D Gary Grady

Jan 9 2001

The TRV900 and all cameras in more or less the same range have very small CCDs, typically 1/4 inch or 1/3 inch nominally. with actual image dimensions even smaller than that implies.

Other things being equal, smaller chips have more depth of field than larger ones -- a lot more. If you like deep-focus photography this is good. But if you want to use selective focus, you have to resort to very wide lens openings (compensated for with a neutral density filter if needed) and longer focal lengths.

Smaller chips are also more prone to loss of resolution caused by diffraction (light's tendency to spread out when it goes through a narrow opening). It's a good idea to avoid apertures smaller than f/11 with 1/3-inch chips (as in the Canon XL1 and the Sony VX2000 and PD150). With 1/4-inch CCDs (Sony TRV900 and PD100, Canon GL1/XM1), diffraction can start becoming objectionable with lens openings as large as f/8.

THE TECHNICAL DETAILS IN SUMMARY

You can skip the formulas in this section without losing much, but you might want to at least skim the text.

The formulas I use here are reasonably accurate, especially at the wide end of the zoom range, but I used approximations to make the formulas simpler. For a more detailed discussion, see a text on photographic optics or the useful tutorial at photo.net.

Depth of field isn't precisely defined, because there's no sharp division between "apparently in focus" and "clearly out of focus." To define depth of field we have to declare what we consider an acceptable "circle of confusion" -- the maximum amount by which a point can be blurred into a disk without being perceived by the eye as "out of focus." This depends on the size of the image area behind the lens. For our purposes we can limit the circle of confusion to 1/500 of the height of the image on the focal plane.

A 2/3-inch CCD such as those found in many professional video cameras has an image area with an 11 mm diagonal. For conventional 4:3 format television, this works out to a picture 8.8 by 6.6 mm.

(You'll note that none of these dimensions is very close to 2/3 inch. In fact, pretty much nothing about a 2/3-inch CCD is 2/3 of an inch; a 2/3-inch CCD is simply one that has the same imaging area as a 2/3-inch television camera tube. For what it's worth, those dimensions happen are very close to those of 16 mm movie film.)

Obviously, 1/3-inch and 1/4-inch chips are correspondingly smaller, and here I assume that they are smaller in direct proportion, which is possibly not the case, although the results of my calculations seem to correspond closely to those of manufacturer's specs where I have been able to check.

If you focus a lens at infinity, the closest distance that will still be in focus is called the "hyperfocal distance." Hyperfocal distance can be computed as

h = f^2 / Nc

where f^2 is the the square of the focal length of the lens, N is the f-stop, and c is the size of the circle of confusion.

From the hyperfocal distance we can compute the depth of field. If we call the lens focus setting s, then the depth of field runs from

near depth of field = hs / (h + s)

far depth of field = hs / (h - s)

(If h - s is zero or negative, then the far end of the depth of field is infinity.)

Notice that if s = h, that is, we set focus the lens on the hyperfocal distance, everything from 1/2 that distance to infinity will be in acceptable focus.

WHAT'S A "NORMAL" LENS? WHAT'S A "TYPICAL" SHOT?

There are a lot of variables here, including focal length, depth of field, and focusing distance. To simplify things, let's assume we're talking about a "normal" lens for the format, with "normal lens" arbitrarily defined as one that yields a vertical viewing angle of 27 degrees, such as we get with a 50 mm lens in 35 mm photography. If we take 5 feet (1.5 m) as a "typical" shooting distance for a human subject, this viewing angle would cover about 30 inches (75 cm) in height.

"TYPICAL" DEPTH OF FIELD

For different formats using a "normal" lens at "typical" distance as defined above, here some depth of field ranges for different f-stops.

35 mm slide (50 mm lens) focused at 1.5 meters (5 feet) f/2.8: 1.4 - 1.6 m (4.6 - 5.3 ft) f/4: 1.3 - 1.7 m (4.4 - 5.5 ft) f/5.6: 1.3 - 1.8 m (4.2 - 5.8 ft)2/3-inch video (14 mm lens) focused at 1.5 meters (5 feet) f/2.8: 1.2 - 2.1 m (3.8 - 6.9 ft) f/4: 1.1 - 2.6 m (3.5 - 8.4 ft) f/5.6: 0.9 - 3.6 m (3.1 - 11.7 ft) 1/3-inch video (7 mm lens) focused at 1.5 meters (5 feet) f/2.8: 0.9 - 3.6 m (3.1 - 11.8 ft) f/4: 0.8 m - infinity (2.3 ft - infinity) f/5.6: 0.7 m - infinity (2.3 ft - infinity) 1/4-inch video (5 mm lens) focused at 1.5 meters (5 feet) f/2.8: 0.8 - 6.8 m (2.8 - 22.3 ft) f/4: 0.7 m - infinity (2.3 ft - infinity) f/5.6: 0.6 m - infinity (1.9 ft - infinity)

SHALLOW DEPTH OF FIELD WITH 1/4-INCH CCDs

Since depth of field is incredibly deep with a 1/4-inch chip, even at f/2.8 for a normal lens, you might have to resort to a longer focal length if you want to make use of shallower focus. But longer focal lengths also tend to give the picture less of an appearance of depth, so you probably don't want to go overboard. Here are some depth of field numbers for an image 30 inches (75 cm) high at the focused distance, assuming f/2.8.

Field depth with 1/4-inch video at f/2.8 10 mm lens focused at 3 m (10 ft): 2.2 - 4.9 m (7.1 - 16.1 ft) 21 mm lens focused at 6 m (20 ft): 5.0 - 7.5 m (16.5 - 24.4 ft) 31 mm lens focused at 9 m (30 ft): 8.0 - 10.3 m (26.1 - 33.9 ft)(For comparison purposes, the equivalent 35 mm still focal length is about 10 times the focal length shown.)

Frankly, achieving a shallow-focus effect without screwing up something else is just hard to do with these cameras. I think it's probably best to watch some Orson Welles movies and learn to appreciate the wonders of deep-focus cinematography and its possibilities for composition in depth.

DIFFRACTION WORRIES

As a rule of thumb, resolution is diffraction-limited to about 1500 / N lines per millimeter, where N is the f-stop. (Some sources give 1200 / N or some slightly different formula. As picking a circle of confusion, there's some arbitrariness here.) If we want to preserve 500 lines of vertical resolution, then our f/ number needs to be no greater than 3 times the chip's image height in millimeters. (Proving this is left as an exercise to the mathematically inclined reader.)

For a 2/3 inch CCD with its 6.6 mm image height, we should avoid f-stops much higher than f/19.8, which isn't a big problem. But if the images of smaller chips are in proportion, then for a 1/3-inch CCDs we get into diffraction trouble at f/9.9, and for 1/4-inch chips even f/7.4 can be expected to hurt us!

Hence for maximum sharpness we should avoid stopping the lens down much below f/5.6 with a GL1/XM1 or TRV900/PD100. We can go about one stop smaller on an XL1, VX2000/PD150. Note also that lenses in general tend to be at their best when stopped down a couple of stops from their maximum aperture, so the middle of the exposure range is best.

(Of course, diffraction may fuzz an image up much as a diffuser does, so it might be that a smaller aperture may produce a more pleasing, less harsh image in some circumstances. There's a use for almost anything.)

D Gary Grady

Back to TRV900 page.