DOF and hyperfocal distance considerations (Fuji GW690iii, Mamiya 7 etc)

[david b]

Following on from this theme, especially due to recent threads from

Leica and Mamiya 7 users, I am posting updated experiences with my

Fuji GW690iii. Obviously all three systems have outstandingly good

lenses.

 

Yes, I now have a thorough understanding of the theory. Yes, I

realise lens calibrations are an approximation based on an assumption

with certain criteria which are not always valid in every set of

shooting/display conditions.

 

When I started using the Fuji, I sometimes used hyperfocal-focussing

technique, assuming it would give the max DOF in landscapes from near

foreground to infinity.

 

Doing this, the infinity was always slightly but obviously soft

(heavy tripod used, Fuji Velvia film). Then I tried setting a

hyperfocal focussing point some two stops lower (i.e. focussing for

f/11 if using f/22 set aperture), still soft, albeit less so. Still

much worse to my eyes than focussing on infinity.

 

Two points come out of this:

 

Only actually focussing at infinity gave truly sharp transparencies,

and any extent of hyperfocal focussing gave less than optimum

sharpness. How many people realise this, I wonder?

 

The most interesting point, though, is that subjectively - to me and

a few others at least � is that if the infinity is totally sharp,

then even if the nearest point (which obviously is not quite as sharp

as if the hyperfocal distance had been set) appears to be sharper

overall. Perhaps the brain compensates for the near foreground if the

expected infinity sharpness is present?

 

Any other thoughts?

[Spearhead]

Can you post some scans that show this, this should be visible even at screen resolution.

[tim_klein1]

This point has been discussed quite a bit in various forums. Joe Englander discusses it briefly in <a href="http://www.englander-workshops.com/documents/depth.pdf">this pdf document</a>.

 

There's also a website that goes into this phenomenon quite a bit, but I can't seem to locate it right now. It's quite interesting, so I'll post it if I can track it down.

[doug_brightwell1]

I�m really interested in this issue of focus and depth of field. I

shoot landscapes with a Rollei with a 40mm Super Angulon,

among other lenses. My subject matter tends to have a great

amount of depth between foreground and infinity. In an ideal

world, I would use a camera with tilt movements, but for various

reasons a view camera (which I own) is not usable in most of my

shooting situations, and I just didn�t want to go the Fuji 680

route.

 

I recall from photo classes in my college days that one was

supposed to focus about 1/3 the way into a scene... that depth of

field was greater behind the plane of focus than in front of it. Boy,

was I in for a shock when I shot that way with the 40mm! The

foreground was very sharp, but the background on towards

infinity was soft. Occasionally I see comments on the web that

with a 40mm lens depth of field is so great that focus isn�t

critical. In my experience, it just ain�t so. In fact, the depth of field

at around f8 is minimal.

 

I read the acrobat file referenced in a previous post, and have a

question. The file is actually a bit mapped scan of the article and

is a little hard for me to read. I believe that Englander is

suggesting that if infinity is involved in the shot, the optimal point

of focus is �1/3 the distance _beyond_ the infinity focus point.�

I�m guessing that his choice of terms is based on bellows

extension, since 1/3 beyond infinity doesn�t otherwise make any

sense. So, the question is, is he essentially saying to focus 2/3

of the way into the scene for optimal depth of field?

 

Initially my goal was to be able to make a 20x20 print from a full

frame 6x6 color negative. If I�m reading Englander right, I�m

totally hosed since the Super Angulon simply doesn�t stop down

far enough. I typically shoot at f16, thinking that I should avoid the

smallest f22 aperture to avoid the diffraction effect.

 

The Zeiss Distagon verson of the 40mm stops down to f32. I

wonder if that extra f-stop would make a significant difference in

extra depth of field before encountering defraction problems. Or

would the degradation due to diffraction be the same on both

lenses at any given f-stop?

[val_a]

DOF is assumption and you know that it is based on the circle of confusion. You are correct about "two points come out of this". Discussion about infinity focus and DOF is on the following page (Tim Klein is probably referring to this one):

 

http://Fox.nstn.ca:80/~hmmerk/DOFR.html

[david b]

Thanks for the responses! I'm sorry but have no means of scanning the 6x9 transparency (or any other transparency for that matter), or I'd have uploaded a sample to illustrate.

 

I found that focussing 1/3 or 2/3 into the scene gave soft infinity, in fact anything other than focus at infinity gave some apparent fractional degradation at the far point (i.e. "infinity").

 

The DOF with a 90mm lens will be a fair bit less than the 40mm of course...

 

I have also previously seem the site refered to above, and I guess my findings bear this out. I also avoid smaller apertures than f/22 (f/16 ideally) due to diffraction issues.

 

Get it all right and the Fuji 90mm is so sharp you could shave with it.

