In basic photo, we [unfortunately] learn that D of F is directly affected by three things: 1) aperture, 2) distance from subject, and 3) focal length of lens. Additionally, we also learn sometimes that film format affects it. Actually, it is only 1) f stop, and 2) magnification. Distance from subject and focal length of the lens can affect magnification, and thus affect D of F. Given the same angle of view/composition, film format also affects it indirectly by affecting magnification. However, they do not do it directly, as we learn in the basic photo classes and books. Changing either distance or focal length or film format will have no effect on D of F *unless* doing so also changes the magnification. The higher the magnification, the less D of F there is.
The idea that D of F is defined as the front-to-back area that "is acceptably sharp" in the print has never set well with me as a useful definition. This means that in a large enough blowup, there could be absolutely no D of F, since even the plane of critical focus can fail to be "sharp", due to "overenlargement". I personally define it as the area in the image that is apparently *just as* sharp as the plane of critical focus. Thus, I view it as a comparison of the plane of critical focus to the rest of the image, rather than a simple definition of what is sharp on the print. Viewing it this way, something can be unsharp, yet still fall within the D of F if it is apparently just as unsharp as the plane of critical focus. Thus, it again comes down to magnification, not just print size. Viewing distance and size of print simply affect magnification (only this time it is your eyes/brain, not a lens/film), which affects D of F.
Distance and focal length combined are magnification.
Originally Posted by 2F/2F
The most important thing about DOF, by far, is that it is a perceptual thing. Not an 'autonomous' entity with an absolute dimension. (Despite all the formulae and calculators people like to let loose on it).
The best definition is that as "acceptable unsharpness".
Very clear about that it is a judgement, a perception, and not a measurement or dimension.
Still pretty vague, since it provides no clue about what sharpness is.
You can increase DOF dramatically, not by changing magnification or f-stop, but by changing how 'sharp' the sharp bit is. A good film and good lens produce shallower DOF than a bad lens on bad film. And vice versa. Simply by changing the difference, the visual contrast between realy sharp and not so sharp.
Another undefined thing is viewing distance.
The final magnification counts, so DOF gets less if you blow a negative up more.
But only if you do not increase viewing distance accordingly.
So a giant print can appear to have greater DOF than a small print of the same negative, if only you view the giant print from relatively further away than the small print.
That's my point exactly. It is explained to beginning students that FL and distance from subject are two of the main things that affect D of F, and not explained that they are actually things that affect magnification, which in turn affects D of F. It is rarely explained that moving to a longer lens alone does not give one less depth of field if you move back from the subject and keep aperture and subject size the same in the composition. It affects perspective (thus affects how depth is rendered; how closer objects appear in relationship to farther-away objects), but not D of F. In order for a change in focal length alone to affect D of F, the size of the subject in the viewfinder must also change. Same with simply changing distance. It only affects D of F if the change in distance also affects magnification; for instance, if you move closer to your subject while keeping the same lens on the camera. If you change to a wider lens when you move closer, such that you keep the size of the subject the same, however, D of F does not change. If it was simply explained that magnification is what is really controlling D of F, and that FL and distance are what is affecting magnification, I think it would be easier to understand and to apply.
Originally Posted by Q.G.
I also agree about the other paragraphs, most importantly the part about D of F not being an absolute.
I view D of F as the area which for ones intents and purposes appears as sharp as the plane of critical focus, not simply as the area which appears acceptably sharp. To use your term: the "acceptable unsharpness" matches between the plane of focus and the areas that simply appear sharp.
Like you said, in the same way that FL and distance from subject indirectly affect D of F in the camera by affecting magnification, print size and viewing distance indirectly affect D of F by affecting magnification when looking at the pix. A tiny print has less D of F under a loupe. A big print has more D of F from across the room.
BTW, if you want to get into depth of field scales, there are some informative Internet (and print) articles, and forum discussions. In short, they are often, if not always, "incorrect" for "critical work". My favorite article is the Ken Rockwell one, in which he briefly covers D of F and diffraction, and then explains how to get the sharpest shot possible for a given situation, considering both issues. It is especially helpful for medium and large formats.
As pointed out above, there is no doubt that depth of field is based (to a significant degree) on a perceptual judgment.
However, it is possible to define an objective scale. Here is an example of how one might do it.
First, define a normal viewing distance. There are lots of choices. Pick one.
Then define a size for the print. There are lots of choices. Pick one.
