|1/60 sec||1/125 sec||1/250 sec|
At 1/60, you can see severe motion blur in the dancers at the left who are bowing and slight blur in the dancer who is standing still but raising her arms (click through to see a larger version). At 1/125 there is blur in the hands, right leg and face of the running dancer (but not in the planted left foot). At 1/250 there is blur in the dancer who has just landed a jump. You can see it in the closeup crops below: the face has motion blur (because the whole body is still moving downwards) but the right foot is very sharp (because its motion has been stopped by the floor):
|Crop of 1/250 sec (motion blur)||Crop of 1/250 sec (sharp)|
Now let's see some photos where motion blur is not a problem:
|Motionless pose at 1/125||Crop of Motionless at 1/125||Fast Motion at 1/400|
|Jump at 1/500||Crop of Jump at 1/500||Another Crop of Jump at 1/500|
What can you can do to minimize subject motion blur?
|Jump at 1/125||Crop (top of jump; sharp)||Crop (bottom of jump; blur)|
|Pan at 1/40||Crop of pan at 1/40||Crop of pan at 1/40|
|1/160 sec||Crop of 1/160||Crop of 1/160|
|1/320 sec (photo by Arsky)||Crop of 1/320||Crop of 1/320|
Hand-held: tucked in
As a rough rule, you can expect to hand-hold a lens without noticeable shake at a shutter speed at least as fast as the reciprocal of the focal length. For example, with a 200mm lens, this rule says you should be shooting at 1/200 sec or faster. Assuming you have a half-frame sensor and/or you expect to crop the image, this will make the image relatively larger, thus multiplying the effects of shake. To be safe then, the rough rule would be to double that figure to 1/400 sec. On the other hand, image stabilization in the lens coupled with good technique (as outlined above) can allow you to shoot three or four stops slower (1/50 to 1/25 sec.). In the end, for dance photography, if you have image-stabilization and good technique you are more likely to suffer from motion blur than from camera shake.
If the subject does stay still, with good hand-held technique you can shoot at very slow shutter speeds (click through to see the detail):
|1/15 sec at 200mm||1/10 sec at 200mm||1/4 sec at 135mm|
Focus is an artistic decision; it is usually bad to have the main subject out of focus, but it can also be bad to have too much distracting background in focus.
The first thing you have to be aware of is the width of the focal plane--how much of your picture will be in focus. This is called the depth of field, and it depends on four things: the distance to your subject, your aperture setting, the focal length of the lens, and your degree of fussiness or scrutiny. The chart below left shows some sample depths of field for different combinations of distance, aperture, and focal length (with average scrutiny). On the right is a table of the three factors, and a diagram where the focal plane is centered on the middle butterfly, and everything within the depth of field is in focus, while the near and far butterflies are outside the depth of field and are out of focus.
For example, if you have a 200mm lens set at f/2.8 and you sit 20 feet from the front of the stage, then your depth of field is just 4 inches for a dancer all the way downstage, and 1 foot 3 inches for a dancer 40 feet away upstage.
What does this all mean? It means you have to be careful with the 200mm portraits, but needn't worry much about the 50mm shots. By careful, I mean taking care with where you place the focus selector point (the little dot or rectangle in the viewfinder--on most cameras it lights up red). Suppose you want a portrait of a downstage dancer who is in a pose with her head leaning backwards. If the focus point is somewhere on her torso, then her face will not be in sharp focus. What can you do? You could change your aperture to f/5.6 (or maybe even f/8), giving you more depth of field. Or you could zoom out to 100mm, in which case the depth of field is about 2 feet 7 inches and you'll be fine although you get a wide shot, not a portrait. (I should note at this point that a wide angle lens only gives you more depth of field if you consider the whole shot. If you crop the 24mm shot so that a dancer occupies the same portion of the frame compared to a 200mm shot, then the 24mm and 200mm will have the same depth of field. See, for example, Michael Reichmann's explanation of this.)