[thorsten_merz]

David -

 

When I first moved into medium format I applied the same hyperfocal distance "rules" that I had been using quite succesfully with 35mm landscapes to my medium format landscapes. And boy was I in for a surprise - every image that I had applied this technique to had soft backgrounds. And this despite the fact that I aplied the same size for the circle of confusion as I had been using in 35mm (the conventional wisdom being that it can be bigger for MF).

 

For some time I struggled to find a cause for this - was it my lenses, was there something wrong with the film back or had I made a mistake in calculating the hyperfocal distance (I made little credit card size charts using MS Excel for each of my lenses).

 

Then I came across the "Depth of Field Revisited" article by Harold M. Merklinger (referenced above at http://Fox.nstn.ca:80/~hmmerk/DOFR.html). Since reading this article and applying the principle discussed therein, I no longer have soft backgrounds and am now quite happy with the results I am achieving.

 

I have carried out some focus-bracketing (where one shot is taken applying the conventional hyperfocal-distance techniques and a second applying the principles discussed in "Depth of Field Revisited") and have noticed a remarkable difference between both images.

 

Of course, like much in photography, experimentation is key to success and the above should'nt be taken as gospel. It works for me, but the same may not be true for you!

 

Cheers,

Thorsten.

[micah_marty1]

David Bickerdike wrote: "The most interesting point, though, is that

if the infinity is totally sharp, then even if the nearest point

(which obviously is not quite as sharp as if the hyperfocal distance

had been set) appears to be sharper overall. Perhaps the brain

compensates for the near foreground if the expected infinity sharpness

is present?"

 

Excellent observation! You raise some great points and clearly your

post has struck a chord with many of us who have struggled with the

same issues.

 

I've fretted at length over your exact quandary because I have two El

Cheapo large-format handheld cameras, each with fixed focus (i.e.,

they are preset to hyperfocal distance), an 8x10 camera that uses a

120mm lens and an 11x14 camera that uses a 150mm lens. Finding an

acceptably sharp (yet realistically handholdable) hyperfocal setting

for those focal lengths has presented a real challenge.

 

I've found that the larger the format, the sharper the background

("the infinity") needs to be (unless of course you choose to throw it

totally out of focus, an entirely different matter). I suspect this is

because with larger formats (maybe 6x6 on up?) the viewer is much more

likely to "zero in" close on the picture simply "to see how much I can

see" ("That license plate is readable a block away!"). With 35mm, I

think a general overall blurring of the distinction between grain,

resolution, and focus makes the viewer more forgiving (many of

Cartier-Bresson's pictures, for example, definitely aren't "sharp" at

infinity but then you wouldn't call his photographs "soft"

either--perhaps because you don't zero in as closely with 35mm to

inspect for such things?). (And yes, I know of HC's contempt for some

photographers' "insatiable craving for sharpness of images"; for this

thread's purposes let's just say his photographic goals were different

from the goals of many who use medium- and large-format. But his

example does illustrate why street-shooting is harder in MF and

LF--it's more than film handling and film costs that cause problems.)

 

When I bought the 8x10 cigar-box camera (a "Hobo" from

www.bostick-sullivan.com)--I modified it after seeing my first batch

of pictures precisely because of the focus issues you talk about. I

had been told that if the camera was focused at 12-18 feet, everything

from maybe 4 feet to infinity would be in focus at some ridiculous

aperture (by LF standards) like f/8 or f/11. Nonsense. The first thing

a viewer does when encountering an 8x10 transparency is to put his or

her nose as close as possible to it--with a loupe, if available!--to

see how sharp all of those buildings or trees are a quarter-mile away.

At this test my first pictures failed miserably.

 

I trundled over to Schneider's depth-of-field charts

(http://www.schneideroptics.com/large/depth/depthof.htm) to ponder

this. (These charts are made for LF lenses, but esp. the 6x9 settings

listed will at least show MF photographers what I'm talking about;

similar charts for non-LF formats are available elsewhere on the www.)

I found that Schneider's d.o.f. settings were too optimistic, perhaps

because they were based on circle-of-confusion measurements which I

found too generous but also perhaps because they didn't take into

account the increased expectations for sharper infinities of larger

film sizes (see beginning of this post, above).

 

I modified the preset focus on the 8x10 Hobo to much deeper than 12-18

feet and compensated accordingly when designing my homemade 11x14 box

camera. Two things I learned from Schneider's d.o.f. charts,

applicable I presume to other charts for other formats, inc.

medium-format (I'm no optics expert, just a hack photographer, so

correct me if I'm wrong!):

 

1. When possible, I use the charts for a smaller format than the

camera in question (e.g., for 8x10 I use Schneider's 4x5 tables, and

for 11x14 I use their 5x7 tables). This is, of course, the same thing

as stopping down one or two additional stops from what your lens

markings (or d.o.f. tables) tell you; the point either way is to not

regard lensmakers' d.o.f. calculations as gospel.

