhance it to show not only mid-tone details, but highlights and shadow details as well. Although the idea is simple, its implementation is not. You will need a specialized software and enough processing power to compute all the pixels of the resulting picture based on three originals.

Unfortunately, there is one more problem that makes the idea even more difficult to implement. This problem is how to show the picture you have just synthesized. Indeed, even if the resulting picture (which is nothing more than just a file with picture data) has now enough data in light and dark areas, the monitors (CRTs, LCDs) as well as prints we use to render such data are not capable of doing this (in other words, they have their own DR limitations)! Obviously, there is no real solution to this problem, but, as always, there are methods to alleviate it. All these methods rely on what is called "tone mapping."

The word "mapping" here relates to modifying picture data that are not reproducible by the output device in such a way, that it becomes visible to human eye. There are many specific ways of doing it (and that's why there are many HDR software packages). For our purposes, it is suffice to say that almost all of them rely on "compressing" range of data in the HDR file into a shorter one without clipping the ends. The net result is a kind of faked high dynamic range - you can see shadow and highlight details on a picture that in reality (roughly speaking) posses the same dynamic range qualities as any other picture created with a digital camera.

The latter is the origin of a great deal of controversy surrounding HDR which often results in heated discussions where one side, pro-HDR, tries to elevate the technique to some kind of the future of digital photography. Whereas the other side tries to prove that HDR is useless if not harmful (see our "HDR Battle" text box). As often the case, the truth is always in the middle. HDR is just a technique - if used correctly, it can at times significantly contribute to your work and convey your artistic ideas to the audience. If not used correctly, the result can be horrible.

So, what are the HDR pros and cons? I think that apart from some general concepts (see the "HDR Pros and Cons" text box), it all depends on what you want and need. For example, if you are working for a busy real estate publication selling residential homes, you probably are not going to use HDR for homes in the low price range (and quite well use it for $10+ million dollar homes). However, if you a digital artists selling prints, it is quite possible that you will want to spend extra time playing with HDR to see how it helps you transform a picture in an avant-garde piece of art.

For or against it, I can tell that for many photographers HDR is a source of fun. Equally, HDR galleries of popular sites like PBase, Flickr and others are very popular. I should also say, HDR can be very addictive for both authors and viewers. Like almost everything, if well crafted, HDR can be wonderful. If not, it can become cartoonish and surreal. Anyway, instead of talking and arguing, why not to try HDR and see if you like it, and it works for you?

Let's do it together and try to go through HDR steps using various software to compare results and the convenience of the workflow process. To start we need images, and this is the first step in the HDR technique. Usually, at least three images are used; although depending on specific conditions it could be a good idea to use more. In fact, in some cases five, seven and even more images can be used.

Since all we need to do is to explore the HDR techniques, as a source of my pictures I decided to use our family room with big windows that presents a classic case for using HDR. I used Canon 1D Mark III with EF17-40mm f/4L lens. Pictures were taken at focal length 17mm, f/4.5, auto white balance, and ISO 200 as large (L) JPG to avoid additional step of converting RAW files (I have to note though that some HDR software packages can take RAW files directly). The first image was taken with the exposure of 0.3 (EV 0), the second with 1/50 (EV -2.64) and the third with 0.8 (EV +2.68).

It was a rather gloomy day at around 5:20 pm, I did not try to pursue any artistic accomplishments but rather to illustrate the essence of the HDR and results that can be obtained by using it. The only real goal to achieve was a faithful reproduction of the room and view from its windows as I remember it at the time of shooting. Nothing exiting but quite indicative for the HDR exploration.

Top image has regular
exposure (before), the
bottom image is the result
of HDR merge of 3 picture
in PS CS3 with some
additional contrast and tone
adjustments (after).

The most convenient way of taking images for HDR is to use what is called AEB - Auto Exposure Bracketing. If you never used it, check your camera manual: chances are that you have this feature. It provides a convenient way of taking almost identical pictures where the aperture, focus, and white balance are locked and the only difference is the exposure (also, if you have any automatic image enhancements turned on in your camera, you need to disable them). If you do not have AEB, you can still try to use the HDR technique, but understandably it will be more difficult.

Now it's time to try our first HDR image. If you use digital cameras, there may be a chance that you have Photoshop (where HDR images are supported since CS2). If you have it and never tried HDR, it's time now. In Photoshop CS3 use the following menu options: File -> Automate -> Merge to HDR. Select file using the Browse button and click on OK. The time needed to process images depends on the size of your images and your computer; on mine it took about 20 seconds to process three 3888x2592 pictures before you will see a kind of preview window. You can disable here source images to see their contribution to the resulting image and do some other adjustments. If you want to create a real HDR image, you have to use 32-bit/channel bit depth and later on (when the image is generated) to do tone mapping by using Image -> Mode -> 16Bits/Channel... options of the main menu.

And that's what pros will do. However, since our goal is a "simplified" HDR, we are going to take a shortcut and select 16-bit/channel bit depth option here and click on OK. This will result in what is called sometimes exposure blending which in many respects is significantly different from "real" HDR. However, for the purposes of preserving details in shadows and highlights, exposure blending can provide good enough results.

On my machine it takes another 5 - 7 seconds to complete the conversion process, and you will see an HDR merged picture. You will also have a dialog box to set the conversion method (Exposure and Gamma, Highlight Compression, Equalize Histogram, and Local Adaptation). Some conversion methods allow for tuning of additional parameters - play with them to see what you like. The most sophisticated conversion method is Local Adaptation which allows to fine tune your results with a greater degree of control. There is bog problem though with Local Adaptation - you really need to know how to use it which may take time to learn and master. Again, since our goal is a "simplified" HDR, for my images I selected Highlight Compression (it simply compresses the highlight values in the HDR image so they fall within the luminance values range of the 16-bits-per-channel image file), and here we are: the result of the HDR merge.

Let's examine the image. It's rather dull and flat, but, indeed, it has details in highlights and shades previously absent on the photo with regular exposure. Using Photoshop let's do some additional quick adjustments to increase the contrast and add the warmth to the image to bring it closer to the original typical for Canon's 1D output. The result is the HDR merged image after additional adjustments. You can also see side by side comparison on left side.

If you have an old version of Photoshop without a direct support for HDR as a menu option, it is still possible to merge several images using layers and various blending techniques with the result similar to the one obtained above (in other words, no real HDR). However, it goes beyond the scope of this article. One of the reasons is that there are plenty of affordable software tools (at reasonable prices plus open source and freeware tools) you can use for HDR photography. I will review some of them in next installments of this article. Stay tuned!

Some Popular HDR Software Tools

Part II - High Dynamic Range Photography - Photomatix Pro
Part III - High Dynamic Range Photography - "HDR Look" and Pseudo HDR vs. HDR
Part IV - High Dynamic Range Photography - Qtpfsgui