But the best solution is to place the focus point exactly where you want it. Your camera has a controller (dial/buttons/joystick) that allows you to select one of several focus points. The image below replicates the view through the viewfinder. I used the joystick to select the top focus point (the red rectangle) and placed it over the dancer's eye while half-pressing the shutter. I waited for a pause in her motion (because this scene is rather dark and I only had a shutter speed of 1/125 sec) and then fully-pressed the shutter to capture this shot.
|Focus point on eye (f/2.8, 1/125, ISO 1600, 200mm, distance about 20 feet)|
You can see that the depth of field is shallow: the dancers in the background are very blurred, and even the main subject's fingernails, which are only about a foot in front of the focal plane, are somewhat blurred. If I had left the focus point on the default center position (which you can see is on the side of the neck), I would not have gotten the face in sharp focus.
Here are some reasons why the autofocus system can miss, and things you can do to minimize the problems:
Consider the "Focus point on eye" image above. It was taken at f/2.8, 1/125 sec, ISO 1600 and is well-exposed. As we will see in the next section, If we took it at f/11, 1/500 sec we would have to use ISO 102,400. Some cameras go that high, but the results will be extremely noisy. So we have a never-ending trade-off between noise, motion blur, and depth of field. Sometimes there is plenty of light and you have nothing to worry about; sometimes a scene is dark and you must decide which factor to compromise on.
Why are ISO 102,400 pictures noisy? And what is ISO, anyways? ISO is the Internatinal Organization for Standardization, and in 1979 they set up a scale to standardize film speeds; that scale has been transfered over to digital. You can think of ISO 100 as the base rate, and every doubling of the ISO number (to ISO 200, ISO 400, ISO 800, etc.) as telling the camera to count each photon of light twice. So in a perfect world, a photo of a still subject at 1/500 sec at ISO 400 would look the same as a photo at 1/125 sec at ISO 100 (the former is 1/4 the time and counts each photon 4 times). But we don't live in a perfect world; we live in a random quantum world in which there is always fluctuation in the amount of light gathered. At ISO 102,400 each little fluctuation is magnified by a factor of 1024, making them very visible.
With my full-frame Canon 5D II camera, I'm confident starting out at ISO 1600, but if the light changes to a brighter scene I take advantage of it and switch the ISO to 1250 or 800. In dimly-lit scenes I will switch to ISO 3200, but I try not to go beyond that. Here are some real-world examples--they look pretty good even with the crops, but if you click through to the even-larger versions, you can see noise in the 3200 and 5000 shots.
|ISO 1600||ISO 3200||ISO 5000|
|Crop of ISO 1600||Crop of ISO 3200||Crop of ISO 5000|
The thing that bothers me most about my camera is that I need a two-touch sequence to change ISO: press a button and turn a dial. I wish my camera had three dedicated dials: one for aperture, one for ISO, and one for exposure compensation (in aperture priority mode) or for shutter speed (in manual mode).
Another option is auto-ISO. I haven't used this much because I find it guesses wrong too often, but maybe I'm not using it right. Newer cameras seem to do a better job; for example the Canon 5D III and many Nikon cameras allow you to set the minimum and maximum ISO values for auto-ISO; I will experiment more with this and might grow to rely on it.
The web site DxOMark tests cameras and reports the highest ISO speed at which cameras continue to deliver excellent picture quality. Here are some of their ratings, to which I've added the sensor area and approximate price:
|1800||Canon 5D II||full-frame||$2200|
|1200||Nikon D5100 / D7000||1/2 frame||$900|
|1150||Pentax K5||1/2 frame||$1300|
|1100||Sony Alpha 580||1/2 frame||$900|
|900||Nikon D3100||1/2 frame||$700|
|850||Canon 7D||1/2 frame||$1300|
|800||Canon T2i / T3 / T3i / 60D||1/2 frame||$600-$900|
|550||Nikon D3000||1/2 frame||$500|
|550||Olympus EPL2 / E510||1/4 frame||$600|
You can see that the larger the sensor area, the better the
performance. You probably can't notice the noise difference between the Nikon D7000 (ISO 1200) and D3100 (ISO 900), but you will
easily notice that the D7000 has lower noise than the D3000 (ISO 550) and
more noise than the D700 (ISO 2300). Judgments are subjective, and
DxOMark is run by camera snobs (and I mean that in a good way) so you
will probably find that you're getting very good images at a stop higher ISO
than their rating (with some noise in the shadows if you look hard).
