A question on hypefocal distances.

[alexn]

Hi everyone.

I'm pretty sure I know the answer is yes, but just to be certain..

In a hypothetical situation, lets say a calculator tells me the HFD for my camera/lens/aperture is 1.3m. That 1.3m measurement needs to be taken from the subject to the film plane?

I have had conflicting answers from here and there, Some telling me its from the subject to the front element, others from the front element to the center of the lens etc..

I was certain it was to the film plane.

Thanks.
Alex.

[Heroique]

In everyday, practical landscaper’s terms, your “1.3 meters” is the distance from the HF point to the camera.

This makes a lot of assumptions, of course. For example, leveling & movements.

To be more exact and comprehensive, this answer will get very, very long, heated, and controversial!

Focusing at this point keeps infinity in acceptable focus, plus everything else that is no closer than 1/2 this distance (.65 meters).

---
BTW, I’ve always noted that AA’s The Camera illustrates this distance as being from the HF point to the front of the lens. This, too, is for simplicity’s sake.

[Lachlan 717]

The theory says from the front glass; however, this assumes a single element of no thickness.

Reality is somewhere around the nodal point, but even this can be hard to define.

If your closest object for focus is not super critical to the (roughly) 65cm point, just assume the front of the glass, thus keeping infinity in focus.

[Helen Bach]

Just to add a little to what Lachlan said: A DoF calculator (and the usual formulae) gives the distance from the lens (specifically the front or first nodal point as Lachlan says) to the subject. If you are using a lens that has a focusing mount it is likely that the distances on the mount will refer to the film plane - subject distance, which is unambiguous because the film plane is usually clearly marked (most movie cameras even have a tape attachment point there to help with measurements). Maybe this is the reason for the different stories you are hearing.

Best,
Helen

[ic-racer]

The hyperfocal distance to the subject for most critical large format systems (small CoC) is far enough away that it does not matter where you measure it. In large format systems I consider the hyperfocal point merely an intellectual curiosity. Best bet is usually to focus on the near object, focus on the far object, split the difference on the standard and set the aperture based on your spread in milimmeters.

[Leigh]

Just to add a little to what Lachlan said: A DoF calculator (and the usual formulae) gives the distance from the lens (specifically the front or first nodal point as Lachlan says) to the subject.
Sorry... wrong.

The DoF calculator gives the distance from the FILM PLANE to the subject.

It has no knowledge of what lens you're using, nor where the front nodal point (first principle plane) might be.
In many cases the front nodal point is not even inside the physical lens. It cam be moved by hundreds of mm.

- Leigh

[Emmanuel BIGLER]

The DoF calculator gives the distance from the FILM PLANE to the subject.
It has no knowledge of what lens you're using

If this is so, then the guy who designed the calculator is simply illiterate in optics and we strongly recommend him to come here more often and learn ;)

Various (compound) lenses exhibit various distances between their nodal (or principal points).
A "caricature" for this can be found in telephoto lenses.
And all classical models for geometrical DOF refer to the cardinal elements of the lens : foci and principal (or nodal) points. And the hyperfocal distance is measured in object space with respect either to the front focal or the front nodal point of the lens (see below).
Moreover, formulae for asymmetric lenses differ from the classical model, valid only for symmetrical lenses. So, not only for a general coupound asymmetrical lens you should know the inter-nodal distance, but even more, you should also know the pupillar magnification ratio and you should apply the most general formulae.

There are different issues there
1/ the engraved distances on helical focusing rings for small and medium format cameras, by tradition, refer to the object-to-film-plane distance, hence we can trust the engineers to have taken into account the inter-nodal distance for a specific lens. Helicals designed for view camera lenses do not care for the inter-nodal distance which is usually small for wide-angle and standard view camera lenses. And of course regarding DOF engravings, do not care for the pupillar magnification, view camera lenses that you mount on an helical are most often wide angle or standard lenses of quasi-symmetrical design.