 

2. I noticed that, when looking at Schneider's tables, at least with

the apertures I was talking about (f/22, f32, etc., routine for LF)

the receding distance of foreground focus wasn't so bad as the

infinity focus increased. In other words, on the 5x7 chart at a given

aperture (f32) with a given lens (150mm), the chart told me that if

the lens was focused at 20 feet everything from 9 feet to infinity

would be in focus, while if the lens was focused at 100 feet the near

focus increased to 14 feet. For the kind of scenes I had in mind and

the formats I was dealing with, the latter was an acceptable tradeoff;

I'd settle for a noticeably sharper infinity even if the only

"penalty" was a loss of 5 feet at the near distance. Obviously for an

MF photographer struggling with focusing on both the tree within

touching distance and the distant hills, that solution won't be fully

workable, but my point is that the charts should be scrutinized for

such "curves" rather than just glanced at. (Boy, there's a fun way to

spend an afternoon!)

 

One final observation that might be useful to those who have bothered

to read this far, something I learned when rapidly shooting city

scenes with a LF camera that offered only focus and rise--no other

movements. Make sure that something in the scene that will logically

attract the viewer's attention is in perfect focus even if it's not

the "subject" of the picture and even if it means that some less

prominent parts of the scene then won't be in perfect focus. That

colorful sign across the street, the one with the small lettering and

pictures on it? Don't just set your hyperfocal for the overall scene

and be done with it; make sure that sign is in sharp focus, because

you can bet viewers are going to try to read that tiny writing (and

find the entire picture "soft" if they can't). A similar example

showing the importance of infinity focus that David alluded to: you've

got a beautifully curved bridge girder 15 feet away through which you

can frame the city skyline (an infinity away). Where do you focus? A

compromise so that your lens claims both 15' and infinity are in

focus? Sure, that's fine--but also try one frame at infinity, even if

it means the girder will be slightly softer. With an enlarged

photograph made with the larger formats, being able to see every

window of every building in the distant skyline just might be of

greater importance to viewers than the sharpness of the bridge girder.

Completely hypothetical example of course(!), but you get the idea.

 

Bottom line: with landscapes and cityscapes I'm a strong advocate of

having something in the scene in perfect focus rather than merely

going for a compromise between near and far. When in doubt I make at

least one exposure where that "something" is infinity, because--as

David suggested--in many cases (especially at bigger enlargements)

that will determine the apparent sharpness of the overall photograph.

[hans_bernhard_beykirch]

There a nice DOF calculator for PC and Palm Pilot on http://www.dl-c.com/dof.zip. After some tests I evaluated that my Pentax 35mm lenses are calculated on the base of 30 lines/mm and the Mamiya 6 G-lenses on the base of 20 lines/mm. To get 40 lines/mm, which is sufficient in my point of view, I correct the Pentax lenses by 1 f-stop and the Mamiya ones by 2.

 

What is your experience with x lines/mm?

[tim_brown5]

<a href="http://fox.nstn.ca/~hmmerk/HMbook14.html">Here</a> is the Merklinger article on DOF

[tim_brown5]

<a href="http://fox.nstn.ca/~hmmerk/DOFR.html">This</a> is directly to Merklinger's article on DOF

[garry edwards]

I have a very good DOF chart on excell, works with any format and any lens at any distance, will gladly email it to anyone who wants it - but please email me straight away as I am going on holiday Friday p.m.

 

The point that I really want to make is that DOF is no more than an optical illusion - an area of apparent sharpness. The laws of physics are immutable and do not allow any point except the point on which the lens is focussed to be absolutely sharp.

 

If a point on which the lens has not been focussed appears to be sharp then this is really just a subjective judgement, although of course it is affected by all the usual factors of effective aperture, circle of confusion and magnification. Witness the Russian lenses, that seem to have more DOF than others - they simply apply lower standards vis-a-vis circle of confusion, and so can claim more DOF.

[john_lawson]

I also find that the eye wants to see details at the far distance as well as close and to do so requires biasing the focus point much farther than the hyperfocal distance. There are a couple of tricks which I use. I tilt the lens on a folding camera about 1.5 deg and leave it there. This is about half the amount required for full Scheimpflug tilt for a 3 inch focal length lens at about 60 inches off the ground. For my Mamiya Press camera with tilting back I made metal shims which give me 1.5 deg tilt. This is much quicker than

tilting the back while looking at the image on the ground glass.

Another possibility is to take the shot in two parts with one

focussed at infinity and the other close up and stitch them together

with a program such as Panarama factory.

[micah_marty1]

An addendum: Please note that in late 2001 Schneider redesigned their web site and the depth-of-field URL given in my post above is no longer functional. They are now located at:

 

http://www.schneideroptics.com/info/depth_of_field_tables/