In part it depends on whether you want to print small 4x6" prints versus
big 20x30" posters.
If you can tolerate the noise in ISO 6400 pictures, great
for you! That means you can use the slow f/5.6 zoom lens that came
with your camera rather than having to upgrade to a more expensive
refers to the amount of light gathered for an image. A camera has
three controls that you must deal with:
(Jargon note: the word stop originally meant a physical object, such as the shutter blades we see above that stop light from entering the camera; now "stop" refers to the abstract concept of doubling (or halving) of light-gathering ability, in either aperture, shutter speed, or sensitivity. Two things are confusing about the f-stop numbers. First, f/1.4 seems like a small number but it is a big hole, while f/16 seems like a big number and is a small hole. That's because the "/" in "f/16" is a division sign; the larger the denominator of a fraction, the bigger the number it represents. Second, we don't just have whole numbers in a doubling pattern like 2, 4, 8; we also have intermediate numbers like 2.8 and 5.6. That's because the f-stop numbers refer to the diameter of the circle, but the amount of light is determined by the area, which is proportional to the square of the diameter. So a doubling of diameter (say from f/4 to f/2) is two stops (quadrupling the area), while a doubling of area (a single stop), would change the f-stop number not by a factor of 2 but by the square root of 2, which is about 1.4.
The lens that came with your camera is probably something like an 18-55mm zoom with a maximum aperture of f/3.5-f/5.6. This means a maximum of f/3.5 at the 18mm (wide) end and f/5.6 at the 55mm (telephoto) end. Lens also have a minimum aperture (usually in the f/16 to f/32 range) which is relevant for landscape and macro photography, but not for action.
The camera simulator below, generously made available by camerasim.com, lets you see the effects of varying aperture, shutter speed, and ISO sensitivity. Can you get a "nice/fine" picture with "bright indoors" lighting? You need to get just the right balance to avoid a "grainy" or "blurred" image. With "dim indoors" lighting, there is no way to get a "nice" picture with the available settings (you could do it with a f/1.4 lens rather than a f/2.8 lens, or with a camera that can handle ISO 1600 without getting grainy.)
We'll first consider a dimly-lit scene at EV 6. If we cut the shutter speed in half 6 times we go from 1 second to 1/2, 1/4, 1/8, 1/16, 1/32, and finally 1/64, which we round off to 1/60. So EV 6 is f/1, ISO 100, 1/60 sec
Of course, most of us do not have an f/1 lens, and we know that 1/60 sec would likely give us severe motion blur. The principle of reciprocity says that if we simultaneously double the ISO and half the exposure time we end up with the same exposure value. Same thing if we half the aperture area (which corresponds to multiplying the f number by 1.4) and double the ISO. (Note: in the song Popsicle Toes (which happens to involve a Pentax photographer), Michael Franks writes "When God gave out rhythm / He sure was good to you / You can add, subtract, multiply / and divide by two." And in fact, multiplying and dividing by two is all a photographer needs to do.)
