DOF determined by focal length, aperture and...?

[jaapv]

Advertisement (gone after registration)

There are different approaches to DOF. Of course getting different image elements acceptably sharp even when shooting from the hip makes the concept very interesting for street and reportage shooters. However, for general photography other considerations are as important.

 

Firstly the subject. Not only does DOF seem to be more deep for low frequency ( large defined objects) than for high frequency ( finely structured) images, the focal length has another effect depending on the way they are used. Mostly images shot with longer focal lengths are single-subject compositions, usually needing less DOF and more subject separation, whilst wideangle shots are often multi-subject or image-in-image type of photographs where we want a modicum of sharpness at different distances. This is aided by the difference in DOF between focal lengths.

Subject contrast is important as well. Low contrast is deeper apparent DOF.

 

All this leads me, not being a street shooter at all, to reject the mathematical approach. I focus exactly on my main subject and stop the lens down to obtain the subject separation I am looking for for that particular image, depending wholly on my experience to judge the effect of aperture, focal length, subject distance and subject matter.

Shoot and look and shoot and look until it is second nature.

 

Of course using the non-mathematical approach presupposes an understanding of the nature of the phenomen, making discussions like this, albeit quite tedious and all too frequent, quite useful.

And forget those silly DOF markings on the lens barrel...

[luigi bertolotti]

Actually, all else being equal, a smaller sensor has less DOF, not more. It sounds counter intuitive but it's not. The increase in DOF seen when using a smaller sensor is created by other factors (moving further away or shooting a wider focal length) in order to get the same framing. At rhe same distance, focal length and aperture a smaller sensor has less DOF. Grab your DOF calculator of choice and try it.

.....

 

EHM !!! I remember very HEATED and LONG threads about this... let's hope for NOT one more...

[01af]

Sensor size. Small sensor = more depth of field

Oh puleeeze! Not again ...

 

Smaller sensor = smaller COC diameter = less depth-of-field.

 

 

Actually, all else being equal, a smaller sensor has less DOF, not more. It sounds counter-intuitive but it's not.

Exactly.

 

 

The DoF scales on lens barrels appear to be identical across the range, also from datasheets. I think they just use the calculated DoF (indeed how would you measure DoF—not very easy to give a precise criterion). Is that true?

No, it isn't.

 

Look at the depth-of-field scales on the barrels of the various Leica M 50 mm lenses and also at the DOF tables printed in the data sheets. Here is the hyperfocal distances at f/16 for several lenses, according to Leica's data sheets:

 

• Elmar-M 50 mm 1:2.8: 5.108 m (inconsistent with scale on lens barrel)
  
• Summarit-M 50 mm 1:2.5: 4.784 m (consistent with scale on lens barrel)
  
• Summicron-M 50 mm 1:2: 5.218 m (inconsistent with scale on lens barrel)
  
• Apo-Summicron-M 50 mm 1:2 Asph: 5.128 m (inconsistent with scale on lens barrel)
  
• Summilux-M 50 mm 1:1.4 Asph: 5.087 m (inconsistent with scale on lens barrel)
  
• Noctilux-M 50 mm 1:0.95 Asph: 5.279 m (consistent with scale on lens barrel)
  

As you can see, no two 50 mm lenses have the same depth-of-field. And moreover, for many lenses the DOF indicated by the scale on the lens barrel is different from the values printed in that lens' data sheet. In the case of the Summarit, my own sample has a DOF scale that is different from that on the sample used for Leica's official product photo. I don't know if these differences are for real or just spurious (e. g. from using different formulas or from different COC diameters or from just being lazy). Anyway—the numbers above as well as the scales on the barrels suggest that the Summarit has more DOF at the same aperture than the Summilux, and in actual pictures this apparently is true indeed.

 

I suppose all the DOF scales and tables are computed, but a lens maker can consider way more DOF-affecting factors for his own lenses than any universal DOF formula can. So I guess the computations are much more complex—and more accurate for the given COC (whatever that may be)—than the usual formulas that mere mortals are using. Still a few inconsistencies are there ...

 

 

I tend to think that the concept as such is firmly established, useful, and reasonably exact for 'standard' lens use [...]. [...] understanding the concept is very useful—especially if you use a rangefinder.

Yes, definitely.

 

 

My point above was that theory is all very well but translating it into actual images is a less exact science.