2/ the hyperfocal distance "H" for symmetrical lenses (let's forget about telephotos and retrofocus lenses) can be defined with respect to the front focal point F or the front nodal point.N. The difference between the two lengths is simply equal to one focal length "f"
In classical models, the "true" hyperfocal distance between the object and the font nodal point N is equal to (f^2/(N.c)) + f where "f" is the focal length, "N" the f-number, and "c" the chosen diameter for the circle of confusion.
Usually, the term f^2/(N.c) is referred-to as the hyperfocal distance.
Stricly speaking, we should say : f^2/(N.c) is the "focal hyperfocal distance" and f + f^2/(N.c) is the "nodal hyperfocal distance".
Of course nobody cares for that. Let's simply say : H = hyperfocal distance = f^2/(N.c)

3/ Now if you insist on having the true value for the distance between the object and the film plane at the hyperfocal setup, then the total distance is
H + f + NN' + f + 1/H = (H + 1/H) + 2f + NN'.
Fortunately enough, "NN'" is most often small with respect to "f", so we can neglect "NN'", and we get the simplified formula actually valid for a single thin lens element : (H + 1/H) + 2f;

Again, 1/H is most often very small with respect to H, and eventually we get:
H = f^2/(N.c) is the hyperfocal distance measured to the front focal point F
H + f is the hyperfocal distance measured to the front nodal point N
H + 2f is the hyperfocal distance measured to the image plane, when H is much bigger than 1/H i.e. "all the time" for landscape shots.

As a conclusion, there is a very high probability that those simple formulae are actually in use in most DOF calculators, defining the hyperfocal distance as usual. The software designer should simply tell the user which reference point is:considered : either the front focal point (probably not), the front nodal point (high probability) or the film plane (like for engravings on helicals according to a secular tradition), and which approximations are in use.

[Helen Bach]

Sorry... wrong.

The DoF calculator gives the distance from the FILM PLANE to the subject.

It has no knowledge of what lens you're using, nor where the front nodal point (first principle plane) might be.
In many cases the front nodal point is not even inside the physical lens. It cam be moved by hundreds of mm.

- Leigh

Leigh, I'm fairly familiar with both the rigorous and the simplified versions of lens formulae, and with a number of analogue and digital DoF calculators, and I believe that you are wrong about this. Emmanuel has already given an excellent answer. I'll give a shorter one. The person making/writing the calculator does not know the internodal space, apart from anything else. If you look at the popular online calculators (at least the ones whose authors appear to understand the subject), they state that they are giving first nodal point - subject distances. You need fewer assumptions.

If the "front nodal point is not even inside the physical lens. It cam be moved by hundreds of mm" then whoever is designing or writing the DoF calculator doesn't stand a chance of getting the subject-film plane distance accurate: the internodal space will be quite large.

Here is what the popular dofmaster website (http://www.dofmaster.com/faq.html) has to say:

"Is the 'subject distance' measured from the front of the lens or from the film plane?

This question is relevant only for close-up and macro photography. For other photography, any error caused by measuring subject distance from the wrong location is neglible.

The depth of field equations are derived from the "thin lens" equation, which assumes a single lens element with no thickness. And, subject distance is measured from the thin lens. So, technically, subject distance is measured from the front of a lens.

However, that doesn't apply directly to a photographic lens. These lenses have many elements, and the front of a lens isn't necessarily the location that subject distance is measured from. The actual location is something called the "front nodal point" of the lens. The location of the front nodal point isn't documented by lens manufacturers, nor is it easy to find experimentally.

I measure subject distance from the front of the lens. I believe that the nodal point would actually be somewhere between the front and rear elements. But, by assuming it is at the front of the lens, I get a conservative estimate of the depth of field from the calculator. In other words, the depth of field calculation shows less depth of field than will actually be seen in the photo."

Of course we are lucky: many of the modern lenses we use do have the nodal points given by the manufacturers, not that it matters a whole lot most of the time. It's also quite easy to find if high accuracy is not required.

I only mentioned it as a possible explanation of the conflicting advice that Alex had heard. There are many reasons why it is an approximate science in addition to those given by ic-racer. (basis for max. acceptable CoC; simplifications in the formulae etc.)