According to the principle of reciprocity, EV 6 is equivalent to any of the following (I have added annotations for the amount of blur for a moving subject and the amount of noise):
|Aperture||Shutter speed||Sensitivty||Reason this is EV 6||Blur||Noise|
|f/1||1/60 sec||ISO 100||+6 stops in time from EV 0||severe||none|
|f/2.8||1/60 sec||ISO 800||reciprocity: -3 f-stops, +3 stops ISO from previous line||severe||slight|
|f/2.8||1/125 sec||ISO 1600||reciprocity: -1 stop time, +1 stop ISO from previous line||noticeable||slightly more|
|f/2.8||1/250 sec||ISO 3200||reciprocity: -1 stop time, +1 stop ISO from previous line||slight||more|
EV 6 is risky territory: assuming you have something like a Nikon D3100 or Canon T2i camera that performs well up to about ISO 800, then no matter what choice you make in the reciprocity trade-off you risk either motion blur or a noisy photo, or both. And that's with a f/2.8 lens! If you only have the f/5.6 lens that came with your camera, your choices are dismal at EV 6:
|Aperture||Shutter speed||Sensitivty||Reason this is EV 6||Blur||Noise|
|f/5.6||1/60 sec||ISO 3200||reciprocity: -2 f-stops, +2 stops time from previous line||severe||more|
|f/5.6||1/125 sec||ISO 6400||reciprocity: -1 stop time, +1 stops ISO from previous line||noticeable||severe|
|f/5.6||1/250 sec||ISO 12800||reciprocity: -1 stop time, +1 stops ISO from previous line||slight||extreme|
But you can turn that around by investing $100 in a 50mm f/1.8 lens (or $400 in a 85mm or 100mm lens) shooting at f/2:
|Aperture||Shutter speed||Sensitivty||Reason this is EV 6||Blur||Noise|
|f/2||1/250 sec||ISO 1600||reciprocity: +3 f-stops, -3 stops ISO from previous line||slight||slightly more|
|f/2||1/125 sec||ISO 800||reciprocity: +1 stop time, -1 stop ISO from previous line||noticeable||slight|
(You could also invest $2200 in a full-frame camera that does a stop better in ISO (among other benefits). Or you could wait two or three years and the same high-ISO technology will make its way into the sub-$1000 cameras.)
Now let's look on the bright side: in scenes that are lit at EV 8 you're in good shape with any lens:
|Aperture||Shutter speed||Sensitivty||Reason this is EV 8||Blur||Noise|
|f/1||1/250 sec||ISO 100||+8 stops in time from EV 0||slight||none|
|f/2.8||1/250 sec||ISO 400||reciprocity: -2 f-stops, +2 stops ISO from previous line||slight||almost none|
|f/2.8||1/500 sec||ISO 800||reciprocity: -1 stop speed, +1 stops ISO from previous line||almost none||slight|
|f/2.8||1/1000 sec||ISO 1600||reciprocity: -1 stop time, +1 stops ISO from previous line||none||slightly more|
|f/5.6||1/250 sec||ISO 1600||reciprocity: -2 f-stops, +2 stops time from previous line||slight||slightly more|
|f/2||1/1000 sec||ISO 800||reciprocity: +1 f-stop, +1 stop ISO from 2 lines above||none||slight|
|f/2||1/500 sec||ISO 400||reciprocity: +1 f-stop, +1 stop ISO from previous line||almost none||almost none|
So much for the theory of exposure. But what do you do in practice? You have four things to worry about:
Metering and intent: Your camera has a built-in automatic exposure meter, which usually does a pretty good job of deciding the exposure value of a scene. Let it. Although some people succsfully use manual exposure for dance, I find that for the performances I shoot the lighting changes too rapidly and too much from scene to scene and from one point on the stage to another. However, there are a few metering situations that the camera consistently gets wrong, and you will have to step in and make a correction by twisting a dial.
What do I mean by "wrong"? I mean contrary to your intent. There are many ways to interpret a scene. If there is a dimly-lit dancer to the side of the stage, the camera doesn't know if you want to darken the image to make that dancer disappear into the background, or lighten the image to make that dancer one of the points of interest. You can change the camera's exposure to match your intent, but the camera can't always guess what your intent is.
