M Lens DOF question

[63strat]

Advertisement (gone after registration)

I have a M8.2 and Summilux 50/1.4 ASPH, and need a wider angle lens with good DOF in the middle F-stops. From what I've read, certain lenses offer greater DOF than others of the same focal length. Speaking of DOF at mid F-stops, how do the Summicron 28/2 and the Elmarit 2/2.8 compare? Does one offer greater DOF than the other at say, 5.6 or 8?

 

And in a related question, Leica says this about their new Elmar 24/3.8:

 

"At the medium f-stops, the depth of field is so large that it can even reproduce entire scenes in sharp focus."

 

Speed notwithstanding, does this lens offer greater DOF at 5.6 than the Summicron 28/2?

 

Thx.

[giordano]

We often get into delicious tangles about depth of field, and this is a great question to start one:D

 

I have a M8.2 and Summilux 50/1.4 ASPH, and need a wider angle lens with good DOF in the middle F-stops. From what I've read, certain lenses offer greater DOF than others of the same focal length.

 

Not really. A less highly corrected lens may give a more gradual transition from "sharpest" via "acceptably sharp" to "out of focus" than a very highly corrected one and thus appear to have more depth of field in some circumstances. But its "sharpest" focus will be less sharp than the other lens too.

 

Speaking of DOF at mid F-stops, how do the Summicron 28/2 and the Elmarit 2/2.8 compare? Does one offer greater DOF than the other at say, 5.6 or 8?

 

If you mean the Elmarit 28/2.8, no.

 

And in a related question, Leica says this about their new Elmar 24/3.8:

 

"At the medium f-stops, the depth of field is so large that it can even reproduce entire scenes in sharp focus."

 

Speed notwithstanding, does this lens offer greater DOF at 5.6 than the Summicron 28/2?

 

Yes - but so does any 24mm lens.

[SJP]

Just to clear up some of the mess:

• the depth of field of lenses of the same focal length and at the same f/value is the same - so a Noctilux at f/2 is the same as a 50 cron at f/2 in terms of depth of field.
  
• the depth of field of a shorter focal length lens at a certain f/value will be greater than the DOF of a longer focal length at the same f/value. So a 28 at f/4 & 5 m has a much larger DOF than a 135 af f/4 & 5m.
  
• if you want a big DOF you need to stop down.
  

 

Speed notwithstanding, does this lens offer greater DOF at 5.6 than the Summicron 28/2?

If you mean the 28/2 cron at f/5.6 then yes, but not too much more.

[lct]

...Speed notwithstanding, does this lens offer greater DOF at 5.6 than the Summicron 28/2...

On the same camera, any 24mm lens has roughly the same DoF at f/5.6 that any 28mm lens at f/8.

[63strat]

Thanks all -- this is what I thought on all these DOF issues, but started to question it all after reading too much.

[lars_bergquist]

DOF depends on two parameters only: (1) the f-stop used, and (2) the reproduction ratio -- the ratio of the subject's size on the sensor to its real size, which is determined by subject distance alone. Focal length in itself has nothing to do with DOF.

 

Take a picture of a subject, first with a short tele and then with a wide angle lens, without changing your position and using the same f-stop. If you crop the wide picture until its edges are the same as those of the tele image, you will find that both images, the tele one and the cropped one, have the same 'perspective' (i.e. the same internal image geometry) and also that the depth of field is the same! The reason why we use other focal lengths than superwide ones is simply that drastically cropped pictures lose quality (they are less sharp, have more grain etc.)

 

When the 9cm Elmar was introduced in 1931, the stated reason was that it produced a higher quality picture than a 5cm image cropped to the same size. Later, photographers discovered that the 9cm lens at two meters gave a portrait a more pleasant perspective than a 5cm lens shoved up the victim's nostrils ... which was true, their mistake was that they thought the improved perspective was due to the focal length, while it was really produced by the longer subject distance.

 

The old man from the Age of the 9cm Elmar

[SJP]

Advertisement (gone after registration)

Mmmmmm... need to think whether that is right, Lars

 

Meanwhile click here Depth of field - Wikipedia, the free encyclopedia the ultimate expression is:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

With N aperture, c circle of confusion, f focal length, s the subject distance

 

Based on that if you keep N, c and s constant the DOF goes down with increasing focal length. This is the 'normal interpretation, wide angle has a larger depth of field than a tele.

 

Back to your proposal. If you crop a wide angle image to a tele perspective and then blow up to the same size of the comparison tele lens you need to change the c value. The magnification is roughly proportional to the focal length so if you go from 28 to 90 you need to blow up the 28 mm image by a factor 3 to get the same perspective & print size. This is equivalent to a sensor crop of the same amount, and that means that the circle of confusion has to be reduced by that factor to get the comparison to work. So in Lars' example c/f is constant (as are s and N). Using that I am still left with an effect of focal length on the DOF, to first approximation decreasing linearly with focal length (easy to see with the expression on page 8 of this document)

[ho_co]

DoF occurs in the brain. It doesn't have a physical existence. It is simply the range of image in front of and behind the actual plane of focus that the eye _perceives_ as "sharp."