You're right. Understanding the concept is one thing; applying it to actual picture-making is another. Lots of misconceptions, misunderstandings, and grim debates arise just from not properly making this distinction.

[Jeff S]

Oh puleeeze! Not again ...

 

Smaller sensor = smaller COC diameter = less depth-of-field.

 

I think the (circle of) confusion often arises because the issue is usually not framed (so to speak) in the same way. I think this article explains the issue better in terms of requirements (section titled Depth of Field Requirements) regarding the steps needed (so to speak) to fill the frame equally when comparing sensor sizes. And its accompanying link on DOF provides info regarding various COC issues.

 

At least I think this provides the right perspective (again, so to speak).

 

Jeff

[ismon]

The simple answer to the simple question of what are the three factors that affect DOF, they are : Focal Length, Aperture, and Distance. It always drives me nuts to read a review of the latest wide-angle wonder as having "enormous depth of field". Nonsense. A 15mm lens @ F/4 has the same DOF as a 90mm lens @ F/4 -- when covering the same picture area. The 15 can cover the subject @ 1.5 meters, the 90mm covers the same area @ 9 meters. Choice of lens is a matter of working distance and the perspective of your composition. It's like choosing which golf club is best for a putt or long drive--you don't need a computer.

[gyoung]

Most of that is periferal vision. Look at the screen in front of you. How much are you really looking at - concentrating on? Individual letters? Pixels? Words?

 

A lot too depends on spectacles, if any, varifocals dive a narrowing field of sharp focus as you drop down the lens, so you have to turn your head more. Even so, just like a camera lens, you only focus on one distance at a time. and things are less sharp in fornt of and behind that plane

 

I'll set jaapv off by wondering what depth of field the eye has and how to work it out,

 

my circle of confusion varies and sometimes is infinite....

 

Gerry

[01af]

Advertisement (gone after registration)

A 15 mm lens at f/4 has the same DOF as a 90 mm lens at f/4—when covering the same picture area. The 15 mm can cover the subject at 1.5 meters, the 90 mm covers the same area at 9 meters.

Oh puleeeze! Not again ...

 

The 15 mm lens at 1.5 m focusing distance will indeed capture the same field-of-view as a 90 mm lens at 9 m. But the depths-of-field will differ considerably. The 90 mm lens will have more (!) DOF than the 15 mm lens before the respective plane of focus but less behind it. The loss behind will be greater than the gain before so in total, DOF will be less with the 90 mm lens.

 

I've said it before, and now I'll repeat it once more: The assumption that DOF will be the same when magnification at the plane of focus is the same is accurate enough for practical intents and purposes only when depth-of-field is small in relation to the width and height of the field-of-view.

[pico]

Oh puleeeze! Not again ...

 

The 15 mm lens at 1.5 m focusing distance will indeed capture the same field-of-view as a 90 mm lens at 9 m. But the depths-of-field will differ considerably. The 90 mm lens will have more (!) DOF than the 15 mm lens before the respective plane of focus but less behind it. The loss behind will be greater than the gain before so in total, DOF will be less with the 90 mm lens. [...]

 

As you wrote (and I clipped), the front and rear DOF points will be a tiny bit different for the longer lens. Is this perhaps because with a true telephoto the nodal point is in front of the lens?

[Jeff S]

Note the 'Clarification" section in my second link above for discussion of this concept.

 

Jeff

[01af]

... the front and rear DOF points will be a tiny bit different for the longer lens. Is this perhaps because with a true telephoto the nodal point is in front of the lens?

No, it isn't. It has nothing to do with the positions of the nodal points. Don't think too complicated. Whether the difference in the front and rear DOF points is tiny or huge depends on the difference of the angles-of-view.

 

When the angle-of-view of one lens is small and the other's is even smaller then difference in DOF (at the same magnification at the respective plane of focus) will be anything between tiny and negligible. If one lens has a small angle-of-view and the other has a wide one then the difference in DOF will be anything between significant and huge.

 

It's just basic geometry. When the angle-of-view is wide then the gradient of magnifications along the optical axis will be steep. Make a thumbsketch and it will jump out at you. It's easy.

[AlanG]

There are some complex explanations here using COC, etc.

 

The basic idea behind the depth of field scales was based on statistical testing of subjects viewing 8x10 prints at a normal standard viewing distance with controlled illumination and figuring out on average at what point the test subjects no longer could tell if one object was sharper than another. From this they determined what size circle could no longer be differentiated from a point.