A common problem in dance photography is when there is an area of the image that you would like to be dark black. The default metering scheme of the camera tries to pull detail out of every image; if there is a large black area the camera will try to make it gray rather than black. If you truly want it to remain black, you need to tell the camera that. The way you do it is exposure compensation, which should be set up on a dial operated by your right thumb on the back of the camera. To keep a dark area black, dial in negative exposure compensation: tell the camera you want the picture to be darker than the camera guessed, by moving the dial. In the two images below, I used an exposure compensation of -2 stops.
|f/2.8, 1/160, ISO 1600 (EV 6), EC -2||f/2.8, 1/125, ISO 640 (EV 7), EC -2|
Another problem is when a white (or other light-colored) outfit is large in the frame. The camera can't tell if the outfit is supposed to be gray or white; it guesses gray and exposes accordingly. You fix that mistake by dialing in positive exposure compensation. In the photos below this was +1 stop.
|f/2.8, 1/400, ISO 1250 (EV 8), EC +1||f/2.8, 1/250, ISO 800 (EV 8), EC +1|
Mechanics: Given that you're not going to try to manage exposure manually, you have a choice of what are called "shooting modes," but should be called something like "exposure priority modes." I almost always set (and leave) the dial on top of the camera to aperture priority mode ("A" in Nikon; "Av" in Canon). This means that I can adjust the aperture with the dial near my right index finger.
Here's what happens: suppose I am shooting the white dress picture above right. I have already set the controls on the camera to f/2.8, ISO 800, and exposure compensation +1. When I press the shutter button the camera's exposure meter measures the incoming light as EV 7, and adds the +1 compensation to arrive at EV 8. The camera then determines that the shutter speed should be 1/250 to achieve an EV 8 exposure, and takes the image with those settings.
The camera engineers couldn't get the automatic exposure metering right for every picture, so they compensated by giving you several choices of automatic metering systems. (That way, if the exposure is messed up, the engineers can always say "you should have used the other mode!") The metering modes are:
I set the "Av" mode and the metering mode ahead of time and tend to leave them alone. But for each individual shot, I have the option of changing the aperture with my right index finger (to get more depth of field), and I have the option of setting exposure compensation with my right thumb (if the scene has a lot of black or white that will throw off the metering system). Also, if the lighting changes to be much more bright or dim, I will change the ISO (with right index finger button press and then dial adjustment).
There is one more dial that I use regularly: my camera has a shooting mode dial that contains three "custom" modes. I set up C1 to be my regular settings as described above, and C2 to be my "blur" setting: shutter priority mode ("Tv" in Canon) with a shutter time of 1/5 second, ISO 100. I can then switch back and forth easily between blurs and norrmal shots.
So much for dials and buttons; now let's concentrate on what the camera tells you. The great advantage of a digital camera over film is that you can see each image on the camera's LCD screen. You can quickly tell if an image is all wrong, however you won't be able to accurately judge precise focus or exposure on the small screen. Two things are more important than the image itself:
First, set up your screen to show blinking highlights. The idea is that if there is a portion of the image that is so bright that your sensor can't record details, that portion will blink on and off. Look at the display, and if the portion is something that doesn't matter--like a spotlight off stage--no problem. Below we see blinking highlights on the dancer's shoulder (due to a hot spotlight there). This image was shot at +1/3 exposure compensation; it probably should have been about -1/3 to preserve these highlights.
|Blinking highlights and LCD display|
The second thing is your most important guide to judging your image: the histogram. The histogram shows the relative number of pixels at different light intensities from very dark (on the left) to very light (on the right). If the histogram has a bump (or bumps) in the middle, you're good. If there is a bump crunched up against the left edge that is a potential problem. It means that some details are lost in shadow. That might be fine (you want the unlit edges of the stage to be black) or it may mean that you need to add positive exposure compensation (to get detail back). At the other end, if a bump is crunching against the right edge, you may need negative exposure compensation to bring the details back from the white highlights.
