DOF resolution and perspective

[lct]

Advertisement (gone after registration)

..If a sensor has a crop factor compared to FF then to get the same perspective you need to compensate the focal length, so a 28 mm lens on the M8 'looks like' a 35 mm lens on film...

But why do you need to get the same DoF as that of another lenses (FF or whatever) when you photograph? You just need to know the DoF you'll get with your actual lens at such or such aperture don't you?

[mjh]

But why do you need to get the same DoF as that of another lenses (FF or whatever) when you photograph? You just need to know the DoF you'll get with your actual lens at such or such aperture don't you?

If the M8 happens to be your first and only camera, then indeed you wouldn’t care. But then there are those of us who started with a 35 mm camera. Their experience is tied to the world of 35 mm photography and cannot be transferred unchanged into the digital world. When you are accustomed to getting a certain angle of view and a certain depth of field from a 50 mm lens at f2.0 on, say, an M6, you have to learn that a 38 mm lens at f1.5 (or, as such a lens is not available, a 35 mm lens at f1.4) would give you the same angle of view and the same depth of field on an M8. The notion of an equivalent aperture (resulting in the same depth of field despite the smaller sensor and the correspondingly shorter focal length) is quite useful for this type of photographer.

 

If you are looking for an invariant across different sensor sizes, how about this one: For two combinations of sensor and lens, both capturing an identical angle of view, the depth of field is identical if and only if the entry pupil of the lenses is the same. This is true for any size of sensor and any focal length.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[lct]

Well i find this artificial and confusing personally but it's just me after all so if it helps you actually -- but does it really? -- it's OK by me folks.

[SJP]

For two combinations of sensor and lens, both capturing an identical angle of view, the depth of field is identical if and only if the entry pupil of the lenses is the same. This is true for any size of sensor and any focal length.

 

Michael, thanks for the comment! This boils down to the same thing as the form-follow-format expression and basically says

f-crop.A-crop = f-fullframe.A-fullframe. This is NOT true (see table in appendix).

 

For instance it is wrong if the circle of confusion is assumed to be governed by the sensor capabilities instead of using the 'print standard' (20 micron for M8 versus 23.5 micron based on the crop factor). It gets worse if you take diffraction into account. Finally it is NOT true for large apertures below f/2 or so, again see the table. This is basically caused by equating sin(x) to x which of course is OK for small apertures, and otherwise fails horribly.

 

Taking the same entrance pupil is fine as a first estimate.but that is all.

 

Of course actually the question is whether we think the difference between 1.2 and 1.45 is significant or not for reformulating the depth of field. Factor 1.45 is close to 1 stop, 1.2 is about ½ of that.

 

For myself I would not care as I am happy with the M8 as it is.

[mjh]

For instance it is wrong if the circle of confusion is assumed to be governed by the sensor capabilities instead of using the 'print standard' (20 micron for M8 versus 23.5 micron based on the crop factor).

Yes, it is based on the assumption that the DoF is proportional to the sensor diagonal. It is immaterial whether it’s assumed to be 1/1000 or 1/1500 of the sensor diagonal or whatever value one prefers, as long as it is some fixed percentage of the sensor diagonal. I happen to think this is a sensible assumption.

 

It gets worse if you take diffraction into account.

Yes, but then standard depth of field calculations never take diffraction into account. That’s not to say that it doesn’t matter, just that it is a separate issue. If we assumed the CoC to be limited by diffraction, more diffraction would imply a wider depth of field. Technically that is true and one could choose to see it that way, but I am doubtful many photographers would. This is similar to treating the CoC as dependent on the pixel pitch – lowering the resolution would then increase the depth of field, but I don’t believe this to be a particlarly practical perspective.

[SJP]

Yes, it is based on the assumption that the DoF is proportional to the sensor diagonal

 

This is fine as an assumption if that is what you think is appropriate but even then the rule does not hold below f/2. For f/1 the ratio is 1.49, f/1.4 ratio 1.41, f/2 ratio 1.37, further on it is close to 1.33 which is the ratio you are proposing to use (1.33 = 1/0.75 where 0.75 is the M8 crop factor)..

 

So you cannot just use the rule of thumb under all circumstances even under the assumption above..

 

Yes, but then standard depth of field calculations never take diffraction into account.

 

So who claims we should be standard? Maybe it is something worth doing better then.

[ho_co]

Advertisement (gone after registration)

Stephen--

I'm glad you're working this out. I hope my summary of Karbe's "Form Follows Format" didn't mislead you. I realized after sending it that your mathematical capability far exceeds mine.

 

 

Michael--

Since I don't follow mathematical arguments well, do you basically agree or disagree with Osterloh's explanation of Karbe's topic in the LFI article? Thanks!

[SJP]

Stephen--

I'm glad you're working this out. I hope my summary of Karbe's "Form Follows Format" didn't mislead you. I realized after sending it that your mathematical capability far exceeds mine.