 

For that reason, DoF will always generate discussion.

 

If your eyes are better than mine, you'll see less depth of field than I will in the same image. When you stand closer to the print, you'll see less depth of field than when you stand further back.

 

Both Lars and Stephen are right within reason and make good points.

 

But because the formulas are all based on _assumptions_ of magnification and the eye's resolution, they can only be approximate. That's what was so genial about Peter Karbe's discussion of lens equivalency in LFI.

 

And consider another fact with digital: When you print with an inkjet printer, you're depositing points on a piece of paper, and the entire concept of depth of field as a gradual slide in and out of focus is passé.

 

We all grew up with fixed rules about photography. They grew out of the Thirties, Forties and Fifties. But in those days we didn't have 12mm lenses or fisheyes or MTF charts. Rangefinder focusing accuracy was demonstrably greater than that of SLRs except at longer focal lengths.

 

That has changed and so have the rules we learned. The rules are still good approximations over limited ranges; use them, but don't think they are infallible.

 

That's advice from someone with no math training and no sense, so take it or leave it.

[lct]

...The magnification is roughly proportional to the focal length so if you go from 28 to 90 you need to blow up the 28 mm image by a factor 3 to get the same perspective & print size. This is equivalent to a sensor crop of the same amount, and that means that the circle of confusion has to be reduced by that factor to get the comparison to work...

Not sure if i understand what you mean. The CoC does not change with the focal length does it?

[ho_co]

Stephen--

I admire and respect very greatly the knowledge and effort you've demonstrated in creating and publishing your Excel files.

 

But it's interesting to me that here you find that Lars' contention doesn't hold up as focal lengths grow shorter. That was the same result you had with Peter Karbe's contentions in LFI as I very poorly summarized them previously.

 

I wonder if we don't just reach a point where the old formulas just no longer hold.

 

Another forum member commented that he didn't accept Karbe's ideas because they didn't coincide with his own training. But on the other hand, I thought that the brilliance of Dr Karbe's thought was exactly that he found a totally new way to derive depth of field from our standard concepts.

 

I don't want to take a side between your and Lars' viewpoints, because both show a great deal of logic.

 

I am fascinated to hear the justification Lars says Leitz gave about the introduction of a 90mm lens. There are still people who don't recognize that perspective depends only on standpoint, not on focal length. But like you, I hadn't considered pushing the idea of perspective through the idea of depth of field.

 

As the Dalai Lama has said, when we find out that our religion is in conflict with science, we have to change our beliefs. [buddhism traditionally teaches that the sun and the moon are heavenly lights. But as a boy, His Holiness had built his own telescope and observed shadows on the moon. So, he said, Buddhist teaching had to be updated.]

 

Just a thought.

[SJP]

Not sure if i understand what you mean. The CoC does not change with the focal length does it?

Normally not, if you use the whole image of the lens on the sensor. In Lars' example he proposed to blow up a wide angle shot to such an extent that the image would cover the same field as that of a tele lens. That procedure is the same as taking a smaller sensor with crop factor given by the ratio's of the focal lengths. So then you need to change the CoC to keep comparing oranges with oranges.

 

Example focus = 5 m

f=35mm, f/2, COC = 16, DOF = 4.41-5.57 = 1.36 m

f=70mm, f/2, COC = 32, DOF = 4.69-5.35 = 0.66 m

 

This is from the spreadsheet, note that the ratio is roughly the factor 2 that I had predicted.

 

I don't know what others think but for me this is a substantial difference.

[StS]

I guess the issue is indeed the mix of mathematics and subjective factors, as Howard points out.

 

The maximum tolerable size of the circle of confusion is based on experience. For 24mm x 36mm, a circle of 0,033, for 60mm x 60mm a circle of 0,060 was seen as tolerable according to my old handbook. A larger negative need less magnification. Even then, magnifying a picture to a larger poster will mean it will seen by a larger distance, hence the tolerance for larger circles of confusion should be higher.

 

However, the same handbook points out these acceptable circles of confusions are 'rough guidelines' resulting in an angle of about 3 angular minutes. Of course, this is a guideline as well, assuming a 'typical' observer.

 

I must have missed the LFi edition with Mr. Karbe's article. Could you please tell me the edition?

 

Stefan

[SJP]

I think Lars made a very valid point that the perspective does not change, and indeed I guess quite a few of us may tend to forget that, so it is good to be reminded.

 

The summary of Peter Karbe if I recall was that cropping a sensor means that you need to magnify the focal length by the crop factor to get the equivalent focal length, and that the 'equivalent aperture' is reduced by that same factor. This then boils down to the argument that your beloved Noctilux effectively is a 75mm f/1.4 lens on the M8, from the DOF point of view. I fully agree with Peter Karbe with the trends, only the absolute change in equivalent DOF is not quite as severe as he proposed.

 

Anyway I tend to believe that numbers are still relevant, even for a fuzzy concept like DOF.

[ho_co]

I must have missed the LFi edition with Mr. Karbe's article. Could you please tell me the edition?

