DOF/Dr. T/Merklinger Book

[P3D Edward Gosfield III, MD <egosfield@xxxxxxx>]

Hi

Some reflections on George's adventures with DOF and ?Harold Merklinger's
book(s).

>------------------------------
> I now believe that the photographer should use
>judgment]...

Always--agreed

>The question is, which is the main
>object being photographed and what is the finest detail required to be
>captured on film?  This question should always be kept in mind when doing
>critical (in terms of sharpness/ resolution) work.

This is true on the face of it, but what really counts ( and what George
probably meant--as Merklinger  does) is the degree of detail (ie resolution)
required in the final image as seen by the viewer.  So 2x2 prints vs. Red
Dot viewer, vs. projection would indeed have different requirements.  On the
other hand, do you want to relax your standards and assume you won't need to
use the image in a more stringent context? Or if, like me, you never
project, are you willing to give up some ultimate resolution--ie use a
larger circle of confusion.  In flat photography resulting in a print, of
course, the choice is relatively easy.  Similarly for projection at a known
magnification which remains constant.

>[...] In this case, it might be better to even focus
>at infinity to get the most out of the city and be satisfied with the
>detail in the canon.
>
>One might object and say that by focusing at infinity I waste all the depth
>of field that is beyond infinity.  That's true but when talking about DOF
>we should not forget that the plane that is in exact focus will show more
>detail than the plane at the edge of the field.  Why place the city at the
>edge when it requires the most detail, just to maximize the DOF and put the
>most detail in a place where it does no good?  Again, what is the main
>subject and what is the finest detail required?

If you choose your circle of confusion correctly, (ie use judgement) then
your 'infinity detail' should be _just as good as you require_, when focused
at the distance whose DOF encompasses the desired range of subject
elements...Assuming the nearest and furthest details are technically doable.
I have a hard time finding any aperature which allows me to include, say, a
2 foot distant tree and a distant mountain, when using a MF camera, say 6x7
with a 90mm lens (=38mm in 35mm format). 

What Merklinger's approach does is allow you to consider the size of subject
detail whose image on the final product will be included within your
'appropriate' circle of confusion choice.  You can already do this with a
properly engraved DOF scale, since you can focus the lens to include the
'zone' of focus desired.  I do this all the time.  It just means you are
applying a predetermined circle of confusion and selecting the zone of
desired acceptable sharpness. On a Leica DR Summicron, i find that for my
satisfaction i have to use the DOF markings for one aperature larger than
the one set on the lens, while for the Kodak stereo i agree with those who
use 2 stops wider aperature (as Eric and others advised us long ago).  This
is based on my use of the Leica for prints of 5x7 to 8x10 viewed at a
maximum of arms length, while for the Kodak i use a Kodak viewer.  If you
use a  Reel-3-d 'cheapy' plastic viewer, of course, you don't need as much
DOF. Leitz probably used the standard Zeiss promulgated choices of 1/1000*FL
or the later 1/1500*FL to determine their DOF scale.  Kodak obviously did
not, nor, i think, does the Realist.


>If you are using f16 and you are focused at 10 feet while your closest
>object is just the ground at 7-10 feet, then you are short-changing your
>infinity detail (if that is important for you).

I agree--you should not blindly(!) focus at the hyperfocal distance.  You
should choose a focal range which includes the desired subject elements.

Merklinger's approach requires (in rigorous application) the photographer
literally to choose what size subject detail (1mm, 1cm, 1 foot, etc.) is the
smallest which need be resolved at the limits of DOF.  I think for tabletop,
studio, etc. this may work fine, but i think in the field it is pretty hard
to figure out how big the twigs are on the trees at the limits of the
subject field.  I suppose one could make some measurements and memorize them
for typical subjects.  Merklinger seems to do something like this in the
examples he posts by assuming some typical subject detail sizes.

Perhaps the books, as opposed to the papers posted on the Web, go into this
in better detail.


>
>>Use the camera's DOF scale, but 2 stops lower than indicated. In
>>other words, use the f/8 markings when shooting at f/16, the f/5.6 mark
>>when shooting at f/11, etc.
>
>The author had an interesting comment about this, which can also be applied
>to properly constructed DOF scales for more demanding applications.  He
>said that his first reaction, after being disappointed by the results of
>the hyperfocal setting, was to use the hyperfocal setting for the next
>wider aperture.  He was surprised to find out that he only got a small
>improvement (2x) even though he expected an order of magnitude (10x)
>improvement. Working the mathematics shows that this practice indeed
>doubles the resolution.  He still was not happy until he realized that,
>instead of focusing at 12 feet at f11, he needed to focus at infinity to
>get the most detail out of the smallest details at infinity.  That is an
>order of magnitude improvement, if not more.

Using the standard formula, a 33mm lens (like the Kodak?) at f/11 has a
hyperfocal distance of 16 feet for a circle of confusion of 1/1500FL.  This
coc is adequate for standard 35mm photography for prints, but possibly
excessively stringent for stereo transparencies and viewers.  I can't
comment on projection.

Using these same standards and formula, BTW, one comes up with the rather
easy to remember sequence of hyperfocal distances(feet): 7.8 (let's say 8),
11,16,22,31(let's say 32),43(let's say 45) for aperatures of, respectively,
f/22, f/16, f/11, f/8, f/5.6, f/4.  How familiar....?magic. This sequence is
the mantra I mutter to myself when wandering about with the Kodak or Realist.

>
>Regarding diffraction effects, the author states that his experience with
>35 mm lenses has convinced him that he gets the maximum sharpness at f8 or
>f11.  Closing the aperture to f16 will resolve less detail, not more.  He
>says that he does all his critical landscape work at f8 or f11 and usually
>focused at infinity.  Many beginners make the mistake to favor very small
>apertures (f16, f22) at ordinary (including infinity) scenes without
>realizing that the excess DOF is not needed, the pictures are not going to
>be any sharper...]

There is good discussion of how to balance DOF, resolution, and diffraction
effects, in Phototechniques a year or so ago. There is, by the way, an ad
for Merklinger's books in the back of each issue, and a reference to his web
site

http://fox.nstn,ca/~hmmerk/

i haven't checked the web page content to see if it still has the stuff that
used to be posted a year ago.. He had some very interesting interactive
diagrams using Java.


>
>Again, the nature of the subject is of primary importance.  For most of my
>table-tops or macro close-ups, depth of field is more important than fine
>detail and I stop my macro lens to the smallest aperture (f22). 

seems reasonable unless you are measuring details for some scientific purpose.
 
>
>The advice of focusing at infinity comes from a landscape photographer who
>works with large prints where fine detail is important.  This is not my
>typical photography but I learned a good lesson by reading his book.

The infinity advice doesn't match my LF experience, but i have never gone
beyond 16x20 from 4x5. Remember--"large prints" in any format require a
different choice of coc, just like shoving your nose in an 8x10 or 11x14 (or
sitting real close to the projection screen). In such cases the DOF is
effectively reduced, and you may end up having to focus nearer to infinity
to encompass it, especially with the longer FL lenses required in LF
photogrpahy.

Would John B. or another DOF maven/wallah like to comment on the appropriate
coc for stereo transparencies viewed in Red Dot or Kodaslide viewers?

(surely we need not take this subject to tech-3d)

best wishes

ted gosfield
-----------------------------------
mailto:egosfield@xxxxxxx


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