Talk:Depth of field/Archive 2

Fancy Italic “f”

Dick, in this context, f is a quantity symbol, and as such, should be set in italics (see ISO 31-0 or NIST Special Publication 811). ANSI and ISO standards all follow this practice. JeffConrad 07:45, 25 January 2007 (UTC)

I don't find where it says that f in f-number is a quantity symbol. I thought it was just a name. Can you point me more specifically? And what about the long hooked f as used in f/#? Is that part of the standard? Dicklyon 02:57, 26 January 2007 (UTC)

I don't think there is any explicit statement, but it seems obvious that f is the symbol for focal length. ASA PH2.12-1961, Sect. 3.4.2, states

The symbol for relative apertures shall be f/ followed by the f-number.

The same statement appears in ANSI PH3.49-1971, again in Sect. 3.4.2. At least to me, it seems obvious that f/# indicates 'focal length divided by number', with f the symbol for focal length. I don't think there's really a long hooked f—it's simply an italic f. That this is so is illustrated in ISO 2720-1974, Sect. 3.1.2. ISO standards are set in sans-serif type; hence, f is simply in sans-serif italic (Actually, it's just oblique, because an italic front is both oblique and cursive. But this is getting a bit pedantic ...). JeffConrad 04:29, 26 January 2007 (UTC)

Jeff, my reading of the history is different. Long before the 1974 standard, the typographical standard of the long hooked f, not the oblique f, had been adopted for f-numbers as in f/8 where the f is literally focal length, but not at all standardized for the expression "f-number" where f is part of a name. I studied an awful lot of old books on this. I'm surprised to hear that the ANSI standard has an ordinary oblique f for both. Very strange. Dicklyon 04:58, 26 January 2007 (UTC)

Well, looking at books again, I must say the typography is much more varied and less regular than the simplified memory I just described. Still, I see no evidence that the f in f-number has ever been taken to be a quantity as it is in f/8. Some use a specifically different symbol, like F-number and f/8 (The Eye and Visual Optical Instruments By David A. Atchinson, George Smith). Dicklyon 06:02, 26 January 2007 (UTC)

Dick, your knowledge of photographic history far exceeds mine. No disagreement that actual practice is all over the map. Logically, though, I can't see how f-number derived from anything other than the quantity symbol for focal length, so I'd have to say that the folks on the ANSI and ISO committees got it right. They're also arguably fairly authoritative. Sidney Ray and Warren J. Smith also follow this practice, and they're fairly authoritative sources. I still wonder whether the long hooked f was anything other than just italics (which often produce a long hooked f, at least in serif typefaces). JeffConrad 06:31, 26 January 2007 (UTC)

Like this: $f\mathrm {-number}$. JeffConrad 09:40, 26 January 2007 (UTC)

Principal planes

And principle planes are not the same as principal planes ... JeffConrad 22:13, 14 March 2007 (UTC)

Aperture diagram (edit of 19 March 2007)

I agree with Girolamo—having the aperture diagram in the aperture article is more than adequate. JeffConrad 19:31, 19 March 2007 (UTC)

Where is distance measured from?

The text below the Nikon lens' photo says this:

A 35mm lens set to f/11. The depth-of-field scale (top) indicates that a subject which is anywhere between 1 and 2 meters in front of the camera will be rendered acceptably sharp. If the aperture were set to f/22 instead, everything from 0.7 meters to infinity would appear to be in focus.

Now my question is: Is the distance measured from the front of the camera? I thought it was to be measured from the point in focus! Please correct me if i am wrong. 165.125.144.16 19:51, 19 March 2007 (UTC)

The measurements on the lens barrel (focusing/DOF scale) are normally taken from the camera's film plane. My old Canon SLRs have a symbol like this: -o- on the top plate of the camera indicating the position of this plane. (Quantities in optical formulas may refer to different origins, like the center of the lens.) -- Coneslayer 20:08, 19 March 2007 (UTC)

The formulae in the DoF article give object distance from the object nodal plane; at moderate-to-large subject distances, the distance between this plane and the image focal plane (i.e., the film plane) usually is negligible, but for close-up photography, the difference can be significant. JeffConrad 20:17, 19 March 2007 (UTC)

The important thing to note in practical applications, such as working as focus puller, is that lenses designed for film cameras give their focus markings from the film plane, while lenses designed for video markets, such as ENG, often tend to have focus marks measured from the front element of the lens. At the end of the day, though, this has no effect on the depth of field itself - only where the focus should be measured from. Girolamo Savonarola 18:10, 20 March 2007 (UTC)

Imperceptible vs. acceptable blurring

Dick, is "acceptable" the right term in the introduction? In this context, I think it serves more to confuse than to enlighten. It would seem to me that in the general sense, "acceptable" could apply to any blurring that met the aesthetic requirements of a particular image, even one employing substantial selective focus. Isn't the concept of DoF that the CoC is a blur spot indistinguishable from a point? I recognize that "acceptable" appears several other places in the article, but in most instances, it means, essentially, "imperceptible". JeffConrad 00:32, 10 April 2007 (UTC)

Perhaps it is confusing in that sense, yes. But while the concept is usually "imperceptible", the actual values used are typically somewhat larger than a just-perceptible blur, perhaps by a factor of two, aren't they? In any case, different people choose differenet CoC values based on what is acceptable to them, like how close to have to look to notice blurring on a big print. Tieing this to perception is a conventional fiction, I think. Dicklyon 01:10, 10 April 2007 (UTC)

I certainly don't see any factor-of-two cushion. A Snellen chart is based on 30 cycles/degree, which is 6.875 cycles/mm at 250 mm. A Snellen chart is very high contrast, so the usual rule of thumb (see Ray 2002, cited in the Circle of confusion article) is to reduce this to 5 cycles/mm (equivalent to a CoC of 0.2 mm) to allow for the reduced contrast of normal subjects. The tools available to Snellen admittedly were limited in comparison to those available today; people at the Smith-Kettlewell Eye Research Institute in San Francisco tell me that resolution on the order of 35–37 cycles/degree to a high-contrast sinusoidal target is more typical. Even so, I still don't see the big cushion when allowance is made for normal-contrast subjects. I've seen some very optimistic claims on the net, but I've yet to see one with a credible explanation or citation of a reliable source.