|Histogram: levels of light|
The histogram helps you with tricky situations like the one below. She has a white outfit, so you have to worry about blowing the highlights (suggesting negative exposure compensation) but also about the automatic metering turning the white to gray (suggesting positive compensation). He has a black outfit, but it is much smaller in the frame, so the main concern is retaining detail (suggesting positive compensation). I shot at -1/3 compensation; was that acceptable? The histogram shows separate channels for red, green, and blue, and one for overall luminance. Together they show that we've done a good job of protecting the highlights (nothing bumps up against the right edge) but that there is a slight problem with the darkest blacks losing detail (they are bunched up along the left, but only a sliver). Not a big problem, but probably this shot would have been better at about 0 or +1/3 exposure compensation.
|Histogram display on LCD|
Another tricky situation was the shot above of a white dress on a black background captioned "f/2.8, 1/250, ISO 800 (EV 8), EC +1 ." You want to protect the highlights but make sure the dress is white and the blacks are black. Does that mean positive exposure compensation, negtive, or zero? I used +1, which was about right; I can tell by looking at the histogram.
Mechanics of the eye: Beyond dials and displays, my main advice for mechanics is: Don't get fixated on the center of the viewfinder. It is all too easy when there is a lot of action to get tunnel vision. If you see something you like in the center of the frame, take a few shots, but then begin to broaden your tunnel vision. First look to the edges of the frame: is there something distracting there? Something that should be included or removed from view? Have we framed the image well? Then look to the camera display at the bottom of the frame. Are all the settings good? If your shutter speed is showing 1/60 and the dance movements are rapid, you're in trouble--you need to bump up the ISO. But you won't know to do that unless you notice your shutter speed is 1/60. And finally, open your other eye, look around the stage and check if there is something else you should be catching. Then, after the shot (or shots), take time to look at the playback display, particularly the histogram. This process of looking at the display has acquired the pejorative name of chimping (after photographers who exclaim "ooh! ooh! ahh! ahh!" at their own shots) but it is an important part of accelerating the learning, correction, and improvement process.
Postprocessing: One school of thought is to get your exposure just right in the camera. Then you don't have to spend time messing around with your photo editing program ("Damn it Jim, I'm a photographer, not a photoshopper!").
The other school is that you should capture as much digital information as possible in the camera, and then manipulate the image to make it look the way you want. If you subscribe to this school there are two main things to do:
Shoot wide, then crop: instead of trying to get the framing perfect, zoom out a little, with the expectation that you will crop the image in your editing program. This makes action photography easier; if you are shooting wider you don't have to worry as much about a hand getting cut by the side of the frame, you just leave plenty of space and remove the excess space with a crop in your editing program.
|173mm: too long;|
can crop in editing program
Expose to the right (ETTR): this technique, originally proposed by Thomas Knoll, the author of Photoshop, recognizes that the highest signal-to-noise ratio can be captured by shifting the histogram as far to the right as is possible without having it actually bump up against the right edge (and thus lose information). If this shift (achieved by positive exposure compensation) makes the image look too bright, then we can fix things by dialing the exposure back down in the photo editing program. The theory is that you get more information (more detail, less noise) if you expose to the right and shift back than if you had exposed "properly" in the first place. There are many good examples of this approach working well, but you do run the risk of clipping the highlights, and it is more work in the editing program.
|"Get it right in camera" histogram||"Expose to the right" histogram|
If you are in the get-it-right-in-camera school, you will set your camera to capture jpeg images. These have the advantage of smaller file sizes, making them easier to store and share, and also making it possible to take a longer burst of continuous images before your camera grinds to a halt. If you are in the postprocessing school, you should capture raw images, which have larger file sizes but allow you more leeway in making edits, particularly for color balance and exposure.
On the next page we will move away from the mechanics and cover the aesthetics.