The dreaded DoF issue again

[howard smith]

"... and the value commonly accepted varies depending on the format."

If I first assume CoC (and therefore DoF) are format dependant, then it will be much easier to prove it true.

If I assume A=B, then apply a true statement that B=C, then logically A=C.  True, if and only if, the assumption that A=B is true.

[howard smith]

"... since larger format means less enlargement for a given print size."  An Assumption and not always true.

I could have taken the same image in my example with a 4x5 camera (LF) with a 150mm lens at f/11 and made the same 8x10" print.  The degree of enlargement would have been exactly the same.

BernardLanguillier, can you take the mathimatical equations for DoF and point to format?  No, unless you point to CoC and make some format dependant assumptions.  Those assumptions need not be made if you understand how DoF works.

[howard smith]

"... given the subject is at the same distance from the lens of course."

"given" is equal to "assuming."  What other assumptions are you making?

The near and far points are a function of the hyperfocal distance, focus distance, and focal length.  For focus and hyperfocal distances much much larger than the lens' focal length, you can drop (yes, assume it doesn't matter) the lens' focal length.

The hyperfoacl distance is a function of focal length, f/stop and CoC.

FWIW, I have read Michael's article several times.

[Jonathan Wienke]

While the DOF formula does not have a "format" parameter per se, choosing the correct CoC value is indeed format-dependent. If you look at most of the web-based DOF calculators on the web, you'll find a variety of recommended CoC values depending on the camera format you're using. The assumed enlargement factor is part of the reason for this, as well as the fact that lenses that are designed to produce a large image circle cannot focus down to as small of an area as lenses designed for a small image circle. Cheapo digicam lenses are capable of a >50% MTF at LP/mm specs that Leica rangefinder glass never dreamed of, because they only need to cover a 6mm image circle instead of the 40+mm required for 35mm format. And said Leica glass can perform at much higher lp/mm than 8x10 large-format lenses. Given that, the actual performance capabilities of the lens and recording medium are far more important when choosing a CoC value tha print size, especially considering that in many cases prints of several different sizes will be made from a single exposure.

So the "right" CoC value should be based on the actual capabilities of the lens and recording medium, and accurately reflect their capabilities. Print size, which is subject to random and capricious alteration after the exposure is made, is nothing more than a red herring distraction that only confuses the issue.

[Ray]

How do you weigh an elephant with a very small scale that can handle no more than a kilogram?

It's possible, but not particularly efficient.

You are welcome to make suggestions.

[Ray]

I've just had an interesting conversation with Howard through a series of private messages regarding this issue of whether or not knowledge of camera format is really essential for calculating DoF on a given size print.

Howard's view seems to be that camera format has no bearing on the matter. He supports this opinion with the fact that many DoF formulas in the text books do not have a 'format' input.

He also reinforces this view by quoting that many photographic experts, including Michael, are of the same opinion.

I have to state, at the risk of being expelled from the forum once again, that I find such a view total codswallop.

It's theoretically true in the most impractical sense possible for any photographer in the field. Basically, if you don't know the format of your camera, you are up sh*t creek without a paddle, as regards DoF calculations.

If anyone wants to dispute this, I'm all ears.

[Ray]

Quite true. Your statements are all correct.

[BernardLanguillier]

Hum... I'll let you and Ray discuss this one.

Just one detail though, a larger CoC will contribute to give the LF camera more DoF...

As a final comment, the usage of a larger CoC for LF cameras is actually not only the result of an arbitrary choice resulting from the fact that less enlargement is needed, I believe that it is also picked so as to stay coherent with the actual definition LF lenses are actually capable of.

Somebody could finally perhaps confirm whether the resolution of 4*5 sheets is as high as those of 35 mm roll film? If it were lower with the 4*5 sheet, this would be another reason for picking a larger Coc?

Anyway, I'll be watching your next exchanges with great theoretical interest.

Cheers,
Bernard

[BernardLanguillier]

Howard,

Sure, it does depend on enlargement and viewing distance.