I don't buy the theory that conventional CoCs are nothing but remnants of 1930s film resolution. The idea behind DoF is the perception of sharpness under normal viewing conditions, and not necessarily the maximum that film or an electronic sensor can capture. If the latter were true, a 4×5 would use the same CoC as 35 mm, and making an outdoor 4×5 image would be all but impossible. I don't disagree with Merklinger's math, but his illustrative examples employ magnifications seldom encountered by anyone but the CIA or David Hemmings. Even absent subject motion, the idea that a CoC can be reduced to any arbitrary value is far greater urban legend than imperceptible blur at DoF limits. Eventually, diffraction overtakes defocus blur, and a few simple calculations suggest that this happens at values not much less than conventional CoCs.

The basic concept of DoF always has been a zone in which everything is perceived as sharp. I agree that the choice of CoC is somewhat arbitrary, but if conventional practice is to be questioned, I don't think the introductory paragraph is the place to begin the attack. For the casual reader, the extra qualification is likely to confuse. JeffConrad 04:06, 10 April 2007 (UTC)

I'm OK taking it out of the lead, but I think it should be admitted that the conventional COC for "sharp" has crept over the years from about 1/700 to 1/1000 to 1/1500 to 1/2000 of the format diagonal. The latter for 10+ MP pixel peepers, of course. Dicklyon 04:29, 10 April 2007 (UTC)

I changed it (differently; tell me if you like it). But then I noticed the first section head is Definition of "acceptably sharp". So shouldn't that notion be in the lead? Dicklyon 06:48, 10 April 2007 (UTC)

Technically, I agree with the "specified" viewing conditions, but practically, I think the use of "normal or specified" still is confusing in the introduction. I've recast things a bit in attempt to avoid the problem; I've also moved the heavy stuff from the intro to the first section to be less intimidating. The intro now is a bit sparse, but everything I tried was little more than fluff. I'm not happy with "misfocus" (in the first section) but I'm afraid that "defocus" would be a bit much for many readers. See what you think ...

This article has been somewhat hijacked by a couple of still photographers; I hope we're still OK with the filming people ...

The topic of "CoC creep" would seem better suited to the Circle of confusion article. Of course, it would be interesting to see why the criteria have changed; I've seen many values cited over the years without much explanation. I don't suggest that the criteria I put in the CoC article are infallible, but at least they can be related to basic principles. For full-frame 35 mm, a 0.03 mm has been around since the 1930s, and is about 1/1440 of the format diagonal. It might derive from the assumption of enlargement to the long dimension of a 8×10 frame. The 1/2000 criterion is new to me ... JeffConrad 09:24, 10 April 2007 (UTC)

I think most of us agree that strictly, we speak of “acceptable” sharpness according to specified criteria. As I've mentioned before, though, this seems a bit technical for the first sentence. Strictly, “area” is probably applicable when applied to a two-dimensional image, but it’s somewhat at odds with the physical concept of DoF as a range of distances from the camera. In any event, reducing the number of words probably improves the readability, and to me, “Apparent sharpness” is less unwieldy as a section title than was “Apparent sharp focus”. See what others think. JeffConrad (talk) 07:08, 25 April 2008 (UTC)

“Portion of a scene” or something similar works for filming and photography but seems less suitable for something like photolithography; of course, “in front of and behind the subject” arguably had the same problem. We could say “the region of object space that appears sharp in the image”, but that seems a bit much for the first sentence. JeffConrad (talk) 22:09, 25 April 2008 (UTC)

Camera Movements and DOF

I hate to add yet another section, but it seems better to introduce camera movements in one place rather than two. I've avoided the term "tilted plane focus" because it isn't standard (try a Google search to see what I mean), but I'm not sure there is a good term: "tilted focus plane" seems to be the winner of the obvious terms with 18 hits ... Good technical treatments of tilt and swing are few and far between, and the authors all seem to use different terminology: witness the term used to describe the axis about which the plane of focus rotates as the lens is focused: "counter axis" (Scheimpflug), "hinge line" (Merklinger) and "pivot point" (Wheeler). I usually call it a "pivot axis" or even "POF rotation axis" ... I've used "rotation of the POF" for the process because it seems most descriptive of what is happening; perhaps there is a more elegant way of stating it. JeffConrad 07:51, 24 April 2007 (UTC)

Diagrams requested

{{reqdiagram}} This article would greatly benefit from diagrams illustrating the various formulas. -- Beland 18:56, 5 May 2007 (UTC)

I've indicated DOF in the diagram in the section 'Derivation of the DOF Formulae', and directed the reader to that section. That diagram illustrates most of the quantities discussed in the basic section, and the picture of the butterfly gives an alternative illustration of the concept. It's certainly possible to provide additional diagrams, but the article is already a bit cluttered. Perhaps others disagree, however. JeffConrad 00:49, 6 May 2007 (UTC)

Edits of 5–9 June 2007

Try as I might, I cannot decipher the paragraph that was added to the Artistic considerations section (originally under Word of caution). Although I can think of several things that may be intended, I am guessing at best. Consequently, I cannot see what this paragraph adds to the article in its current form. Absent a clarification, I am inclined to remove it. JeffConrad 21:26, 6 June 2007 (UTC)

I think he was trying to say that the degree of blur of distant background is not determined by the DOF. But I agree it's a mess and hard to see how to fix, which is why I only took out the section head and waited for you to find it. Dicklyon 00:21, 7 June 2007 (UTC)