Best,
Helen

PS It's a principal plane, not a principle plane, but I'm sure that you know that.

[genotypewriter]

Why hasn't anyone mentioned that HFD is a highly variable value and using it to set the focus will only give accurate results only for only one or maybe two reproduction instances?

[alexn]

so basically, focus on your horizon, then on the foreground, wind it 1/3 towards the horizon focus point again and stop down to F/32 :) Gotcha

[Dan Fromm]

Why hasn't anyone mentioned that HFD is a highly variable value and using it to set the focus will only give accurate results only for only one or maybe two reproduction instances?

Eh? Wot? Please explain further, and while you're at it show us where Prof. Dr. Bigler and Ms. Bach went wrong.

[Oren Grad]

Eh? Wot? Please explain further, and while you're at it show us where Prof. Dr. Bigler and Ms. Bach went wrong.

Prof Dr Bigler and Ms Bach aren't wrong, as far as they go.

I'll guess that genotypewriter was hinting at the problem that DoF and hyperfocal distance calculations necessarily make assumptions about CoC. Those assumptions may not conform to your picture-viewing requirements. I always appreciate it when the author of a DoF calculator allows for the CoC criterion to be set freely.

There's also a subtler point, something Harold Merklinger explained nicely - placing the focus point so that the calculated DoF reflects your CoC criterion doesn't necessarily result in a picture that looks "sharp".

Since we're talking about it, there's another point that's subtler still: the optical formulas on which DoF calculations are based assume idealized behavior in the defocus transition. In fact, real lenses vary in their bokeh. Perceived DoF varies as a result.

The upshot is that it's not possible to calculate your way to perfect focus.

[Oren Grad]

To be clear: used sensibly, together with other information, DoF calculations can help you make pictures that look more or less the way you want them to look. Just don't kid yourself that you can turn the crank on one of these calculators and use the result to make a picture that's sharp from 15.2 to 47.3 feet.

[Dan Fromm]

Oren, thanks for the reply. I'd still like to hear from genotyprewriter. To my literally little mind, "highly variable" and "subject to assumptions" aren't quite the same.


I'll guess that genotypewriter was hinting at the problem that DoF and hyperfocal distance calculations necessarily make assumptions about CoC. Those assumptions may not conform to your picture-viewing requirements. I always appreciate it when the author of a DoF calculator allows for the CoC criterion to be set freely.

I thought everyone knew that. My macro DoF calculator insists on being given a CoC, also points out the diffraction limited resolution ...


The upshot is that it's not possible to calculate your way to perfect focus.

Agreed. But you can find out what probably isn't possible.

Cheers,

Dan "limits are often worth the trouble of learning" Fromm

[Oren Grad]

...using it to set the focus will only give accurate results only for only one or maybe two reproduction instances?

...seemed to me the tip-off. But we'll see - it would hardly be the first time I've read something wrong.


...also points out the diffraction limited resolution...

Aye, subtlety #4. Thanks for adding.

[Emmanuel BIGLER]

The issue of diffraction effects is a subtle point.

Since we are often dealing here with the best lenses ever designed for LF photography, used at their best f-stop, we can consider that residual aberrations and diffraction roughly contribute to the same extent and actually limit the ultimate sharpness, at least for objects that we focus on at best.
For far out of focus objects, we can simplify the model by taking only into account geometrical optics en neglecting diffraction and aberrations. Hence from a very crude, oversized, value of the circle of least confusion, and for objects far out of focus, geometrical optics and classical Dof equations do apply. However if you insert in your DoF calculator very small values for the circle of least confusion, the model fails because small "c" values can be smaller than the minimum diffraction spot size, roughky equal to "N" microns where "N" is the f-number. For example, following the Holy Example of Saint Ansel, with your left hand, you stop down your lens to f/64 ; and at the same time, with your right hand on your palmtop calculator, you enter "32 microns" for the CoC. Doing so your are plain wrong. And a good calculator should warn you against this dangerous behaviour :o (using a palmtop with your right-hand only, while your left hand manipulates the controls of your lens at the same time is the ultimate challenge for any LF photographer)

However (and not kidding) we all know that the transition between sharp and unsharp play a very important aesthetic role. Hence let's experiment and see what happens ;-) I want my swirling bokeh ! Like a natural product, I do not want it to be modelled by a palmtop.

