By Uwe and Bettina Steinmueller
Photographers are painting with light. Sounds good, right? But if you listen to them, you will learn that the light is hardly ever right for them. Why are we photographers so picky? We have to deal with three different kinds of contrast range:
Contrast range of the scene
Contrast that can be captured in the camera (negative film, slide film, or digital)
Contrast that can be printed
Most fine art photographers care most to produce photos that can be presented in prints. The first barrier we have to master is to handle the contrast of a scene and capture it with our camera. Studio photographers produce their own light and can create a contrast that can be fully handled by their photographic medium (film, digital). Landscape photographers don't have this luxury and need to hope/wait for the right light. If you listen to them, the right light of course hardly exists. Studio photographers have a hard time producing the "ideal" soft, yet contrast-revealing light (don't forget the coloring of the light). Now imagine being as picky in the outdoors. That's why many landscape photographers get up early (not all of us love this part ) or photograph late in the evening.
What happens if the scene has a larger contrast range than the camera can capture in a single shot? There will be clipped light in the highlights and/or the shadows. Somehow, lost highlights are the worst because the highlight details are lost forever and will show in a print as paper white. We try like hell to avoid clipping the highlights. This may sometimes force us to underexpose the main content of the photo. Underexposure leads to blocked and noisy shadows. The most efficient solution to this problem is to wait for a better light. We natural light photographers need to be very patient (and, of course, not all of us are). There are also some techniques used in the field that help to tame the contrast. One of the solutions is a gradient neutral density filter. They work fine, but unfortunately the boundaries between the bright areas and the darker ones are not that often just split by a line.
In the end, many of us photographers don't even start taking photos when we think the light is not right. In California, this is most of the day. Also, low contrast (in the extremes like fog) is challenging, but low-contrast photos can be much better handled than clipped highlights and shadows.
Capturing more dynamic range
The desire to capture more dynamic range is not new. Black-and-white photographers captured quite a bit more dynamic range and also knew how to handle the film to tame the contrast. Slide color photography is more restricted, and most digital cameras today behave like slide film. As said, if the highlights are clipped, you have actually captured an image that will hardly be good enough to be used for any fine art print. An old classic solution is to capture multiple exposures (film or digital) and blend these images together. Unfortunately, this is both labor intensive and error prone. Here are some of the challenges:
How to make sure nothing moves (neither the camera nor anything inside the scene)
How to blend (masking)
It is a classic problem that if you cut something into pieces (here, exposures), it is always hard to get back to the original.
In the digital world, there is some hope that in the future our cameras may be able to capture better dynamic range (Fuji made a first step). Of course, until then we won't stop taking pictures.
Our personal road to HDR
We tried in the past some manual blending of exposures and were hardly impressed by our results:
Manual registration of the photos was a pain.
Manual creation of the blending mask was tedious and unpleasing.
Automatic tools did not make us really happy.
At some point we looked into using the Photomatix tonemapping plug-in (Details Enhancer) for single exposures and we started to like some of the results.
"Light in the Dark" Portfolio - Sample workflow #1
Our experience with Details Enhancer made us curious how HDR would work in Photomatix and we revisited some photos that we took two years earlier at Fort Point in San Francisco. These were three-shot sequences (unfortunately only shot at −1, 0, +1 EV; today we would do −2, 0, +2 EV instead). Image ❶ shows the exposures we took.
We call this portfolio "Light in the Dark" because the shots have in common a dark scene that is only lit by light through a window or opening. They also have in common that we hardly have to deal with moving objects. Still, a lot of attention has to be paid to camera movement. It sounds actually easier than it is.
Even on a sturdy tripod, the camera shake by the mirror (yes, we enable the mirror lockup feature but still it has to go up for each exposure) and the shutter will cause small movements. Be careful if the floor is not sturdy enough. It's less of a problem with medium- to wide-angle lenses, but it can be shocking with telephoto lenses. Don't forget to use a remote release and avoid any contact with your camera during all of these exposures.
If we capture multiple exposures, we only use the camera in manual mode. First we decide on the aperture and then find the medium shutter speed for the exposure. Finally, we either use three (-2, 0, +2 EV) or five (-4, −2, 0, +2, +4 EV) exposures using the camera's auto-bracketing mode. It's too bad, but using auto-bracketing mode can sometimes be a problem because some (most?) cameras limit the longest exposure in this mode to 30 seconds and this is easily a limitation for the +4 EV shots. We plan to explore some more advanced programmable remote release for future work.
Once you have your multiple exposures, your fist step is to combine the images. There are two principle ways:
Exposure blending
Generating an HDR image
For us, exposure blending is not the solution as it already compresses the tonal range and we plan to do this via a choice of tonemapping operators. So we first create an HDR image. This can be done in Photoshop, but we are happier using Photomatix Pro. In the following sections, we'll show the steps we use to get our final image.
Generating an HDR image(❸): In Photomatix, you can either open the three files directly (we use, of course, RAW files) or convert them in the RAW converter of your choice (Adobe Camera Raw, LightZone, or other tools). We open the RAW files directly.
Note: The selected files are RAW files even if they have a .tif extension. The Canon 1Ds saved the RAW files using the .tif extension. This was not a good idea and Canon fixed it in later cameras.
Photomatix is quite slow converting the RAW files but it does the job. Once Photomatix finished the conversion, it shows dialog.(❸)
We always check the Align Source Images... option. Photomatix does a nice job for minor alignment problems (mainly vertical and horizontal movements). After this step, Photomatix Pro creates an HDR image.
Don't be too shocked if the HDR image looks like this (❹)on the screen in Photomatix.