Actually, I held off waiting for you to make some cuts :-). If I knew how to fix it, I would do so. My guess was that the intent was to say that the amount of the background or foreground blur doesn't quantify the DOF. Perhaps the intent was simply that foreground/background blur is somewhat of a separate issue from DOF. I'm not sure why we would need this comment, though. When people attempt an attractive background blur, they don't usually give much thought to quantifying DOF (Merklinger's example of blurring a sign is special case). Perhaps the editor (WalrusJR) can clarify, but absent this, I think we should eliminate the paragraph. JeffConrad 01:43, 7 June 2007 (UTC)

I've tried to interpret the edit of 5 June 2007 and make the wording more comprehensible. Although, strictly speaking, the added paragraph is correct, we're really splitting hairs, and I wonder if the distinction merits this much space. I avoided the use of "nicely", because it could be taken to include bokeh. If it is felt important to cover the appearance as well at the amount of background or foreground blur, I suppose yet another sentence could be added, but I would wonder if we're getting too far off topic. JeffConrad 23:34, 9 June 2007 (UTC)

DOF and camera movements—edits of 8 June 2007

There is no maybe about the near and far limits of DOF no longer being parallel when the POF is rotated, as the Scheimpflug principle article or any decent text on view cameras will show.

The meaning of the added paragraph isn't clear. Perhaps the intent is at using swing rather than tilt, but either swing or tilt involves rotation of the POF, so it's not clear what is meant by alternatively". "Width of field" is an invention, even if a clever one, and doesn't seem appropriate for a WP article (should we also say "height of field" when tilt is employed?). If people think this article should say more about movements, it's certainly possible to describe the application of both tilt and swing, but I think such material would be more appropriate in either the View camera or Scheimpflug articles. Even a simple treatment will require several paragraphs and a couple of diagrams, and this article is already a bit on the long side. But whatever people think is appropriate. JeffConrad 20:42, 8 June 2007 (UTC)

Peer review and FAC?

Would any of the primary editors be interested in shepherding this article through the peer review and/or FAC processes? Girolamo Savonarola 20:50, 8 June 2007 (UTC)

If you think it's ready, and we get a couple more votes ... though I think a couple of minor details (discussed in the last few days) should be resolved first. JeffConrad 23:47, 8 June 2007 (UTC)

It can go into PR at any time, regardless of the state it's in. I would do it myself, except that I don't feel as familiar with the optics details nearly as much I'm certain a few people including yourself do. Girolamo Savonarola 00:22, 9 June 2007 (UTC)

Foreground and background blur—edits of 10–16 June 2007

The ratio of background blur to background object size is very similar in concept to Merklinger's object field method, considered on the image side of the lens. The key difference, at least in the context of this article, is that, within the DOF, the blur is imperceptible, so although the ratio still holds, it is irrelevant. In the object field method, the ratio would still be important.

The difference in concept could be regarded largely as a matter of viewing conditions; most examples showing the benefits of the object field method are at much greater magnification than would normally be employed. Alternatively, the difference could be regarded as what the imaging medium is capable of recording rather than what is noticeable under standard "normal" conditions of enlargement and viewing.

A brief mention of some or all of this could be added to the article if people think it is important. However, unless lack of mention is likely to prompt the question, I'm inclined to omit it because the article is already quite long. JeffConrad 04:48, 11 June 2007 (UTC)

I should have mentioned that the comparison in the second paragraph with the object field method applies only to distant objects; although the object field method is scene specific, the criteria for near objects often are less restrictive than the criteria of the traditional approach. JeffConrad

I've restored the comment on the ratio of background blur to imaged detail size because this also is objectively true; I've left out the interpretive comment, however. I also removed the interpretive comment from the Background and foreground blur subsection under derivation.

1. Without mention of the ratio, there is a concerted effort to claim that a longer focal length produces greater blur, which doesn't seem NPOV and may or may not be true. No disagreement that absolute blur is proportional to focal length, but to mention only that seems a selective inclusion of facts.
2. Few backgrounds, including those in most of the images in the article, are at infinity, so concentrating on the effect at infinity is of questionable relevance to to most photographic situations.
3. Omitting mentioning the blur to detail ratio while retaining the link to van Walree's article is potentially confusing, because it's exactly what he discusses.

Using "detail" rather than "object" also seems less precise (the behavior applies to any object at that distance), but I've left it for now. JeffConrad 23:08, 11 June 2007 (UTC)

Strictly speaking, the pupil magnification cancels out of the expression for blur spot diameter if the distance is measured from the lens entrance pupil rather than from the object nodal plane. However, we've expressed all other formulae in terms of the distances from the nodal planes, so it would seem reasonable to do so here as well; I used "lens" in attempt to avoid the issue altogether. If people insist on referring to the entrance pupil here, I suppose I won't quibble, but I think by doing so we would confuse more than we would enlighten. JeffConrad 01:43, 12 June 2007 (UTC)

Jeff, you restored part of what I took out, leaving "For a given subject magnification, detail distance, and f-number, the degree of detail blur is proportional to the focal length." It's very hard for me to imagine what could be changing here with focal length if you're keeping subject magnification and background detail distance fixed; you'd have to be moving the background relative to the subject as you change focal length and move the camera to keep subject distance fixed. So I can't really find any truth in it. That's why I reduced it to only the special case of an infinitely distant background. And the next bit "however, the magnification of the detail is also proportional to focal length, so for a given detail, the ratio of the blur disk diameter to imaged size of the detail is independent of focal length," that's also not true, since m is not proportional to f in general, and it's introducing a concept that might make sense to Merklinger, but not to anyone else. I think we should drop it. Dicklyon 04:18, 12 June 2007 (UTC)

Dick, the magnification of the defocused object changes with focal length:

$m_{\mathrm {d} }={\frac {f}{u_{\mathrm {d} }-f}}$

For $u_{\mathrm {d} }\gg f$, the magnification is essentially proportional to focal length. The ratio of blur spot size to imaged object size is constant with distance, for all distances. The proof, which I put in the derivation section, is very simple.