An additional (marginal) remark for those you love precision in equations (they'd better go out and take pictures, but this is another story ;) )
If you use the general DoF formulae valid in general for asymmetric lenses, you'll find that the hyperfocal distance, namely the distance where you expect DoF to extend to infinity at the back of the object on which you focus "spot-on", is the same as for a plain, simple, symmetrical lens (or a single thin lens element).
Hence people should not care for general DoF equations with a telephoto, mosty used for landscape at long distances : for all lens designs used for far-distant objects, the hyperfocal distance f^2/(Nc) (of f + f^2/(Nc) if you prefer) is the same regardless of the pupillar magnification ratio.

-------------------------------

Reminder : Jeff Conrad's DoF articles explain everything (for those who prefer to read maths prior to take pictures ;-)

part I, Introduction to Depth of Field
http://www.largeformatphotography.info/articles/IntroToDoF.pdf
Part II, Depth of Field in Depth
http://www.largeformatphotography.info/articles/DoFinDepth.pdf

[Leonard Evens]

so basically, focus on your horizon, then on the foreground, wind it 1/3 towards the horizon focus point again and stop down to F/32 :) Gotcha

When you are focused on the hyperfocal distance in the subject space, the distances from the focus point to the near point and far point (infinity) on the rail will be just about equal. Whenever you use the near and far points on the rail to focus, theoretically, you should focus so the distance from the rear standard to the rear principal point of the lens is the harmonic mean of the corresponding distances for the near and far point. But it is almost always true that this point is so close to the midpoint between the near and far points that you might as well use the midpoint.

If you favor the near point, as suggested above, you are biasing the focus towards the foreground, which may be a good idea in some circumstances, but it is not generally so.

In any case, the rule will not set the focus on the hyperfocal distance.

It is possible that the suggestion is referring not to distances along the rail but to subject distances. But in that case, it makes even less sense. The distance in subject space from the hyperfocal point to infinity is infinite.

There is another rule of thumb which suggests that one should focus so that the near depth of field is one half the far depth of field: i.e., focus one third of the way into the scene. This rule works when you focus at about one third the hyperfocal distance and for no other distance.

[genotypewriter]

Eh? Wot? Please explain further, and while you're at it show us where Prof. Dr. Bigler and Ms. Bach went wrong.

No it's not "wrong". In scientific writing, "variable" doesn't imply that something is incorrect or it is bad.

I said the values are "highly variable", as in reliant on other factors. And making assumptions on those other factors mean there will be limitations to the meaningfulness of of the HFD values. For example, you might assume a CoC for a 150dpi 8x10" print which will give a HFD of say 10ft. That HFD value is not valid anymore if you were to print the same image at 300dpi.



Oren, thanks for the reply. I'd still like to hear from genotyprewriter. To my literally little mind, "highly variable" and "subject to assumptions" aren't quite the same.

Oren read it right. And "variable" means it is "subject to assumptions", although they're not the same... just like not everything that's "edible when cooked" is "raw chicken".

[Helen Bach]

Thanks for that. I also was wondering what you meant. You had written "Why hasn't anyone mentioned that HFD is a highly variable value..." and yet I had mentioned that it was an inexact calculation in the previous post to yours, including a mention of the variation in the value of the maximum acceptable circle of confusion.

[Dan Fromm]

G, vagueness is a sin.

[genotypewriter]

Thanks for that. I also was wondering what you meant. You had written "Why hasn't anyone mentioned that HFD is a highly variable value..." and yet I had mentioned that it was an inexact calculation in the previous post to yours, including a mention of the variation in the value of the maximum acceptable circle of confusion.

Yes, "CoC" gets mentioned almost every time people talk about HFD or DOF but that wasn't my point. The post you mentioned as well as the other posts at the time were trying to go in to great depth about HFD calculations. My post was merely trying to pointing out the fact that HFD values are so variable and therefore why anyone should nit pick that much to find out whether it's given from the film plane or the lens.