The "enlargement" part is responsible for most of the feeling that format has to do with it since larger format means less enlargement for a given print size.

Discussing whether the enlargement is the actual driving force, or whether the format is the driving force (as it defines the required enlargement) is in the same league as discussing which of soccer and basketball is most exciting, isn't it?  

Regards,
Bernard

[howard smith]

If I have one camera (let'sassume it is a 35mm), one lens (50mm), I always shoot at f/11 (to get what I think is maximum sharpness), I always make 8x12" prints and I never crop, and I always veiw my print from arms length - a lot of ASSUMPTIONS - then isn't DoF merely a function of focus distance?

[sergio]

Howard's view seems to be that camera format has no bearing on the matter. He supports this opinion with the fact that many DoF formulas in the text books do not have a 'format' input.

He also reinforces this view by quoting that many photographic experts, including Michael, are of the same opinion.

Lets try and put this one to rest once and for all.
Depth of Field is a phenomenon exclusively dependant on the F stop used. Michael has a nice test that demonstrates this somewhere on this site.
HOWEVER, format, enlargement factor, and focal length do affect perception of focus, hence of DoF. Why? because though the focus relationship between a nearby object and a far one is exactly the same with a wide angle or a telephoto lens at the same given aperture, perceivable sizes in relation to the frame are not, thus making us believe the short focal length gives us more Dof and the telephoto less Dof. For practical and real world photography I count on this impression of extra Dof. But for test chart photographers it probably won't cut it.

Maybe in this difference between what is perceived and what is measured is where all this circle of confusion is originated. ::

Sergio

www.sergiobartelsman.com

[sergio]

DoF is a phenomenon exclusively dependant on F stop not on focal length or format or anything else given the subject is at the same distance from the lens of course.  
Check out Michael's test here.


Watching the test makes it easier to understand what I mean by the measured focus relation between two objects and the perceived relation between them.

[Tim Gray]

As much as I hate myself, I'm just a sucker for these DOF debates...

The coc is a measure of the size at which a "point" starts be be big enough to be perceived as a disk.  I think there's no controversy in that depending on print size and viewing distance and the magnification factor of the microscope used to view the ink dots.   So the controversy is around whether the format (size and resolution?) of the capture medium has an impact?

So here's a thought expirement:

Imagine, if you will, a sensor of a gazillion photosites (a technical term) that's 8x10 and I mean feet, not inches.  Take a shot.  Take the same shot, ie distance, f stop, fl, with a 20D.  Print both same size - say 8x10 (inches) and view from the same distance (I know the field of view won't be the same).  Now print both 8x10 (feet) and view from the same distance.  If you think that the size of the range of acceptable sharpeness is NOT the same in both of these large prints, then I think you would conclude that format has an impact?

[Ray]

Well, it all seems perfectly clear to me.

1. You have enlarged (or reduced) CoCs on the print, depending on whether the enlargement is bigger or smaller than the film/sensor.

2. You have on the film/sensor many CoCs, the size of which is determined directly by lens FL, f stop and distance to subject adjustments.

3. In order to produce a print of a given size it is necessary to know what it is that's being enlarged. To put it another way, a single CoC does not a picture make.

[Ray]

So the "right" CoC value should be based on the actual capabilities of the lens and recording medium, and accurately reflect their capabilities. Print size, which is subject to random and capricious alteration after the exposure is made, is nothing more than a red herring distraction that only confuses the issue.

Maybe so, but not only are there are no inputs to the standard DoF equations for fim/sensor and lens quality, most photographers probably don't even know the specifications of their lenses. Some films and lenses don't even have published MTF graphs of their resolving power. It's easy to get information on the pixel pitch of your camera's sensor, so making the CoC equal to the pixel pitch is a way out, but doesn't seem to me entirely satisfactory.