The only constraint that I restored was that of constant subject distance, emphasizing that the relationship between focal length and blur diameter applies at all background distances rather than just at infinity. Most backgrounds aren't at infinity, so I think this is important.

You mean constant subject magnification, yes? That's what it says, anyway. And in this case, the conventional approximation is that the DOF is nearly independent of focal length, meaning the blur diameter is constant, not proportional to f, at those distances at the edge of the DOF. I realize that's not exact, but it's certainly nothing like proportional to f. One of us is confused. Dicklyon 07:08, 12 June 2007 (UTC)

The only reason that background blur spot size is of interest to the average photographer is for isolating the subject from a busy background, and I believe that is what led us to add the section on background blur. The subject is isolated from the background because the background is blurred and the subject isn't; what's not so easy to determine is how blurred the background must be for adequate isolation of the subject. It seems a general rule of thumb that when the blur spot is approximately the same as the imaged object size, the object is difficult to recognize. I'll admit here that I'm relying on John Williams's Image Clarity, and he's really a secondary source—I haven't directly consulted the references in his bibliography. In any event, the concept didn't originate with Merklinger; where he differs from most others is that he applies the concept within the DoF, where most others (including me) consider the blur undetectable, at least under normal conditions of enlargement and viewing. Merklinger's method well may be appropriate for surveillance photography, where the main objective is in identifying objects, employing the maximum usable enlargement.

Certainly, reducing recognizability of the background plays a part in determining isolation of the subject, though I don't think it's as clear cut as Merklinger's example of the "PROWLER" sign on his neighbor's RV, i.e., I don't think it's essential that the background be completely unrecognizable. However, I think it's just as speculative to imply that absolute blur size is the only criterion as it would be to insist that a background object be unrecognizable. Van Walree's examples illustrate this quite well; to me, the image with the long-focus lens appears better separated from the subject, but I'd maintain that it's because of the narrower angle of view than the amount of background blur. This is entirely consistent with my experience in macro photography with long-focus lenses—though I like them primarily for working distance, they're also mighty handy for hiding a busy background simply because they include less of it. I never gave much thought to greater background blur, because I've never been convinced there's much subjective difference compared with a lens of shorter focal length.

In summary, if we're going to have a section on foreground and background blur, I strongly think we should mention the ratio of blur to imaged object size. I don't suggest that it's always the governing factor, but I think it's often as valid as absolute blur size, and to mention one without the other would be very disingenuous and misleading. I originally included the subjective comment just to make clear that neither criterion was necessarily absolute. We could add a comment about differing angles of view; although it's a bit off the topic of DoF, it's probably no more so than many other comments in the article, and in any event, it's an important consideration in isolating a subject from the background. JeffConrad 06:55, 12 June 2007 (UTC)

OK, but let's clear up the objective part that separates us first, and reference Williams on the other.

What is the objective part on which we differ? That should be simple to fix. As for referencing Williams, I'm not sure that's appropriate, because the criteria he cites are quite specific, and relate to legal criteria for identification of suspects and admissibility of evidence. I don't suggest that it's anywhere near that clear cut, but simply that the blur disk size, without consideration of its relationship to the object itself, isn't the entire picture when it comes to isolating a subject. Isolation of subject from background is necessarily subjective; although blur disk size ostensibly is objective, its relationship to background isolation is the only justification for its inclusion in the article. Perhaps a one-sentence comment would adequately cover what I had said in three or four sentences. JeffConrad 07:44, 12 June 2007 (UTC)

I've made a few subtle but important changes to the text; in particular, I've eliminated reference to "proportional". I've also eliminated the "however" in introducing the ratio of blur to object size, so the presentation should be more neutral. Everything in the section is objective, and now, hopefully, strictly correct. Although it might benefit from a sentence or two mentioning why the two measures of background blur were included, I think the material could stand as it is. See what you think.

I've expanded the treatment in the subsection under Derivation to support the claims of how the blur disk changes with focal length. JeffConrad 12:42, 12 June 2007 (UTC)

Thanks. The "proportional" was the main problem, since it didn't account for the variation in s and D. Dicklyon 14:38, 12 June 2007 (UTC)

I am afraid that the present text ("For a given subject magnification, f-number, and distance between the detail and the subject, the degree of detail blur varies with the focal length (varying s and D together as focal length is changed).") is not entirely clear. Somehow the article should say in words that the influence of focal length on the blur disk diameter is small when D is close to s, but sizable when the two quantities are well separated. Odo Benus 17:49, 12 June 2007 (UTC)

Yes, I agree. It's what the equation implies, but it's not yet clear, and it really is the point. For D not too far from s, the D in the denominator nearly cancels the f and there's not much effect, but for D much greater than f, the D in the numerator starts to cancel the one in the denominator, leaving the f to make a difference. Maybe a plot or examples would help. Have a go at it. I'm still not convinced that Jeff's focus on the size of background elements is very relevant to the objective issue here. Dicklyon 18:01, 12 June 2007 (UTC)

The ratio of blur to object size is every bit as objective as the absolute blur; it's just a different criterion. Whether one criterion is better than the other at "isolating the background" is necessarily subjective. I don't think I've ever seen a definitive authoritative statement that quantifies what isolates a subject from a background; van Walree makes as good a case as I've seen. I'd arrived at the same conclusion mathematically, but his images make for a more convincing presentation. One advantage of the ratio is that it's independent of focal length and subject distance, depending only on the defocus of the foreground or background. I don't suggest that it's the "right" criterion, and don't really care whether or not I can read a license plate—I just want the background not to be distracting. There simply is no way to make that an objective call.