G, vagueness is a sin.

And the irony is funny :) Actually the term "variable" is quite well understood in scientific circles... of no confusion :cool:

[Dan Fromm]

Y'know, in the labs I've worked in we always spelled things out. My collaborators and I have always been as clear as possible. "x is a highly variable value" isn't very informative. "x is determined by y, z, ... " is.

I've looked at your profile, see that you say you're "a scientist." That covers many fields with different standards.

As I said, vagueness is a sin. It sometimes, but not always, reveals ignorance (see your question on extension at 1:1, posted earlier today, for an example of a question that reveals lack of knowledge) and attempts to mislead. Telling strangers about whom you know little "I'm a scientist. Shut up." is insulting. Unwise, too.

No one knows everything. Trying to fill in gaps in knowledge and understanding is good. Trying to use credentials earned in one field to gain credibility in another isn't.

[Oren Grad]

However (and not kidding) we all know that the transition between sharp and unsharp play a very important aesthetic role. Hence let's experiment and see what happens ;-) I want my swirling bokeh ! Like a natural product, I do not want it to be modelled by a palmtop.

Swirliness around the periphery because you want to exceed the covering power of your lens isn't the issue. Within the zone that is optically well-corrected, the difference between smooth defocus transitions that retain form and rough focus transitions where ni-sen turns everything to frizz throws all calculations to pot.

Regardless of whether one is in a New-Agey frame of mind. ;-)

[Helen Bach]

The post you mentioned as well as the other posts at the time were trying to go in to great depth about HFD calculations. My post was merely trying to pointing out the fact that HFD values are so variable and therefore why anyone should nit pick that much to find out whether it's given from the film plane or the lens.

I believe that point had already been made. The original question asked whether the distance given by a DoF calculator should be measured from the film plane or the lens. Simply saying 'you don't need to know because the calculation isn't that accurate' may not satisfy everyone's curiosity. Is it really 'nit picking' to explain that there are two different practices, and why they are different?

[ic-racer]

I think one does not need to know if the hyperfocal calculation if from the lens, or film because it does not make any difference. Say the hyperfocal distance is 65 feet. So what. No one is going to be measuring that distance. It is just an intellectual curiosity. If you were going to measure a focusing distance, one would just focus the lens.

[rdenney]

And the irony is funny :) Actually the term "variable" is quite well understood in scientific circles... of no confusion :cool:

In the scientific circles in which I circulate, "variable" nearly always means "described by a random variable"--stochastic rather than deterministic. I don't think that applies to hyperfocal calculations. There are lots of factors in play, and it's hard to consider them all, with the result that calculations are usually imprecise. And they are based on a standard of sharpness defined by the user for a particular application, which might range widely. For example, DOFMaster, near as I can tell, is based on an 8x10 print. I routinely cut the circle of confusion values in half before using it for 16x20 target prints. That leads to much less calculated depth of field an a more distant hyperfocal point.

What might be stochastic (because it is subject to a range of subjective factors) is what people interpret as appearing to be sharp. There is not some binary boundary between sharp and unsharp. A print made a bit too large might still look acceptable to most people, and some critical viewers might not accept any of the usual standards of sharpness (many of which originated in the Zeiss Formula).

And that leads to the practical truth that outside the macro range, it probably doesn't matter whether the measurement is made to the film, lens, or tripod center column for that matter. And focusing at the hyperfocal distance is always outside the macro range.

DOFMaster (with their 8x10 print assumption) claims that the hyperfocal distance for a 47mm Super Angulon on 4x5, at f/16, is 4.7 feet. A couple of inches (i.e., the distance from the film to the lens) one way or the other makes not much difference. At f/32, it's 2.4 feet--still more than an order of magnitude greater than the focal length. The hyperfocal distance they report for a 300mm lens on 8x10, at f/32, is 47 feet. The difference of a foot between the lens and the film is not significant. Stricter standards of sharpness as indicated by selecting smaller circles of confusion just make the hyperfocal distance that much greater with respect to the focal length.