Then define an amount of allowed "blurriness." There are a number of choices. Pick one. For example, one might base the choice on what amount of blur is barely perceptible by a person with good vision when viewing and the object at normal viewing distance. This amount of blur would correspond to a certain spot size for a slightly out-of-focus point.
Then, calculate the distance fore can and aft of a well focused object that would produce an out-of-focus point of the diameter determined above. This involves a chain of calculations or measurements. In the best case one could do this by a theoretical calculation, assuming a perfect lens. An imperfect lens will always perform worse than this.
That will give you a depth of field.
Is this the only possible approach to defining depth of field? No!
Another approach would be to base the calculation on the minimum spot size of the lens in question. This spot size is limited by aberrations and diffraction. Then, one would arbitrarily pick a factor, let us say 1.4 or so. (Don't like that factor? Then pick another one.) The depth of field would be determined by determining how far fore and aft of an object would allow an object point to be imaged to a size of 1.4 times the well-focused limit.
Both of these approaches can use objective criteria. The criteria may be arbitrary, but they can be well defined.
The first will produce a result that depends, among other things, on the ratio of the photo size (linear dimension) to the viewing distance. If you decide to change that distance (strictly speaking, that ratio), then the depth of field changes.
The second method is independent of factors such as magnification, viewing distance, photo size, etc. I figure that in most cases the second method will give a shallower depth of field than the first method.
By the way, I am not claiming that these methods are those in common use. I only give them to show that one can define depth of field in more than one way, but that the definition can be based on objective criteria. I suspect that most depth of field calculations are based on something closer to the first method than the second.
It's important to realize that the DOF's being a function of format size, is only an artifact of the way apertures are expressed in photography. The f/ratio is a complicated way to express aperture that is used in photography because it results in uniform exposure between all format sizes. The rest of the optics world (microscopy and photolithography are what I have experience with) uses Numerical Aperture to define aperture.
When defined in this more geometrically simple way, DOF-per-given-aperture is not a function of format size. However if cameras expressed aperture in Numerical Aperture, different apertures would not give the same exposure value across different format sizes. A given aperture number would be slower on larger cameras, but would give the same DOF across all format sizes. The F/ratio system is used because, it is felt, that it is more important that a given aperture number give consistent exposure across format sizes than it is that it give consistent DOF across format sizes.
In microscopy exposure isn't as important because you can just turn the light up a bit, but being able to calculate DOF for any possible magnification is very simplifying. Since exposure is important in photography, photography uses the F/ratio to express aperture, which takes into account the combined effect of aperture area and the inverse way that light falls off as the focal length increases. Using this system, f/8 gives the same exposure on 8x10 as on 35mm, but the DOF is much different between the formats.
Now, the way that DOF itself is quantified using certain criteria of unsharpness and so on, is another matter, but I hope that explains a bit about why DOF varies with format size in photography.
What the dpreview guy says is true! The smaller the format the larger the depth of field at a given aperture.
I agree with what BetterSense says, except for one quibble, which is that the term "Numerical Aperture" is not correctly used by BetterSense in this context. Numerical Aperture is essentially equivalent to f-number in the sense that both numbers are ways of expressing the angular aperture of a lens.
The f-number is twice the cotangent of the half angle that defines the angular aperture of the lens. (This is strictly true only if the principal surfaces are planes.) The Numerical Aperture is the sine of the half angle.
In the small angle approximation sin(angle)=~1/cot(angle), hence 2*f-number=~1/numerical aperture.
At lower f-numbers this approximation breaks down.
Someone can check my math, to make sure I have the factors of 2 placed correctly, but in any case I believe the functional relationships are correct.
I think what BetterSense meant to say was that the depth of field depends on the lens diameter, not the numerical aperture. (I'm overlooking some subtleties here, such as telephoto or retro-focus designs, in which the physical lens diameter does not equal the effective optical diameter.)
Now here's a rare opportunity to reply to something, saying that it is just semantics, and be correct (and not in disagreement, by the way).
Originally Posted by 2F/2F
"Lots of choices. Pick one." "Define" (in the "pick one" sense). "[...] perceptible by a person"
Originally Posted by alanrockwood
And still "an objective scale"?
The second approach, based on arbitrarily chosen ("let us say") dimensions, does no better.
As you say: "The criteria may be arbitrary"
And there the whole things come falling down. DOF is not an objective thing.
A statement like "The second method is independent of factors such as magnification, viewing distance, photo size, etc." even cannot be a thing related in any way to DOF.
DOF, in essence, is dependent on factors such as magnification, [etc.]