I find the statement "varying s and D together as focal length is changed" a bit confusing, because we really aren't actively varying $D$. Assuming we're dealing with given scene with fixed positions of subject, foreground, and background, $D$ varies with $s$ simply because the its distance from $s$ is fixed. It may be easier to illustrate this by calling the distance between subject and foreground or background object the defocus, given by

$x_{\mathrm {d} }=\left|D-s\right|$

We then could speak of fixed $x_{\mathrm {d} }$, and

$b={\frac {fm}{N}}{\frac {x_{\mathrm {d} }}{s\pm x_{\mathrm {d} }}}$

Using the ratio of blur to object size,

${\frac {b}{m_{\mathrm {d} }y}}={\frac {m}{m+1}}{\frac {x_{\mathrm {d} }}{Ny}},$

which clearly shows the dependence on defocus and object size. I don't suggest that this is the "right" criterion, but simply that it may be as valid as any other.

It's certainly possible to add a plot, but with all due respect, of all of the many quantities that could be plotted, this would seem far from the top priority. I'm not even sure what a plot would really illustrate. It seems to me the implied question is, for a given scene and subject that I want to photograph at a certain magnification, what lens will give the best background blur? If the absolute blur is the criterion, the longer focal length certainly will give the best blur. But why would I want to blur the background? Presumably, to isolate the subject, and toward that end, I'm not sure the objective criterion is that obvious. Images make the case far better than formulae or plots, as van Walree's treatment shows. The effect of defocus distance could be illustrated with examples somewhat more subtle than van Walree's. I would not be surprised if the long-focus lens gave better isolation, especially with increasing defocus. However, I think the result would be due as much to the narrower angle of view than anything else. There would be one obvious result—it's nearly impossible to isolate a subject from a nearby foreground or background. But isn't this almost self evident? There already is a statement to this effect in the section on Artistic considerations.

Photographing the same subject with different focal lengths will result in differences in blur spot size, differential magnification with distance, and angle of view; all contribute to isolation of subject from background. I think it's a bit arbitrary, and possibly misleading, to concentrate on one to the exclusion of the others.

But in an article on DOF, it would also make sense to ignore things other than how much blur you get due to misfocus. We can't know what size background objects there will be, or whether more or less magnification of them will be a good thing, which is why I didn't take discussing them to be "objective". I do like your new equation that suppresses D in favor of the separation. Dicklyon 23:07, 12 June 2007 (UTC)

Maybe yes, maybe no. By definition, DOF separates what is perceived as sharp from what is perceived as unsharp; it's unavoidably subjective. We make certain assumptions about visual acuity, enlargement, and viewing to estimate what "appears to be in focus"; if we were truly objective, we'd not be able to arrive at an "acceptable" CoC, and we'd be unable to define DOF at all. Ostensibly, blur diameter is completely objective, but again, we'd have no reason to even explore the topic were it not for the unavoidably subjective issue of isolating the subject from the background—if you will, the degree of separation of sharp from unsharp.

It's easy to crucify a subject on the cross of objectivity. Recall 30 years ago when audio equipment manufacturers focused on minimizing total harmonic distortion, an objective and easily measured quantity, and touted THD on the order of a thousandth of a percent, when it was easily demonstrated that hardly anyone could detect THD on the order of several percent. The obsession with minimizing THD led many manufacturers to virtually ignore other issues such as phase shift and limited slew rate, which were easily demonstrated to have very noticeable effects on perceived sound quality. Like THD, both phase shift and slew rate were objective, easily measured quantities—it was largely a matter of what someone considered important and decided to measure. The situation here may be analogous.

In any event, I've given the "relative blur" only one sentence, and have not said anything that isn't strictly correct. I haven't suggested that any particular amount of "relative blur" is good or bad; I've pointed the reader to van Walree's article, and the reader can decide for herself whether it's relevant. JeffConrad 03:10, 13 June 2007 (UTC)

I also wonder if we aren't getting carried away with this topic (I'm certainly as guilty as anyone) and giving it far more emphasis than it merits. This isn't to say that it's unimportant, but so are most of the other sections. The article is already fairly long, and several comments have been made about the surfeit of images. JeffConrad 22:54, 12 June 2007 (UTC)

Sure we are. That what wikiaddiction does to you. Dicklyon 23:07, 12 June 2007 (UTC)

Dick, the "proportionality" holds only for fairly distant background objects; with the expansion to include foreground objects, it totally breaks down, and I neglected to adapt for the change. One way to use "proportional" in the restricted sense and still cover backgrounds short of infinity might be to say that, for a reasonably distant background,

$b\approx {\frac {fm}{N}};$

we rely on many similar approximations to arrive at the simplified equations. JeffConrad 22:54, 12 June 2007 (UTC)

Yes, that's the thing to do. I figured it was obvious when I wrote the version that only mentioned the infinitely distant background, but you're right that being explicit about the approximation is a good idea. Dicklyon 23:07, 12 June 2007 (UTC)

I'll make that change. Upon further thought, Odo Benus's suggestion to make the comment in the text probably is the simplest approach—one sentence should suffice. I'll also add the defocus distance that I mentioned and see people think. JeffConrad 23:40, 12 June 2007 (UTC)

Well, I took two sentences, but in them I also stated some assumptions that may have been obvious to us but not to the casual reason. I hope the current form addresses Odo Benus's concerns; I don't show the basis for the claims, but I don't know how to do so without substituting for the subject distance, and I think that would be getting pretty complex for the "basic" formulae section. In any event, see what you think. JeffConrad 03:10, 13 June 2007 (UTC)

Jeff, thanks. I think it's much better now. Dicklyon 04:11, 13 June 2007 (UTC)

Yup, it is much better now. However, the sentence "Outside the DOF, the blur increases with the distance from the subject." may confuse people if they interpret "Outside the DOF" as a condition for the blur increase. For the blur also increases with the distance from the subject inside the DOF. Odo Benus 16:33, 14 June 2007 (UTC)

Hopefully, the last change will eliminate the potential confusion. The case could even be made for giving the ratio $b/c$, with the implication than when $b/c<=1$, $b/c=1$. One could describe the blur in terms of CoCs; such a ratio would format independent. I'm not sure that it would add much in this context, though.