Dofmaster seems to have tied up everything nicely. If you consider that an acceptable CoC on the print, whatever the degree of enlargement, is 0.24mm, then Dofmaster's choice of CoC for the various formats all relate fairly accurately to a standard 8x12" print. They seem to have used the long dimension for the enlargement factor. For example, 0.03mm for 35mm but 0.2mm for 8x10 (instead of 0.24) and 0.1mm for 4x5.

Knowing that their quoted CoCs relate to an 8x12" print (and accepting that 0.24 on a print is adequate), it's easy to make the appropriate CoC adjustment depending on the intended enlargement. For a 6x4 print from 35mm, for example one could choose a CoC of 0.06mm (instead of 0.03), or for a 16x24" print, 0.015.

If you want to raise the standards, then consider Dofmaster's CoCs all relate to a 5x7" print.

[Gordon Buck]

Smiling while writing this because of the opportunity to espouse one of my favorite theorems....

Eventually all engineering equations involve a constant.  An engineering constant is generally defined as

      (actual result) divided by (my computed result)

so that in the future when "my computed result" is multiplied by the engineering constant, the final computed result is equal to the actual result.  All in all, a very useful and practical approach.

The problem, of course, is that the "constant" is not really constant.  In other words, many -- and different -- constants are needed to satisfy the variety of situations that occur in real life.

The "Circle of Confusion" is such a constant.

I'm often reminded of the joke about the pig judging contest at the county fair.  The pig would be placed on a balance beam and rocks stacked on the opposite end of the beam until balance was achieved.  The judges would then guess the weight of the rocks.  Someone should actually weigh the pig!

[BernardLanguillier]

- You use a lever?
- you chop up the elephant in small pieces?
- you assess visually the rate of distruction of the balance on which on put the elephant and use a constant to compute the weight?
- ...

cheers,
Bernard

[BernardLanguillier]

Circle of confusion is a central element in DoF formulas, and the value commonly accepted varies depending on the format.

This results from the fact that the larger the format, the less you will have to enlarge...

Other factors, like the focal lenght angular coverage also influence indirectly since longer focal lenghts will typically be used on larger formats.

Regards,
Bernard

[howard smith]

OK, let's assume DoF is format dependant.  I take my 35mm camera out and see a scene I want to photograph.  I look at the DoF calculator I found online at DoFMaster.  I plug in 35mm format, I plug in my 150mm lens, I plug in f/11 and hit the go button.  It comes up with a hyperfocal distance of 100', near point of 68.6' and far point of 184.3'.  Great, this is fine, so I focus the lens at 100', dial in f/11, let the autoexposure (aperture priority) pick a shutter speed.  Beep, image made.

I run home and make an 8x10 print.  I look at it from arms length (I'm holding it).  Perfect.  DoF is just right.  (I conclude Ray is right and DoF is format dependant.)

I look again at the print and see it is so beautiful, I will use it on the cover of my new coffee table book.  I e-mail the printer and say run with it as the cover.

A while later I see the cover of the 16x20" book.  The Dof stinks, way too shallow.  What happened?  After all, I only changed the enlargement of the image, certainly not the camera or film format.  MAYBE, just maybe, the ASSUMPTION that DoF is format dependant is wrong.

Disgusted, I toss the book on the coffee table and walk away.  I look back from the kitchen door, and the image, DoF and all, looks perfect.  I must have changed the camera or film format while going for that scotch/rocks?  Of course not.  I changed the viewing distance.  MAYBE, just maybe, the ASSUMPTION that DoF is format dependant is wrong.  Seems now it may depend on the enlargment AND viewing distance.

And the beat goes on and the evidence builds that DoF MAY NOT BE format dependant.

[howard smith]

"Depth of Field is a phenomenon exclusively dependant on the F stop used."

That certainly isn't true UNLESS I assume I have only 1 focal length lens, always use the same CoC, and always focus at the same distance, and always make the same degree of enlargement and view the print under the same conditions.  Because many readers here have zoom lenses, the assumptions are not likely true.  (In fact, Ray says he uses zoom lenses to make sure the view finder is always filled at the correct composition.)