This looks quite horrible on normal monitors, right? Everything is actually fine. Remember that HDR files can store way more dynamic range than your monitor or any printer can handle. The HDR image needs to be tonemapped to fit into the dynamic range that our normal output devices can handle. For now, we save the HDR image in either Radiance RGBE (.hdr) or OpenEXR (.exr). This is only needed if we plan to perform other tone mappings in the future or use different tools (e.g., Photoshop).
The next step is tone mapping(➀). Photomatix currently supports two different tonemapping algorithms:
Details Enhancer
Tone Compressor
We almost always use the Details Enhancer because it is a local operator and helps to create better local contrast/detail.
The Details Enhancer in Photomatix is quite a complex control. Its use requires some experience as well as some trial and error.
Note: Because Details Enhancer is a local operator, the result depends a lot on the image. With some images it works great, and with other images not so great.
Also, with this tool we try to avoid clipping at nearly any cost. The settings we show can be a good start for your own images. We expect some more intuitive tone-mapping operators with future versions of Photomatix.
We save it as 16-bit TIFF file(➁). You will find it too soft and you are right. We intentionally want the image to be soft and with the lowest amount of clipping at this point. We will enhance the contrast in further editing. It is very hard to soften files that show too strong a contrast, and as mentioned earlier, clipping cannot be undone at all (in 8-/16-bit files). For the final editing we use LightZone and Photoshop CS2/CS3. We consider the result from Photomatix Pro to be like the result from any RAW converter. We always want some sort of soft image from the RAW converter, and then we fine-tune it with Light-Zone and Photoshop.
Note: While we work in Photomatix, we are not really working color managed. We try to get pleasing colors and fine-tune them later.
Here is one possible interpretation we got from this file (we use a technique we call "colorizing" that blends a black-and-white and color version of the image).
"Light in the Dark" Portfolio - Sample workflow #2
The timing could not have been better. Just shortly after we got a first very nice result from HDR images (see workflow #1), we had a unique photo workshop on Alcatraz. Now we are certain that we would have missed many Alcatraz shots without using HDRI. This time, we again used three shots but at a wider range, −2, 0, +2 EV.
Already the previews of the three RAW files show that these three shots were good enough to capture the scene contrast. How's that? The underexposed shot (−2 EVs) shows no clipping in the highlights and the overexposed (+2 EVs) does not show any blocked shadows.
The procedure was exactly the same as in our fist example.
The main challenge with this and other color images is to define the color temperature. There is no right way to do it because we want to show the warmth of the light but also not be over the top. In the end, we most often like more muted colors best.
This photo also shows some ghosting. Moved by the air, the transparent tape moved slightly during the different shots. We think in this case it's not a problem, and it may even add a nice touch to the image. We just wanted to mention this and warn readers that ghosting is lurking everywhere.
We shot a full portfolio on Alcatraz but would have found the workshop worth it for this shot alone. Since then, multiple exposures are a fixed element in our photographic tool set.
"Light in the Dark" Portfolio - Sample workflow #3
In November 2006, we conducted our fourth annual "Fine Art Photography Summit" in Page, Arizona. Page is most famous for its slot canyons: Lower and Upper Antelope Canyon. Down in the canyon, you have no direct light. There are two light sources:
Direct sunlight bounces off the red rocks, and while losing intensity, it gets more and more reddish
Light from the blue sky (also called air light) supplies other areas in the canyon with a very blue light
In this case, what your camera records is closer to the truth than what you can capture with your eyes. This means that the final images have kinds of made-up colors while still capturing real color (real, more in a technical sense than to our eyes). Our eyes (and brain) try to neutralize these colors in the canyon.
This time the contrast is so high that using five exposures seems better than just using three.
As you can see, the most underexposed shot (#2) barely gets away with not showing any clipping (but is clearly good enough). On the other end, the most overexposed photo (#5) is the only one that does not show blocked shadows. Again, we think we captured the essential contrast with these five shots (three would not have been sufficient).
These five shots again get combined with HDR with Photomatix.
This time, we worked with selective white balance in LightZone to enhance the effect by bouncing sunlight and air light. We made some area of the image warmer in terms of white balance and other areas cooler.
"Light in the Dark" Portfolio - Sample workflow #4
Recently we wanted to see how well cascades would work using HDRI techniques. The classic solution is to use longer exposures (we like about 1/5 to 1/15 of a second for high-speed cascades. But of course it depends upon your artistic goals; that is, how you would like to present the cascades. In most cases, it can be said that completely freezing the moving water does not add to the picture.
So we started photographing cascades with multiple exposures.
We experimented with bracketing of about (-1, 0, +1 EV).
Note: All shots were done in the shadows. Still, you will be surprised by the dynamic range of that scene.
Remember that you also have bracketed single shots. The first thing to do is to check whether the single shots can be used by themselves. Bracketing was always a proven technique to get the best possible exposures. Now you can use this technique and get shots for HDRI at the same time.
We combined the three exposures in Photomatix (version 2.4) to an HDR image and then tonemapped the HDR file in Photomatix.
Again our goal is to keep the file soft. The final tuning is done this time in Photoshop CS3. Here is the version right out of Photomatix.
We brought this low-contrast image to life using the following steps:
Additional local contrast enhancing with Akvis Enhancer
Sharpening with DOP EasyS Plus
Some extra contrast and saturation tuning Here is the version we finally came up with.
It makes some sense to use HDRI even for moving objects like water and clouds. Otherwise, moving objects will cause a lot of trouble.
Summary
HDRI is still a new technique, but there are many photos that can be improved using multiple exposures and tone mapping. We expect in the future cameras that help capture higher dynamic range in a single shot and improved implementations to remove ghosting. We personally hope for many more ways to perform tone mapping, as well as ways that offer better control. HDRI is clearly here to stay, so we should all start experimenting with it now.