On unattached participles: put "$b$" after "differentiating" if you must, but I don't think it would add much—there is no ambiguity about what is being differentiated. I had originally thought of using "differentiation", but, like many -tion words, it seemed unnecessarily abstract. Use of the participle in this context in mathematical texts is common. The objection to unattached participles usually derives from ambiguity; that certainly isn't the case here. JeffConrad 23:28, 14 June 2007 (UTC)

Ambiguity arises with wrongly attached participles. Your participle is unattached in a "sentence" without a noun. It is true that dangling participles commonly appear in mathematical texts; that is why the author instructions of many scientific journals expressly warn against them. Anyway, there are probably more important things to discuss. Odo Benus 11:31, 16 June 2007 (UTC)

Many expressions like this have long been accepted usage; it could even be argued that "differentiating" is a gerund in this context. In any event, let's not quibble about whether this particular usage is acceptable; I've added the "$b$", which is also common form for mathematical texts. I've also added a sentence explaining the significance of the sign of the derivative; if it's superfluous, get rid of it. JeffConrad 21:47, 16 June 2007 (UTC)

Not all limited-DOF photographs involve separating flowers or cats from distracting backgrounds. Upon further thought, it seems to me that some mention needs to be made of the recognizability of objects, as much for situations in which an object must be recognizable (e.g., evidence and surveillance photography) as for situations in which they should be unrecognizable—I've made many photographs of the former type myself. It would be nice to look at one of Siljander's books, but absent that, Williams probably would suffice as a reference. I'm not sure the discussion belongs with the formulae, however.

On a related note: would the section "Artistic considerations", which initiated the fascination with foreground and background blur, be better titled "Selective focus"? This isn't to suggest that artistry isn't involved, but that "selective focus" is the common term (at least in the U.S.A.), and its use makes content of the section more obvious to a reader scanning the article. Also, is "differential focus" sufficiently common in British English that it should be mentioned as a synonym? JeffConrad 00:17, 15 June 2007 (UTC)

Good idea on the new section title. I hadn't heard of "differential focus", but looks like it is pretty common. Thanks again for all your concientious work on this article. Dicklyon 03:52, 15 June 2007 (UTC)

Chabacano's new figure

This figure is interesting. Mighty large, and not particularly helpful at illustrating where the size of the blur circle comes from, since the crossing of rays in front of and behind the focal plane is a tiny hidden detail. For now, I'll just shrink it a bit and copyedit the caption. Dicklyon 17:29, 15 June 2007 (UTC)

I tend to agree; the diagram is certainly more attractive than mine, but I'm not sure that it really illustrates what's happening. When the request for additional diagrams was made, I had thought of adding a couple of diagrams similar to the one in the Derivations section, but less cluttered. This could be added if people think it it would help.

The captioning illustrates once again the rather casual alternation between DOF and background blur. It's hardly unique to here; a quick review of every photographic text that I have confirms that for good or ill, it is almost universal practice to speak of employing shallow DOF (limited DOF/selective focus/differential focus) to isolate a subject from the background. Consequently, I think that's the terminology we should use. Though DOF and blur are different quantities, shallow DOF and background blur are really two ways of looking at the same thing. What usually isn't stressed is that the amount of background blur also depends on the distance between the subject and background. I'm working on a way to say this succinctly; perhaps a diagram also would help, though getting a reasonable scale might be tough because of the horizontal space required to illustrate the distant background. I'm also mindful of the comments about the clutter of too many images; I assume that, at some point, this would apply to diagrams as well. JeffConrad 19:48, 15 June 2007 (UTC)

Whatever the figure's merits, I find it rather confusing in the subsection "Focus and f-number". Logically, it should be close to the section "Effect of f-number", but without some further shrinking, I can't find a graceful way of putting it there. JeffConrad 22:07, 16 June 2007 (UTC)

Hi, I drew this figure because I tried to understand the relation between aperture and DOF and after browsing the internet some time I did not find good simple figures and explanations for it, so finally I did this draw for es.wikipedia, and eventualy placed it here too. I thought that an isometric diagram would be better, because I have noticed that most of people just skip standard optical diagrams and formulae, as they find them obfuscate, or hard and then they just memorize recipes like "big aperture->small DOF" as if it was product of magic :) (that is the only explanation I can came up to explain why so many photographers do not understand what is going on with this things apart from "reducing aperture will increase depht of field" when the try to explain it in their blogs or home pages). If you have improvements, please tell me and I will modify the diagram. Maybe the crossing lines in front of and behind the image plane should be emphasized in the caption, or maybe a mark (a small black circle) could be placed to emphasize them. I also dislike it in the focus and f-number section, but I couldn't find any better place. Finally, if you think that would be better to remove it, do it without hesitation.