Remember, the thread was about hyperfocal distance, not more generally about depth of field.

Rick "who has spent a big chunk of his career defining variability" Denney

[Helen Bach]

So you don't think that it was worth mentioning the two different practices at all? I often ask myself if there is any real point in answering questions on the internet.

Best,
Helen

PS The online version of DoFmaster allows you to choose your own max. acceptable CoC if you wish.

[rdenney]

So you don't think that it was worth mentioning the two different practices at all? I often ask myself if there is any real point in answering questions on the internet.

I did not say that. I always enjoy detailed answers, and I've never believed a thread had to be limited to the conception of the OP. Your quote of the authors of DOFMaster was spot-on as far as I'm concerned.

But those who said it didn't matter were not without justification for their point of view. Saying it doesn't matter in practical application, however, isn't the same thing as saying nobody is interested in the different approaches.

Yes, as I said, I always cut the circle of confusion value used by DOFMaster in half. In the iPhone version, which I use, I select my desired C of C from the table, rather than choosing a format.

Rick "just trying to keep things in perspective" Denney

[genotypewriter]

Y'know, in the labs I've worked in we always spelled things out. My collaborators and I have always been as clear as possible. "x is a highly variable value" isn't very informative. "x is determined by y, z, ... " is.

This is what's sad about online forums... once you point something out for the sake of others, someone comes and demands you to explain it like it's your duty. Good students go do their own homework.


I've looked at your profile, see that you say you're "a scientist." That covers many fields with different standards.

As I said, vagueness is a sin. It sometimes, but not always, reveals ignorance (see your question on extension at 1:1, posted earlier today, for an example of a question that reveals lack of knowledge) and attempts to mislead. Telling strangers about whom you know little "I'm a scientist. Shut up." is insulting. Unwise, too.

No one knows everything. Trying to fill in gaps in knowledge and understanding is good. Trying to use credentials earned in one field to gain credibility in another isn't.

lol... look up the term "false dilemma", Dan. Remember the time you were pounding your chest when I claimed that a 65mm lens gives half the field of view of 150mm lens and you were insisting that it's should be a 75mm?

http://www.largeformatphotography.info/forum/showthread.php?79297-Vignetting-Analysis-47mm-to-1200mm-Lenses&p=762329&viewfull=1#post762329

;) But when I make a slip at 1.45am or something you go to say it "reveals ignorance".



In the scientific circles in which I circulate, "variable" nearly always means "described by a random variable"--stochastic rather than deterministic.

I don't know which circles those are but variable != random. Random also doesn't always mean variable... in several senses.

[Helen Bach]

I did not say that. I always enjoy detailed answers, and I've never believed a thread had to be limited to the conception of the OP. Your quote of the authors of DOFMaster was spot-on as far as I'm concerned.

But those who said it didn't matter were not without justification for their point of view. Saying it doesn't matter in practical application, however, isn't the same thing as saying nobody is interested in the different approaches.

Yes, as I said, I always cut the circle of confusion value used by DOFMaster in half. In the iPhone version, which I use, I select my desired C of C from the table, rather than choosing a format.

Rick "just trying to keep things in perspective" Denney

Thanks for the reply. I often find it difficult to judge the appropriate way to reply to questions - is it too much, is it too little, is it too vague, does it require too much self-assembly...

Best,
Helen

[Emmanuel BIGLER]

The distance in subject space from the hyperfocal point to infinity is infinite.

Well, now I understand why It always takes me a long time to properly focus; converting distances into time, even at the speed of light, can't we explain the phznomenon by the famous Law of Allenian Physics "The trouble with eternity is it drags towards the end".

[Oren Grad]

The distance in subject space from the hyperfocal point to infinity is infinite.

Well, now I understand why It always takes me a long time to properly focus; converting distances into time, even at the speed of light, can't we explain the phznomenon by the famous Law of Allenian Physics "The trouble with eternity is it drags towards the end".

To see a world in a grain of sand,
And a heaven in a wild flower,
Hold infinity in the palm of your hand,
And eternity in an hour.