 

So as you change the print size, or how you view the print or screen image, these lines marked on the lens no longer apply. But from the data about human vision you can calculate depth of field requirements (COC) for any lens, print size, and viewing distance.

 

It probably is a little tougher to do on-screen as pixel size and pixel density vary so much.

[SJP]

Here is the hyperfocal distances at f/16 for several lenses, according to Leica's data sheets:

Elmar-M 50 mm 1:2.8: 5.108 m (inconsistent with scale on lens barrel)

Summarit-M 50 mm 1:2.5: 4.784 m (consistent with scale on lens barrel)

Summicron-M 50 mm 1:2: 5.218 m (inconsistent with scale on lens barrel)

Apo-Summicron-M 50 mm 1:2 Asph: 5.128 m (inconsistent with scale on lens barrel)

Summilux-M 50 mm 1:1.4 Asph: 5.087 m (inconsistent with scale on lens barrel)

Noctilux-M 50 mm 1:0.95 Asph: 5.279 m (consistent with scale on lens barrel)

Interesting but, in fact not, as they are all approximately the same from my point of view. Or should I get a tape measure out to determine whether the cat is located at 4.784 meters while taking a picture (belated) of the transit of Venus? They must both be sharp of course. Should I move the cat when using my (non-existent) Noctilux at f/16 (unlikely for multiple reasons)

 

It does not work that way (for me), that setting the lens at 5 m at f/16 gives me a large leeway & that stuff is more or less sharp from something closer than 5m to infinite is useful to know, when in happy snapper mode. Still DoF as a guiding principle is rather useful.

 

Remaining question still is how they actually calculate these numbers. I can not think of an obvious criterion. Or is it just the exit pupil size with respect to the sensor? I must admit that I never really thought much about the subtle distinctions between an ideal lens and a composite (thick) lens. Important of course, but irrelevant for first order guestimates.

 

At least we agree on sensor size being some sort of factor although the sign of the effect seems to be contraversial:D

[01af]

Remaining question still is how they actually calculate these numbers. [...] I never really thought much about the subtle distinctions between an ideal lens and a composite (thick) lens.

That's my point. When being scrupulous then depth-of-field becomes a surprisingly complex thing ... and obviously Leica does not just use the common depth-of-field formulas.

 

 

... irrelevant for first-order guestimates.

Sure.

 

 

At least we agree on sensor size being some sort of factor although the sign of the effect seems to be contraversial

The controversy arises only when not being clear about what is changed and what is kept constant.

 

When reducing the size of the sensor and the focal length of the lens so that angle-of-view remains constant then depth-of-field will become wider. When reducing the size of the sensor only and using the same lens then depth-of-field will become narrower. You cannot change sensor size and focal length and then go around and tell everyone the observed result is caused by the sensor size.

[adan]

Finally, after 30 posts, AlanG finally included the key missing factor:

 

How big the final enlargement will be.

 

You can't pick a CoC until you know what sized CoC will still look like a point (i.e., sharp) at the final amount of enlargement. CoC is "undefined" until print/display size is known.

 

Obviously in an 8-x-10-foot (or 8x10-meter) print there will be visible blur circles that in an 8x10-INCH (or cm) print of the identical image will look like sharp points.

 

Spherical aberration can have an effect on DoF (I almost wrote "apparent DoF" - but that would be redundant: DoF is by definition "apparent.")

 

In this X-section, we can see that spherical aberration (top and bottom examples) spreads out the zone of roughly equal sharpness (band of white) over a wider area in front of and behind the actual focus point.

 

File:Spherical-aberration-slice.jpg - Wikipedia, the free encyclopedia

 

However, SA is a function of aperture - it gets reduced as one stops down and eliminates light rays from the outer edges of the glass.

 

When the the 90 f/2 APO-ASPH was introduced, it was noted that it appeared (important word - "appeared") to have less DoF than the non-AA version, at f/2. Backgrounds and foregrounds dropped off to "fuzzy" more rapidly with the AA version, to some eyes.

[AlanG]

Finally, after 30 posts, AlanG finally included the key missing factor:

 

How big the final enlargement will be.

 

 

I tried to hold back but I couldn't take it anymore.

 

Flash Gordon beat me on post 19 but it didn't seem to register.

 

I'm kind of surprised how much interest there is in dof yet how rarely do camera manuals or photo books explain what those lines on the lenses should mean to users.