 

Thanks Howard,

 

No worries I am enjoying myself (ok,ok I know that seems unlikely but I am weird I guess), plus it was a very useful article to focus on some stuff I didn't really understand myself.

 

Stephen

[ho_co]

Stephen--

You seem to be making headway with lct! Congratulations! He and I went through this matter before (though I stuck strictly with the Karbe concept from LFI and with a couple of the Web postings on LuLa), and I was unable to break through.

 

Your greater math competence is a great advantage, and you present the matter much better than I did.

 

Seems to me the concept of equivalent aperture is extremely important; SFAIK, Mike Johnston is the only major blogger who still hasn't signed on to the idea.

 

So since you're doing so well, let me throw some questions, arguments and questionable arguments onto the pile :

 

You raised the question above whether this is at all meaningful, and I would point again to the LuLa article Digital Focusing Part One et seq, where our normal expectations of depth of field are thrown for a loop.

 

Since DoF has no existence in the world, but is only a perception; and since the M8's incredible sharpness has brought us face to face with features of our lenses (like focus shift) we had never noticed before; therefore I wonder whether the whole concept doesn't need re-thinking from the ground up--something I'm glad to see you and Karbe and others doing. It is very easy to say that the old formulas hold, but we're finding a lot of breakdowns there (as pointed out at Equivalent Lenses).

 

In regard to IR--

Since it behaves differently from the frequencies for which most photographic lens design are calculated, and since it behaves differently with different glasses and with different formulas, my feeling is that a 'one size fits all' IR calculator may not fit many. (For examples see the references above to the 280 Apo and the 100 Apo.)

 

When LFI discussed shooting IR with the WATE, I was amazed by the amount they had to refocus to bring the images into focus. Seems to me their recommendation may be best: Shoot trial shots, comparing them on the camera screen; and then using that particular focus shift as standard with that particular lens.

 

As Plato said (Euthyphro, part 04), "And a thing is not seen because it is visible, but conversely, visible because it is seen."

 

 

This is incontrovertibly a superb thread! Thanks especially for the PDF and the spreadsheet. Major effort and as you said, quite enjoyable!

[lct]

...Mike Johnston is the only major blogger who still hasn't signed on to the idea....

Perhaps he does not like those why-doing-simple-when-you-can-do-complex theories that much.

[ho_co]

You're right, lct. He's no dummy. He did catch on to Erwin's "Photography Is Dead" idea quickly and even simplified it to make it more understandable for a lot of nay-sayers.

 

But since he has grasped the "equivalent focal-length" matter and typically remarks "and of course you get more depth of field with the smaller sensor," I have to wonder why he doesn't embrace the "equivalent aperture." The latter says the same thing but says it succinctly.

 

As the LFI article says:

The 6.3-25.2 mm, f/2.8-4.9 lens of the D-Lux 2 becomes the 35mm equivalent of a 35-135 mm, f/16-26 lens.

[lct]

...As the LFI article says: The 6.3-25.2 mm, f/2.8-4.9 lens of the D-Lux 2 becomes the 35mm equivalent of a 35-135 mm, f/16-26 lens.

What's this 'equivalence' for Howard?

I haven't used a D-Lux 2 so far but AFAIK there are no DoF markings on those digicams where DoF is so huge that anything's sharp if you don't use the macro setting.

BTW i did not do the math here but those LFI figures look ridiculously false at first glance.

According to the DofMaster calculator (pic below), for a 2 meters subject distance, the D-Lux 2 has a total DoF of 365 meters at f/2.8 and 6.3mm where that of the 35mm lens at f/16 is only 8 meters.

Again why doing simple...

 

[lars_bergquist]

Lct,

The thing that Merklinger brings forward, is that the basis of all this (35mm with a COF of 0,030 mm) is outdated and was already in 1991. This standard was set up somewhere beginning last century and lenses and output (be it print or in our age screen) have a much higher/better resolution. [/i]

Merklinger is wrong on one count. The sacrosanct CoC value of 1/30th of a millimeter (0.033 mm) rests on the assumption that the main competitor of the Leica would be the ubiquitous 6x9 cm rollfilm folder. Sunday photographers did usually get contact prints when they handed in their films for development and printing.

 

Now assumption #2 was that the largest CoC *in the print* that the viewer would experience as sharp, was 1/10th of a mm. The print CoC is of course the negative CoC times the enlargement. A neg CoC of 1/30th then would permit an enlargement ratio of 3x and a visually sharp print ... measuring 7.2 x 10.8 cm—hooray, bigger than a 6x9 cm contact print!

 

But people like Dr. Paul Wolff were making and displaying what we would regard as exhibition size prints already in the late 1930's. And after the war, even the standard 9x12 or 10x15 cm drugstore print made the old standard hopelessly obsolete. But Leitz and Leica have never dared revise their DoF scales and tables. Lots of people did believe—and may still believe—that a lens with a large 'official' DoF is superior to one with a smaller one ...

