SONY A77: expanding dynamic range with DRO or HDR (or not)1/31/2012 06:07:00 PM
I work near my kitchen (for obvious reasons). As I look to my right from where I am sitting, I see dark shadows down around the cabinets und...
I work near my kitchen (for obvious reasons). As I look to my right from where I am sitting, I see dark shadows down around the cabinets under the counter, and I also see through the big windows to the bright outdoors full of trees, cars, other houses. This is a scene with a high dynamic range. The miraculous thing is, my eyes can see detail in the shadows and in the bright areas equally well and at the same time. Pretty amazing, really.
No camera can match that. Photographing this scene the "old-fashioned way" (say, in M mode), I would have to expose for the shadows (which will result in the windows appearing to be bright white areas with little detail beyond them) or expose for what I can see outdoors (which will mean the shadows inside will become black areas, again, with little detail). It's a dilemma. What good photographers have done for ages is learn how to make the best compromise, so that we get some detail in the highlights and some detail in the shadows, or so that we expose for the part of the scene that matters most to us and simply live with the fact that details at the other end of the light-dark scale are sacrificed.
Note that the problem I'm talking about exists only when there's a large contrast between bright and the dark areas of the scene you are photographing. This isn't always the case, in fact, it isn't often the case. Take a photo in the shade of a tree on an afternoon when the sun is hidden behind light clouds, and you'll probably be dealing with a very modest dynamic range, one that is well within the reach of your camera's sensor. And when dynamic range is a problem, one way to fix it is by balancing the light. You might for example put a flash on the camera, to throw some light into the face of your subject when you're shooting outdoors. These are solutions that don't rely on your camera at all.
|Raw file, unprocessed.|
|Raw file in Lightroom with "show clipping" turned on.|
Building a better mousetrap, er, light-trap
Now, camera makers want to make life easier for us all and they have, for a long time, been trying to solve this problem in the camera, in two different and somewhat contradictory ways. First, makers of sensors like Sony have been working to expand the dynamic range of the sensors. Second, camera makers have been working to squeeze as much dynamic range into their captures as possible. Either way, the goal is to get a more accurate picture of the world, a picture that is — or at least seems to be — closer to what our eyes really see.
Sony is undoubtedly the leader in this area — both in its sensor development and in the work it has done with in-camera processing — and the current state of the art is available on Sony's new DSLT cameras, the A65 and the A77. I'm not going to talk about improving the sensor because the capabilities of any given camera's sensor are a given and you can't get a better sensor without buying a new camera. I am going to talk a little about improvements in how these cameras process images for you.
One way to improve the overall exposure is for the camera to do some quick adjustment of the lighting at the moment of capture. To do this on the Sony A65 or A77, you use the DRO (dynamic range optimization) feature. DRO tries to squeeze more contrast into a single file. Here's a picture of my kitchen with DRO set to "Auto".
Sensing that there was at least a five to six stop gap between the darks and the lights here, I then changed the DRO setting from Auto to "Lv5":
Not a big improvement. The shadows have been lightened a good bit — but the bright areas outside have, too. This isn't worse than the DRO AUTO shot but it's not better either.
In-camera HDR ("high dynamic range") is a different way to solve the problem. Like on-the-computer HDR, in-camera HDR starts with several different exposures of the same scene, then combines them into a single output file in which the well-exposed bright areas from one shot have been combined with the best-exposed dark areas from another, and the composite file has been adjusted to make things look natural. Sony's programmers have written programs that seem to do a very good job — sometimes — of combining the exposures. But a key factor in getting good results, is providing the processor with good source images. The new fixed-mirror (SLT) cameras from Sony are especially well suited to gathering the multiple exposures because, lacking a moving mirror, these cameras can take more shots per second than their traditional reflex (moving) mirror competitors. All of the pictures here were taken hand-held. Computer-based HDR is usually done with a tripod.
When you shoot with the HDR feature enabled, be prepared for the camera to respond differently. Instead of the single shutter noise, you'll hear several shutter noises in rapid succession. And then it takes several seconds for the camera's processor to create the resulting HDR file.
Here's a shot with HDR set to "Auto".
Now, you don't have to use either DRO or HDR. As I said above, you can take steps to reduce the dynamic range, usually by lightening the shadows. For instance, I could have used flash here, or I could have turned the kitchen lights on. You can also shoot raw and fix problems on the computer. This is what a lot of very good photographers do, and it works. The raw data file has a lot more latitude than a jpeg. To compare what I can get from the raw file with what Sony's in-camera DRO and HDR offer, I reshot the scene, saved the raw file, and processed it myself in Lightroom.
I used an adjustment brush to bring the bright areas (the windows) down 1.5 stops, and a separate brush to bring the shadow areas up 1.5 stops. I wasn't doing this for a client so I was content to do this quick and dirty. Even so, the result was pretty good:
|Raw file in Lightroom after adjustment brushes have darkened the highlights and lightened the shadows.|
What's best: DRO, or HDR, or Raw?
So I ended up with three pretty good or at least okay results, achieved in three very different ways.
I'm tempted to say that in-camera HDR is "easy" but that's not quite true. It would be if I could use HDR AUTO, but as I showed above, this produces a result not much better than DRO AUTO. To get the best result with in-camera HDR, I have to calculate (or guess) by what amount the dynamic range of the scene exceeds the capacity of the sensor, and set the HDR level accordingly. This isn't that hard but you can end up making a couple stabs at the shot before you get it right. May still take less time than working on the raw file in Lightroom. Or it might not.
Morever, because in-camera HDR takes multiple exposures and then processes them, achieving a single HDR result in the camera takes about five or six seconds. And you simply can't use it if anything in the scene is moving quickly. Finally, the A77's processor saves the HDR file as a jpeg, by necessity. The HDR file is a composite, a processed result. There is no raw original of the HDR result.
If you really, really want the very best result obtainable, you can use computer-based HDR. This allows you to shoot several bracketed exposures, save them as raw files, then merge them in Photoshop or an HDR program. But that's a lot more work than I went to fixing my one raw file.