 

BTW I may be wrong about this but I think dof tables and markings are calculated on theoretical perfect lenses and thus ignores factors in specific lenses such as CA, resolving power, diffraction, etc. Obviously a soft focus lens won't have the same look as a process lens.

[mjh]

Finally, after 30 posts, AlanG finally included the key missing factor:

 

How big the final enlargement will be.

The standard (and actually quite realistic) assumption of DoF calculations is that the viewing distance is proportional to the diagonal of the print size. In that case, print size cancels out.

[stevelap]

.................this article explains the issue better in terms of requirements (section titled Depth of Field Requirements) regarding the steps needed (so to speak) to fill the frame equally when comparing sensor sizes. And its accompanying link on DOF provides info regarding various COC issues.............Jeff

 

I find Cambridge in Colour a very useful resource too.

 

Handy when threads like this one start going round in circles and become confusing...........(sorry, couldn't rexist;)).

[pgk]

As far as I can see there are two ways in which to consider DoF.

 

First, is when the photographer has either full control over all the viable parameters usable when taking the photograph including subject matter and lighting AND knows to final use of the image, or in scientific applications when the aim point is to produce images with specific information within them. In these cases calculations may be a viability and a thorough understanding of all variables is needed. [This was my background].

 

Second, is when photographs are taken in circumstances which do not allow for calculations or detailed analysis of the situation - the vast majority of images in fact. In these circumstances it is a 'working knowledge' of DoF together with experience which are used to usually utilise a compromise of settings which will yield an optimal result for a variety of uses.

 

The problem here is that the digital revolution has tended to blur the two extremes and it is all too easily forgotten that most practical photography is a visual interpretation of a scene and not a technical dissertation of it. The highly detailed, subtle and complex issues which all go to make up DoF are largely irrelevant when taking most most photographs simply because most of us have little control over them. And at the end of the day its easy enough to compare similar photographs taken on different formats to see which is preferred - if that is, this is of such great importance to an individual.

 

Personally I prefer shooting full frame, can handle a 1.3x crop factor, but don't particularly care for greater crop factors. My suspicion is that this is because I like familiar focal lengths, like fast lenses at their uncropped focal lengths, and have found viewfinders for full frame cameras to be nicer to look through, Some of these factors will change, but oddly enough I haven't really worried about DoF issues from different formats so much. Probably just the way I see things and operate.

[Jeff S]

Finally, after 30 posts, AlanG finally included the key missing factor:

 

How big the final enlargement will be.

 

Discussed in second link I provided some time ago, along with issue of viewing distance, as MJH notes as well.

 

Jeff

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[01af]

Look at the depth-of-field scales on the barrels of the various Leica M 50 mm lenses and also at the DOF tables printed in the data sheets. Here is the hyperfocal distances at f/16 for several lenses, according to Leica's data sheets:

 

• Elmar-M 50 mm 1:2.8: 5.108 m (inconsistent with scale on lens barrel)
  
• Summarit-M 50 mm 1:2.5: 4.784 m (consistent with scale on lens barrel)
  
• Summicron-M 50 mm 1:2: 5.218 m (inconsistent with scale on lens barrel)
  
• Apo-Summicron-M 50 mm 1:2 Asph: 5.128 m (inconsistent with scale on lens barrel)
  
• Summilux-M 50 mm 1:1.4 Asph: 5.087 m (inconsistent with scale on lens barrel)
  
• Noctilux-M 50 mm 1:0.95 Asph: 5.279 m (consistent with scale on lens barrel)
  

As you can see, no two 50 mm lenses have the same depth-of-field.

I just found out that the differences in the hyperfocal distances for all those 50 mm lenses at f/16 are mostly (still not entirely) due to the differences in the actual focal lengths which vary between 50.1 mm for the Summarit and 52.7 mm for the new Apo-Summicron Asph. Another factor taken into actual consideration seems to be pupil magnification. The maximum acceptable diameter of the circle-of-confusion used to compute the DOF tables in the data sheets seems to be the traditional values of 0.033 mm in some tables and 1/30 mm (= 0.03333333 mm) in others.

 

Of course, for practical intents and purposes there isn't any difference between 0.033 and 1/30 ... but when trying to understand how Leica arrives at their numbers in the data sheets then there is.

 

And if now someone feels compelled to make a comment about how relevant this is for taking actual pictures ... no it isn't relevant, everybody knows that, and the practical relevance (or lack thereof) is not the point here. Okay?