by D Gary Grady
Aug 4, 2002
I am fairly confused about the whole 16:9 aspect ratio. I am hoping that someone could explain to me the basic ins and outs.Let's see how much I can cram into as little space as possible while still making it coherent. I'm sure I'll belabor plenty of things you already know, but bear with me.
The traditional proportions of a television screen match those of classic movies, with the height 3/4 of the width. We normally write that the other way around, 4:3. It's easier to compare to other screen shapes if we convert it to decimal form and make it 1.33:1. That is, the width is 1.33 times the height.
Technically, this matches the shape of silent movies. Sound 35 mm films through the early 1950s were slightly wider at 1.37:1. In the 1950s theatrical features started to be made in even wider proportions, with the standard eventually settling down to two shapes: 1.85:1 and 2.39:1. The latter is often incorrectly written 2.35:1 because that was for years the Cinemascope standard; it became 2.39:1 about 1970. Which is of course picky irrelevant trivia if you're shooting DV, because it's very hard to get a good 2.39:1 image out of DV.
Television production is currently moving in the direction of a wider screen shape as well. This was debated all over the place in the early 1980s (a lot of cinematographers wanted 2:1, for example), and eventually a standard of 16:9 (1.78:1) was adopted. This happens to be the geometric midpoint between silent movies and Cinemascope, so there's some logic to it. And it's also nearly identical to the theatrical standard 1.85:1 in which most feature films are shot.
Now, there's nothing inherent in a television signal that gives it any particular proportions to the image.
DV, for example, uses 720 pixels per row and either 480 rows (NTSC) or 576 (PAL). (A pixel is a picture element, the dots that make up a digital video picture.) If pixels were square, as on a computer screen, an NTSC image would be 1.5 times as wide as high (3:2 proportions, like a 35 mm slide). PAL TV would be only 1.25 times as wide as high (proportioned 5:4).
But nothing says pixels have to be square, and both NTSC and PAL standard television screens have the same proportions, 4:3 (which we can also write, rounded off, as 1.33:1). As long as the camera's chips have the same shape as the television screen, the picture looks OK.
(By the way, NTSC is the television system used in North America, Japan, and a few other places. PAL or its SECAM variant is used elsewhere. For several reasons, mainly the greater number of pixel rows, PAL is slightly better for productions intended for transfer to film. But NTSC looks nearly as good.)
If you want to create an image to display on a 16:9 television screen (or to create a 1.85:1 film image), you ideally want to start with a camera that has 16:9 chips. Unfortunately, these are at present not cheap.
An alternative is to use a standard 4:3 camera with an anamorphic adapter lens. This squeezes the image so that a 16:9 picture fits onto a 4:3 chip. (A very similar approach is used to create a 2.39:1 image on 35 mm film.) The result is virtually identical to what you'd get using a true 16:9 camera.
Of course, there has to be a catch, and there are several. First, anamorphic adapter lenses cost about $700. Also, they may keep you from holding focus during a zoom (which you should usually avoid doing during a shot anyway, though), and they may make the picture soft at the telephoto end of the zoom range. Both Century and Optex make these adapters. I've heard slightly better things about the Optex but never used either personally. I've also heard that these adapter lenses are a bit clumsy to work with.
Suppose you can't get a 16:9 camera or even an anamorphic adapter for a 4:3 camera. What then? Well, you can always extract a 16:9 rectangle (or any other shape, for that matter) from a 4:3 image.
Contrary to some things you'll hear, the result will have the *exact same resolution* as an ordinary 4:3 image. (Lopping off part of an image doesn't change the resolution in the rest of it.) But it won't have as much vertical resolution as a native 16:9 image because a 16:9 image crams more pixel rows into the same shape, thus squeezing pixels closer together.
The difference isn't drastic, though, only a factor of 3/4. A PAL 4:3 image "letterboxed" down to a 16:9 shape will have only 432 pixel rows, but that's almost as many as the 480 in a standard NTSC image.
Your camera can actually do this for you. Its electronic 16:9 feature will convert the middle band of its 4:3 image to 16:9 on the fly and record that to tape.
If you have a Canon camcorder, in fact, this is the way to go. Because the conversion takes place before compression, you wind up using more compression blocks on the active part of the image and getting slightly better resolution than in 4:3. But if you have a Sony camera, the conversion adds too much vertical edge enhancement and you end up worse off. (See adamwilt.com for more on this.) If you have a Sony, shoot 4:3 and convert to 16:9 in post-production.
Basically I hope to shoot a low budget film and then blow it up for theater distribution.
Join the club. Not only is everybody and his dog doing this, so
now are gerbils and parakeets. And tropical fish -- you think all
the messages on this list about underwater cases are an accident?
I've been told that even if my camcorder is in 16:9
mode that I still need to block off a half an inch off the top and
bottom of the screen with tape to ensure that the 16:9 is accurate.
Is this correct?
Not even remotely close. Conversion to 16:9 in a camera or NLE
is 100% perfect in terms of proportions. Very few viewscreens
are that precise (most cut off some of the image), and when it
comes to humans and pieces of tape...
Also, for the record, remember that theaters don't project 16:9. They're supposed to project 1.85:1 or 2.39:1, and in practice it's not unusual for them to be way, way off from these proportions. Often a 1.85:1 image is shown with some cut off at the top and bottom; something closer to 2:1 shape is actually quite common.