Also, if you have more ideas for diagrams I would be happy to draw them. Thanks for the copyedit. Chabacano 03:40, 17 June 2007 (UTC)

If you could work on clarifying where the rays intersect the focal plane, and make the fuzzy spots more disk-like, it might help clarify where the circle of confusion comes from. Thanks for offering. Dicklyon 03:56, 17 June 2007 (UTC)

I've made yet another copy edit in attempt to say what the figure really shows; if we're not there, I think we're certainly closing in. I'm assuming that we really don't care whether how aperture size is decreased. I agree with Dick's comment, but I think perhaps the greatest value of the figure is in showing that the spade, heart, and club all appear sharp with the smaller aperture. JeffConrad 07:45, 17 June 2007 (UTC)

I was thinking in adding this closeup of the details, although maybe it just adds complexity. I tried to make the disc thing by increasing the size of the blurred spots. It is less real (now the blurred spots are "bigger", not only blurred), but maybe is more comprehensive and no noticeable. What do you think, better or worse? Thanks for your comments. Chabacano 10:13, 17 June 2007 (UTC)

The closeups certainly are illustrative, but they do add complexity, as you noted. The depiction in the main figure now is much better, though—I think if you simply extended the red rays to meet, the illustration of blur would be more than adequate, especially when viewing the full-size figure. JeffConrad 19:25, 17 June 2007 (UTC)

Accuracy

The diagram discussed above appears to be slightly inaccurate. The circles of confusion are, in the SVG, drawn as blurred ellipses. I'm not sure what convolution kernel the SVG standard calls for for "blur", but it looks like it's probably Gaussian, not a circle. The circle of confusion should be drawn as a lighter circle not as an amorphous blob. —Ben FrantzDale (talk) 17:20, 26 January 2008 (UTC)

Circles?

I'm not well versed in optics and am reluctant to tinker with this excellent article. But something strikes me as an oversimplification. We read: Precise focus is possible at only one distance; at that distance, a point object will produce a point image. At any other distance, a point object is defocused, and will produce a circular image (my emphasis). I believe that this is only true if the limiting aperture (in photographic applications, normally the iris diaphragm) is circular. In Image:Diaphragm-detail.png the diaphragm consists of a number of arcs that approximate a circle fairly well; however, in plenty of cameras that I have encountered -- notably early examples of "shutter-priority" automatic exposure (the photographer sets the speed, the camera sets the aperture) -- this is not so, and out-of-focus point sources of light are rendered as blurry octagons or even pentagons.

The crudest diaphragm would be triangular, and somewhere on the web there's an article that shows what happens to out-of-focus light sources via a triangular diaphragm.

And with mirror lenses you'll have what approximate to annuli (rings), though of course these are indeed circular. -- Hoary 00:53, 12 September 2007 (UTC)

Good point. The assumption of circular aperture is the conventional simplifying assumption, and we should just say so. Dicklyon 00:57, 12 September 2007 (UTC)

<homer>Mmmmm... Donuts of Confusion...</homer> -- Coneslayer 17:52, 12 September 2007 (UTC)

Link to olympuszuiko.com

In a sense, I guess the link is from olympuszuiko.com, though it's actually to a personal blog by the registrant of that domain. I think we already have plenty of these links—we're starting to approach link spam. If someone disagrees and thinks the link should be restored, though, I won't quarrel over it. JeffConrad (talk) 23:30, 11 December 2007 (UTC)

Diagram I made

I made a diagram to replace the fence and flower examples. I don't want to just replace it since it seems like a large change to the page, and I don't know if this is better. The fences don't really help a lot, and here it is easy to see the difference, and I have f/32 as well. I have 2 versions, the rolled-out and the animation, I don't know which people like more. They are of the same images, but the gif had to be indexed. Only one of these should go on the article, of course.

DOF at various apertures, from f/2.8 to f/32 using a macro lens at the minimum focus distance.

Hustvedt (talk) 02:39, 19 December 2007 (UTC)

I can't make much sense of the pictures. Without a familiar subject, it's hard to get a good sense of the depth. Dicklyon (talk) 07:14, 19 December 2007 (UTC)

Hmm, that is quite true. It is more apparent that it is a series of the miniature Christmas tree lights when the aperture is at f/32, so would it be better if I reversed the order of the images, or should I look for something else that would be better? Hustvedt (talk) 17:34, 19 December 2007 (UTC)

I don't think reversing the order would make a great deal of difference. I think the flower images are fine; the biggest objection to the fence images is that the DoF differences aren't quite as obvious as we might like. I agree with Dick that a reasonably familiar subject is essential. JeffConrad (talk) 23:18, 19 December 2007 (UTC)

Ok, how about this? I went for as simple as possible, and here I have a ton of contrast to make it easy to see what is going on. I hope you guys like this version. I only made an animation because the tall version is just huge. Hustvedt (talk) 02:56, 20 December 2007 (UTC)

This diagram illustrates some interesting behavior, such as the light and dark inversion at small f-numbers, but this is pretty esoteric for the average reader. Again, a more familiar subject (preferably, non-macro) would be much easier to comprehend. The two flower images sufficiently illustrate what happens with macro images. JeffConrad (talk) 20:44, 20 December 2007 (UTC)

New section Obtaining maximum DOF

Although the purpose of using the hyperfocal distance or the object field method is probably obvious to most of the major contributors, it may not be so to the average reader. Hopefully, the new subsection clarifies the purpose. I eliminated the reference to bokeh because it really doesn't make sense in the context.

I think this section should also have a subsection on Zone focusing. Although zone focusing often is used to describe the “zone” of sharp focus that obtains from a particular focus and f-number, it is also possible to determine the focus and f-number from the required zone, as is discussed in §7.4, Focus and f-number. I'll see what I can put together if no one has a strong objection.

