Depth of Field and Focus. How do they work?

October 4, 2005 11:52 PM

Selective Depth-of-Field in Six Feet Under, and a question about lens flares.

In the commentary for either the first or last episode of Season 1, the director mentioned a lens technique that would put two subjects in focus when they're each positioned at different depths. He couldn't recall the name of this technique, or even how it worked. What is it called, and how does it work?

I've taken some screen shots of examples here and here. (~150KB jpegs) They aren't the best representations available in the show, but it was tedious to find them. Many are in the kitchen, where the characters are very far apart. In the first example, note that both characters are mostly in focus, while the edges of David's face and the arm of the chair between him and Nate is not. Similarly, in the second example, Ruth's face is in focus, as is the background, but her dress is not entirely in focus.

And, more of a general lens question, what determines the shape of lens flares or point light sources thrown out of focus? In this example, the lights appear as 9-sided regular polygons. I've seen everything from stars to perfect circles. What are the physics/optics behind these shapes?

posted by odinsdream to media & arts (12 comments total)

Focus is not a point, but a range ('depth of field'), dependent on f-stops and shutter speed (at least in still photograpy), so to get 2 subjects in focus, you use a wide depth of field and focus somewhere between them.
posted by signal at 12:15 AM on October 5, 2005

Not sure how you get the two focal planes, though it might be some sort of post-processing effect wherein multiple images are composed somehow.

The 9-sided flares you see are due to the shape of the aperture in the lens (the bit that grows/shrinks like a pupil to control the amount of light entering and therefore exposure and depth of field). The aperture is usually constructed as a set of blades that move in/out. The flare you see is sunlight coming through the aperture, reflecting off a couple of lens elements and then being imaged.
posted by polyglot at 12:15 AM on October 5, 2005

As to the first question, the technique you're looking for is "deep focus," or as Alan Caso, the director of photography on SFU calls it, "split focus." IMDB explains how the technique worked in Citizen Kane, arguably the most famous use of deep focus. You can apparently do this with digital compositing techniques as well, but if the director mentions lens trickery, then it's probably the split lens trick. Basically it seems to work the same way bifocal glasses work: there's actually a section of the lens with a more/less severe curvature to it, thus changing the focal length of that part of the lens. The IMDB article mentions using objects like bedposts to hide the thin line that results from such a lens. As you can now get no-line bifocal glasses, I imagine they're also making camera lenses that can do split focus without the telltale line.
posted by chrominance at 12:23 AM on October 5, 2005

Regarding the two focal planes: Chrominance is correct with the bifocal description, that's essentially what it is. It's generally referred to as a split-diopter lens (or sometimes just diopter lens). It's been used in plenty of movies. Reservoir Dogs, Jaws, and The Butterfly Effect are when I noticed it most.

Since there are two seperate focal planes, objects that exist in focus in one plane will not be in focus in the other (e.g. the woman's dress in the second example).

Here's a bboard discussion regarding split-diopter lens usage.

No idea on the flares, though.
posted by dogwalker at 1:15 AM on October 5, 2005

The pics that you posted are most likely taken with a tilt/shift lens popular in architectural photography as well as fashion photography. Basically it shifts the plane of focus so that is not parallel with the film plane. It becomes obvious when things that would normally be in focus are not.

In example 2, in a normal picture, both the woman's eyes would be in focus because she is facing the camera straight on. The plane of focus has been twisted so that it intersects her left eye and the center of the man behind her. It is tricky to use in video because the actors can't move much without getting out of focus.

Usually, as has been mentioned above, a split focus filter is used for a similar effect but the differences are obvious. If that were the case in these pictures, then the entire face of the woman would be in focus instead of one eye.

With the lens flare, it is as polyglot said. As the aperture is "stopped down", it becomes more of a regular polygon. Usually it is 8-sided I believe. At more wide open apertures, the shape is almost like a circle. With removable lenses, there is actually a trick that you can make your own shape aperture by cutting out a shape in foil for example and placing it on the rear of the lens. Shoot with the aperture wide open and experiment with shutter speed to get the correct exposure. Any out of focus highlights will be the shape of the cut out.
posted by JJ86 at 1:45 AM on October 5, 2005

This article on the differences between two Canon lenses nicely illustrates the effect of the number of aperture blades on lens flares. One lens has eight blades and produces octagons and eight-pointed stars. The other has five and produces pentagons and five-pointed stars.
posted by caek at 3:54 AM on October 5, 2005

This is achieved in camera by using a split lens.

A close up lens that has been cut in half to only cover half of the lens area is placed over the normal lens. Then, you can adjust so that both desired objects are in focus but things between them and outside the depth of field are not.

You can buy these pretty readily. I used to have one like this from Cokin.
posted by aaronh at 6:01 AM on October 5, 2005

Thanks everyone! While I can't really know for sure whether it's a tilt-shift like JJ86 hypothesises, or a split diopter like others have mentioned, I have a feeling it's the tilt-shift, because of some commonalities that all of these shots share. Nearly every instance of it involves two actors, one in the extreme foreground, one in the background, usually standing in oddly similar geometry across all the shots. They usually never move more than a few inches, and often don't even turn their heads.

Of course, it could be both, or it could be digital post-processing. Anyway, thanks for the answers!
posted by odinsdream at 6:43 AM on October 5, 2005

Related - the commentary track for Citizen Kane is fascinating. Roger Ebert points out the various techniques (including split screens) used to keep everything in focus at all times... foreground and background. REALLY interesting if you notice things like this.
posted by kdern at 7:21 AM on October 5, 2005

The reason such trickery is necessary at all, of course, is that the length of the exposure is fixed at 1/24th of a second, because that's how fast film moves through the camera. That necessitates either a lot of light or a large aperture to let plenty of light in -- usually a combination of both. A large aperture means that your depth of field is narrow (the larger the aperture, the narrower the field). Which means it begins to get tricky to have more than one thing in focus at once.

Movie cameras don't have automatic focus. That means they must be adjusted constantly as characters move around to keep them in focus. Which means that a lot of times you'll see the actors moving exactly parallel to the camera. There's a separate guy from the camera operator who handles the focus -- he's called the focus puller and is responsible for snapping the focus from one subject to another or keeping a subject in focus or intentionally letting everything go blurry at the end of a shot.
posted by kindall at 8:58 AM on October 5, 2005

The reason such trickery is necessary at all, of course, is that the length of the exposure is fixed at 1/24th of a second...

Well, these shots I'm interested in are unique in that only the two spots are in focus, while things in between and behind the subjects are blurred. It's a neat effect... I wish I could have offered better examples.
posted by odinsdream at 9:23 AM on October 5, 2005

A couple of additional notes from kindall's answer. Movie cameras don't have automatic focus by design, I think; there's no way a computer's going to know what the director or the DP wants in focus.

Also, it's been said that the mark of a professional shoot (well, one of them, anyways) is a narrow depth of field. It's a desired effect because it blurs everything except what you want to focus on (as opposed to a larger field, which would leave more distance in front of and behind the subject in focus); the background appears to recede, and clutter becomes less noticable because of the blur. This is especially important when you're dealing with long shots; perspective tends to flatten, meaning it's harder to discern the subject from the background.
posted by chrominance at 1:15 AM on October 6, 2005