Sometimes people see me with my camera and ask me, "So what's the big deal with a camera like that? What can it do that my point and shoot can't do?" The simplest answer is almost nothing. But everything that they do, a single lens reflex does better and/or faster than a point and shoot, and in my opinion the price difference (twice as much, more or less) is worth it if you're really interested in photography. For example, SLRs have bigger sensors, better available lenses, more artificial lighting options (flashes), more sophisticated metering, and quicker and more accurate autofocus. All of these things add up to more consistently good images, even if the extra controls on the SLR can be a bit daunting if you're used to a point and shoot camera. This lesson will discuss a couple of those differences, specifically focus and light metering systems. The next lesson, which will be on light and color, will cover flashes as part of that.
Focus
I think the best way to start a discussion of focus is with a bit of physics (an admittedly biased view). Lenses change the direction of light that passes through them. So how do you change the direction? There are two ways to do it: refraction and reflection. Reflecting lenses (ones based on mirrors) are rarely used in general photography because of the way that out-of-focus objects look, but since astronomers are almost always looking at things that are basically infinitely far away, they make sense for them and are used almost exclusively mostly because they have no chromatic aberration (see below), but also because they offer size, weight, and cost benefits.
Refraction is just the bending of light as it crosses the boundary between two areas where its speed is different. What?, you say (or maybe not). The speed of light isn't constant? What about relativity and Einstein and all that? It turns out that the speed of light in vacuum is constant but when it goes through something like air, glass, or water, its interactions with the matter slow it down. But why would changing speeds have anything to do with changing directions? I'll use my favorite analogy to explain it. Let's say you're driving a dune-buggy through a parking lot that is surrounded by sand. To make things simple, let's assume that it's only front-wheel drive.
If you drive straight toward the pavement-sand boundary (perpendicular to it), when you get to the boundary your buggy keeps going in the same direction, only slower because now your wheels are going through sand. Your dune buggy is like the light; the pavement could be air, where light goes fast; and the sand could be glass, where light travels more slowly.
Well, driving straight at the boundary isn't the most enlightening situation, so let's say next time you drive toward the boundary at an angle. This time, one of your wheels hits the sand first, and that side of your buggy slows down while the other side keeps going fast, so your buggy turns toward the side that hit first. It's just the opposite for the reverse direction: come toward the pavement from the sand at an angle and the wheel that hits the pavement first will pull that side of the buggy forward, turning the buggy away from that wheel.
If you know how much a material (like glass) bends light, you can get a lens to focus light that's coming from one side onto the sensor 8 in the camera by making the lens the right shape so that it bends it just how you want it. By moving the lens closer or further away from the sensor, you can bring one plane (it's mostly a plane, anyway) of your vision into focus, and the further things are from that plane, the more blurry they'll be.
How fast things get blurry as you move away from the focal plane is controlled by the aperture (with smaller apertures, a bigger range of things is in focus—see lesson 1), and the range where things are reasonably in focus is called your depth of field. Nature throws us a curve, though, and makes it not quite as simple as just shaping a lens right. Light of different colors has different speeds in any medium except vacuum, so generally red light will get bent less than blue light because red light will go faster than the blue.
It's the phenomenon behind splitting white light into colors with a prism and also the one behind rainbows, where the water droplets act like reflecting prisms. It's pretty, but it makes trouble for lens design because a lens will have a different focal length for each color (see the picture). They call this chromatic aberration (color mistakes). Have you ever wondered why you would need to have something like 5, 10 or maybe even 20 elements (lenses) just to make one lens? A big part of the answer is to correct for this chromatic aberration, and they use all kinds of exotic materials and different lens elements to try to control it.
Autofocus is a pretty recent thing (30 years or so old). There are two types of autofocus systems that are generally used now: contrast detection systems and phase detection systems. Contrast detection is what most point and shoot cameras use.
When you push the shutter button, the camera will actually move the lens back and forth and find where the contrast is best in the area where you want it to focus—it just looks for the sharpest edges in that area. The main disadvantage with this is that you have to sit and wait while the lens gets moved to find the best focus, and that can take a while (meanwhile, your child has stopped smiling and started picking his nose or has closed his eyes).
Phase detection I don't really understand (you've got to have something to learn about, right?), but the main idea is that you split light from opposite sides of the lens, and based on how it lands on the autofocus sensor the camera (usually) knows both how far out of focus the lens is and in what direction you need to move the lens to bring it in focus. The bottom line is that it's much faster than contrast detection.
Metering
If you get the exposure wrong in either direction, you've got problems on your hands. If you underexpose and then try to increase the brightness of the picture, at best your shadows will have extra noise and at worst your whole image will have extra noise.
If you overexpose, you can saturate your photodiode at pixels in highlight areas (the analog to digital converter has a maximum, and if it gets a value higher than its max, it still just reads the maximum value). That means that highlights lost are lost forever.