 

So the demand for a smaller CoC standard has really nothing to do with the size of the M8 sensor, which has just exacerbated an already impossible situation. My remedy is very simple. I point-focus; DoF is negligible with all lenses except super-wide-angles. Sorry about that, folks.

 

The old man from the Age of the Roll Film Folder

[lct]

It is not a matter of film CoC value to me. I mean, 0.03mm works fine for me but whatever CoC value one chooses for film, the M8 CoC will always be 1.33x smaller than the film one right? Then with the same lenses, the hyperfocal distance of the M8 will always be 1.33x wider. And the hyperfocal distance of my APS-C cameras will always be 1.5x wider. And i will still do no so bad focussings when i use the DoF markings of the nearest faster f stop of my lenses hopefully.

[SJP]

The 6.3-25.2 mm, f/2.8-4.9 lens of the D-Lux 2 becomes the 35mm equivalent of a 35-135 mm, f/16-26 lens.

 

I tried with my excel file (including diffraction)

 

6.3 f/2.8 2m Coc 7 micron => 0.97 to infinity

 

35 f/19 2m coc 31 micron => 0.97 m to infinity (note:without diffraction 1.05 to 35.85 m)

 

aperture ratio = 6.8

 

frame factor (1/crop) = 35/6.8 = 5.6

 

So again roughly 20% difference between the 'real' equivalent aperture and the one estimated via the frame factor. Depending on the details of the numbers used no doubt different results will be found.

 

The message as such remains the same - the Dlux at f/2.8 behaves like a FF camera at much smaller f/value. In the car engine business one would say 'there is no substitute for cubic inches'.

 

EDIT: As a final note comparing to DoFmaster values above (use 35mm f/16 2m 30micron) you get from my excel file:

1.145 - 8.590 without diffraction and 1.084 - 15.52 with diffraction. At f/16 including diffraction does make a substantial difference.

[stunsworth]

And i will still do no so bad focussings when i use the DoF markings of the nearest faster f stop of my lenses hopefully.

 

LCT, I lost comprehension of this thread ages ago, but practical experience - using the camera to take photographs - tells me that the one stop trick you mention (aperture at f5.6, use DOF markings at f4) doesn't work with my M8 for large prints. I use a minimum of two stops. For small prints one stop may be ok.

[SJP]

LCT, I lost comprehension of this thread ages ago, but practical experience - using the camera to take photographs - tells me that the one stop trick you mention (aperture at f5.6, use DOF markings at f4) doesn't work with my M8 for large prints. I use a minimum of two stops. For small prints one stop may be ok.

 

For the same print size about ½stop is what you would need but if you want tack sharp large prints then your procedure sounds correct. The whole discussion on equivalent aperture is not relevant if you zoom in to the pixel level - it is all based on the assumption that the print size is the same at the end of the whole process.

[lars_bergquist]

LCT, I lost comprehension of this thread ages ago, but practical experience - using the camera to take photographs - tells me that the one stop trick you mention (aperture at f5.6, use DOF markings at f4) doesn't work with my M8 for large prints. I use a minimum of two stops. For small prints one stop may be ok.

Right, Steve, that's what I do too when using a lens short enough to generate any appreciable DoF. Most of the time I don't want much DoF! The two-stop trick means that you are operating with an assumed maximum CoC of 1/30th / 2 = 1/60th of a mm.

 

The old man from the Age of Roll Film Folders

[SJP]

The two-stop trick means that you are operating with an assumed maximum CoC of 1/30th / 2 = 1/60th of a mm.

No it doesn't - unless you want to believe the simplified rule of thumb. But I agree that the two stop rule is a more or less bullet proof criterion.

 

Even easier regardless of conditions:

set f/16, or smaller if your lens allows, use a tripod, pray, stop breathing, tell your subject to stop moving, this works fine.

 

Achieving maximum depth of field is easy enough, what I think we are discussing is about having a controlled depth of field. In that respect a cropped camera has disadvantages, even more so if the crop is smaller. So your cellphone will never equal a full frame or an M8 which admittedly is not something to get particularly worked up about.

[Advertisement]

Hallo,

Look here: Leica M

[ho_co]

... practical experience ... tells me that the one stop trick you mention (aperture at f5.6, use DOF markings at f4) doesn't work with my M8 for large prints. I use a minimum of two stops. For small prints one stop may be ok.

 

Steve--You're obviously correct. Remember, with film and full-frame, we used as rule-of-thumb one stop extra to get the DoF we wanted. Reason being improvements in emulsions and computations since the time the DoF tables were fixed. Puts, Hußmann, Stefanus all point this out.

 

Now with the M8, yet another extra stop is needed as you say, and the "equivalent aperture" helps point you there: If you consider that, say, f/4 is already the equivalent of about f/5.3 (Karbe's rule of thumb), stopping down only to f/5.6 is obviously almost meaningless, and stopping down to f/8 is necessary.