On an unrelated minor issue: I propose that we replace “formulae” with “formulas” throughout to make the article less pretentiously academic. JeffConrad (talk) 02:20, 12 January 2008 (UTC)

New lead section 11 February 2008

I've made yet another attempt at a lead section; until we get this cleaned up, I don't think the article is a candidate for much of anything. I think there's still plenty of room for improvement, but hopefully this latest attempt will help get the process off dead center JeffConrad (talk) 01:06, 12 February 2008 (UTC)

DOF and Lens Aberrations

I'm not an expert on optics, but I do wonder what is the cause of DOF. My current understanding is that spherical lenses have a type of aberration that causes limited DOF. Perhaps someone who is more knowledgable than I can comment and make appropriate links to the page on lenses, which has illustrations of various types of aberrations. 129.176.151.7 (talk) 14:54, 21 March 2008 (UTC)

The usual approach to DOF ignores both diffraction and aberrations, and just assumes a perfectly converging cone of rays. Put another way, the only aberration considered is the defocus aberration. It's a simple geometry problem then to find out how much the rays are spread for objects at different distances, and to compute the range of distances that keeps that spread within a selected bound. That's all there is to it. Dicklyon (talk) 01:49, 27 March 2008 (UTC)

As Dicklyon says, it's not caused by any "aberrations". It's just a result of geometric optics. If you have a pinhole camera with a big pinhole, it'll make blurry pictures (blurred by the shape of the pinhole). If you add a lens, you'll get a sharp image if it's in focus but you'll go back to the same blur shape if it's out of focus. —Ben FrantzDale (talk) 02:05, 27 March 2008 (UTC)

Betacommand edit of 26 March 2008

I don't consider the Cambridge in Colour tutorial link spam. It's been in this article for quite some, and none of the substantive contributors seems to have had an issue with it. Unless someone has an objection, I think we should restore the link. JeffConrad (talk) 21:48, 26 March 2008 (UTC)

that site in question has been repeatedly mass added by the webmaster and others working for that company since 2006. Recently the spammer has been active again. due to the aggressive and forceful method of the spammer their links have been removed countless times. β^_command 21:52, 26 March 2008 (UTC)

I agree. I have removed their links from numerous articles myself; it seems too spammy. Dicklyon (talk) 22:24, 26 March 2008 (UTC)

35mm MP area

For a 35 mm motion picture, the image area on the negative is roughly 22 mm by 16 mm (0.87 in by 0.63 in). 

That's the camera aperture for Academy (1.37:1). While that may indeed by the image area on the negative, almost no one shoots with that as their intended viewing aperture (or aspect ratio). Is this really what DOF calculations are based on? [Also, it seems weird to give apertures to only two decimals, since normally they are given to three]. Thanks. jhawkinson (talk) 21:44, 3 May 2008 (UTC)

For DOF calculations, you don't need much accuracy in the format size or CoC. If the standard frame is cropped a bit, it doesn't change the format size or CoC criterion or DOF enough to be concerned about. Dicklyon (talk) 23:03, 3 May 2008 (UTC)

The format may have some marginal effect on the DoF mainly because of the method of projection and the fact that most motion picture theaters use a constant screen height and adjust the width of the sides. This means that an Academy ratio film uses the least screen area but almost all of the frame area, while a 1.85 film will crop a significant portion of the frame in the projector, but require a much larger screen area. Therefore, the magnification of a frame will vary within an identical venue depending on which format is being projected. The magnification effect will also be determined by the absolute size of the screen and the distance of the viewer, though, so I would hazard to guess that the CoC number for motion pictures is determined to tolerances able to accommodate many of these parameters where probable. Girolamo Savonarola (talk) 06:09, 4 May 2008 (UTC)

Effect of lens aperture: entrance and exit pupils

I'm all for being strictly correct, but I think we'll simply confuse most readers if we discuss pupils at this point. Except for extreme closeups with highly asymmetrical lenses, the difference in pupil sizes/locations has almost no effect, and I think we quite reasonably confined discussion of pupillary magnification to a separate subsection. JeffConrad (talk) 00:47, 8 June 2008 (UTC)

I agree. I also left you a comment on your talk page before I noticed that you commented here. Dicklyon (talk) 01:05, 8 June 2008 (UTC)

Exposure compensation

I take issue with the text

"For example, if a 35 mm camera required f/11, a 4×5 camera would require f/45 to give the same DOF. For the same ISO speed, the exposure time on the 4×5 would be sixteen times as long;".

I understand that the f-ratio needs to be converted between formats. However, the conversion leads to the same aperture diameter, i.e., assuming the same exposure time, the same total amount of light on the sensitive area. While the intensity (light/square mm) is lower on the larger format, the larger format also has a larger light collecting area. The latter cancels out the loss in intensity. Therefore, I don't see why an increase in exposure time (or, alternatively, ISO sensitivity) is necessary.

Either the text is erroneous or it should be amended by an explanation as to why a compensation is needed. Note that obviously a compensation would be needed if the format stayed the same, i.e., to maintain the same EV from f/11 to f/45, obviously the exposure time needs to be ~16.73 times as long, but the above text refers to a change in formats. ClassA42 (talk) 22:29, 10 November 2008 (UTC)

Exposure depends on light intensity, not total flux. At infinity focus, a given combination of f-number, exposure time, and ISO speed leads to the same exposure, regardless of lens focal length or image format. As focus is decreased from infinity, exposure increase is needed precisely because the same light flux is spread over a larger area, decreasing the intensity. JeffConrad (talk) 04:16, 7 November 2008 (UTC)

Isn't the loss of intensity due to the focus decrease the same for both formats? Are you saying the loss is greater for a larger format? ClassA42 (talk) 22:29, 10 November 2008 (UTC)

The format is irrelevant to the required exposure time increase for lens extension; I mentioned lens extension simply to illustrate that when intensity decreases, exposure time must be increased to maintain the same exposure, whatever the format. Again, it's intensity, not total flux. JeffConrad (talk) 01:09, 11 November 2008 (UTC)

Putting it another way, ISO speed is defined in terms of the required photons per area, not total photons. That's why the text includes the qualifier "for the same ISO speed". If instead you used an ideal photon-counting detector of unlimited resolution, no exposure increase would be needed to get the same image quality from a given total number of photons; but photographers never think that way, because film speed is based on intensity at the film; as a result, a given exposure on a larger format leads to a higher image quality (much better resolution relative to photon shot noise). Dicklyon (talk) 06:40, 11 November 2008 (UTC)