Before you go taking darker pictures, though, know that there are times when you don't care so much about blown highlights and times when you do (one that comes to mind is detail on a bride's dress— it's really better to keep that information if possible).
Since our eyes are so good at adjusting to different lighting conditions to give reasonable results almost anywhere we would need to see, you really need an objective witness to help decide how much to expose your sensor. Back in the day, this meant either being able to put a light meter where your subject was and then choosing your exposure based on how much light it would reflect or pointing a directional light meter at your subject and setting exposure based on the actual light coming off the subject.
Pros still set flash output with light meters, but really most metering has gone into the camera. With a point and shoot, they use the sensor itself as a meter, but SLRs have them up in the viewfinder housing. With most cameras you choose how you want your camera to meter: usually the choices are a spot (called, aptly, spot metering), a bigger spot (center-weighted average), or a "smart mode" where the camera guesses what kind of scene it's looking at and sets exposure based on what it thinks is important (matrix or multi-zone metering).
My opinion on them is that if I shoot raw, matrix metering gets me a usable exposure 97% of the time, but there are times when I prefer the other options because they're easier to predict—specifically when I want to hold detail in a subject that is either very dark or very light. What if (heaven forbid!) you want to hold detail in both highlight and shadow (think black tux and white wedding dress—incidentally, I'd favor keeping highlight detail here)? This is one time when I would say if it's possible don't shoot jpeg whatever you do because they are so unforgiving. Even if you can only shoot jpeg, the beautiful thing about digital is the instant feedback.
If you were doing a wedding with wrong technique with film (say you blew the highlights of the wedding dress in every picture), you'd never know until it was all developed so that could be a huge problem. With digital, you might shoot five before you checked if your metering was right and adjusted for the mistake.
Histograms and Exposure Compensation
There are some situations where the camera won’t choose the right exposure no matter what metering mode it’s in. The camera’s metering system tries to keep things not too bright and not too dark. This means that if you’re taking a picture of a scene with a lot of dark things or a lot of bright things, you have to tell the camera or it will get the exposure wrong.
You tell the camera to adjust its metering with a setting called exposure compensation. It's given in exposure value (EV) steps which are exactly equal to the stops I talked about in the first lesson. A setting of +1 EV is one stop brighter than usual, so the camera will get twice the light it normally would, either by keeping the shutter open twice as long, or making the aperture 1.4 times bigger. To make white snow white, tell the camera to overexpose (with a positive exposure compensation). To keep things dark in a dark scene, tell the camera to underexpose (with a negative exposure compensation)
So do you believe the unfeeling (but wellprogrammed) camera got the exposure right? You can get a feel whether it’s close by looking at the image in the LCD after you take it, but there has to be a way to take out the guesswork. Besides, in bright sunlight your LCD monitor can be so overpowered by the ambient light that it’s hard to see the image let alone check the exposure, or a picture that looks great in the dark on your LCD screen might be underexposed without you noticing. It turns out that there’s something called the histogram that lets you check exposure objectively, no matter the ambient conditions.
What the camera does to build the histogram is count the number of pixels at each brightness level, and plot the number at each level from darkest to brightest. It takes some learning, but once you get the hang of it, using the histogram can save you in tricky exposure situations. Keep in mind that there’s not one “good” histogram.
If you’re taking a picture of a scene with huge dark areas, you’ll have a lot of low-brightness pixels and that will show as a spike on the left of the histogram. A picture of someone skiing might have a huge lump on the right of the histogram caused by all the snow pixels. Using it for a while will give you a feel for how much exposure compensation you have to use in different situations for future pictures.
There’s another reason to use the histogram, though. To reduce noise in the shadows, you want to get as much light as you can from those areas without completely filling up the pixels that are getting light from bright areas of your image.
The histogram lets you “expose to the right.” You flirt with disaster (disaster here is blowing out highlights) by getting the bright pixels as close to the right edge of the histogram without going off of it.
That might mean the dark pixels are bunched up around halfway, but that’s okay since you can always bring the exposure down when you process the images. If you underexpose and then bring the exposure up in processing it’s like you shot the picture with a higher ISO sensitivity, and that’s not a good thing since you’ll see more noise.
If you think the exposure of a scene is tricky and don’t want to spend a long time making sure you get it right, some cameras have a way to bracket exposure: it takes a sequence 11 of shots spread around what it thinks is the right exposure separated by steps that you set.
Shooting Modes
M is manual mode. Here you set both aperture and shutter speed, and the camera still meters to tell you how your exposure is.
A is aperture priority mode. You set the aperture, and the camera adjusts shutter speed to get exposure right.
S or T is shutter priority mode (T for reasons unknown to me). You set shutter speed, the camera sets aperture. P is programmed auto mode. The camera sets both settings. Be careful with this one because it sometimes does some silly things. For example, my camera won't ever open the aperture past f/3.5 or so in programmed auto, so you can't even use some lenses at their best with it.
No comments:
Post a Comment