The article was "Form Follows Format" by Olaf Stefanus, recounting a presentation by Mr Karbe. It was in LFI 3/2006, pp 40-47.

 

 

... The summary of Peter Karbe if I recall was that cropping a sensor means that you need to magnify the focal length by the crop factor to get the equivalent focal length, and that the 'equivalent aperture' is reduced by that same factor. This then boils down to the argument that your beloved Noctilux effectively is a 75mm f/1.4 lens on the M8, from the DOF point of view. I fully agree with Peter Karbe with the trends, only the absolute change in equivalent DOF is not quite as severe as he proposed.

 

Anyway I tend to believe that numbers are still relevant, even for a fuzzy concept like DOF.

 

What fascinates me about this, Stephen, is how much it gets re-hashed. There was the "Lens Equivalents" article at LuLa where two quite competent people fought each other tooth and nail over ideas that we all thought we understood before digital. Next, Rubén Osuna and Efraín García published "Do Sensors 'Outresolve' Lenses?" there, at least attempting to bring those issues to a close.

 

And we have the practical demonstration that what we learned about depth of field in film days simply doesn't hold today, at Digital Focusing Part One et seq.

 

And of course, from the film days, Michael Reichmann's demonstration at DOF2 that "if the subject image size remains the same, then at any given aperture all lenses will give the same depth of field," directly answering the OP's question.

 

It isn't a simple matter, and it's apparently getting less simple every day. And in my case, these things all get squished in my brain.

[lars_bergquist]

My quick and dirty solution to the problem is as follows: With wide angle lenses I halve the apertures on the DOF scale. So if I want to know the DOF for f:8, I look at the markings for f:4. This gives me a more realistic circle of confusion of 1/60th of a mm. The old value of 1/30th, from sometime 1926, is too large even for a 10x15cm print from film.

 

With lenses from 50mm and up, this method does obviously too shallow depths of field to seriously count on. That's bad news, but reality is always to be preferred to illusion. So here I simply forget about DOF and point-focus.

 

With a film or 'full format' camera, you do simply use the next smaller f-stop, like 5.6 for 8.

 

The old man from the Age of Hyperfocal

[SJP]

And of course, from the film days, Michael Reichmann's demonstration at DOF2 that "if the subject image size remains the same, then at any given aperture all lenses will give the same depth of field," directly answering the OP's question.

 

Again this is not exact but probably good enough as an approximation. If I compute the depth of field for a fixed magnification factor then to first order the focal length is unimportant (example at f/2.8: 20mm, 50cm, DOF = 10cm; 400mm, 10m, DOF = 10cm). However this is a very weird procedure as the camera - object distance needs to be chosen differently for each focal length. If you are focussing a 400mm at 10m, then you need to focus a 40mm at 1m. Which defies the point of a wide angle lens from my way of thinking.

 

Anyway, another mess clearing list:

- the depth of field does not change with focal length at constant magnification (image size).

- at a fixed object distance and aperture a wide angle lens has a greater depth of field than a tele.

- a wide angle image can be enlarged and cropped to get the same perspective as a tele lens used at the same object distance, in doing so the depth of field decreases proportional to the focal length.

[lct]

Normally not, if you use the whole image of the lens on the sensor. In Lars' example he proposed to blow up a wide angle shot to such an extent that the image would cover the same field as that of a tele lens....

When we compare say a 35mm and a 90mm lens on the same sensor without cropping the latter for some reason, the CoC remains the same, right? If so, why complicating the matter by cropping considerations?

[SJP]

When we compare say a 35mm and a 90mm lens on the same sensor without cropping the latter for some reason, the CoC remains the same, right? If so, why complicating the matter by cropping considerations?

What Lars was comparing is a 90 mm uncropped and a 35 mm cropped and enlarged to get the 'same' picture or at least the same perspective. The fact that you are enlarging means that you need to redefine the circle of confusion. This is independent of what sensor or film you are using - if you enlarge the CoC changes.

 

The still used standard for film af 31 micron is just based on some standard print size from 1930 and some more or less arbitrary criterion on what they called sharp. This is the way it is, and in fact the criterion is still OK for acceptable 'snapshot' quality. On the M8 the COC gets redefined to 20 micron or so because the sensor is smaller. Obviously you do not want to also start changing the lens depth of field markings, so this gives rise to the 'one stop extra' rule of thumb on the M8. As so nicely summarised by Lars above.

[lct]

Yes i see what you mean thanks but perspective and DoF are different matters.

Never mind i will not repeat what i've been saying for 3 or 4 centuries here but the DoF markings of our lenses are based on the classical CoC values and the latter are good if they work. Do they work? Yes they do with FF cameras, analogue or digital. So where's the problem? The problem comes from crop cameras for which i agree with the rule of thumb you referred to above and i've been repeatedly bothering people about in this good forum. Now we just need to recall that the digital DoF does not behave exactly the same way as the analogue one and that we are in a matter or acceptable sharpness where what is acceptable to me is not acceptable to you necessarily. Suffice it to try.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[SJP]

As natural philosophy degraded into an empirical science some time ago I though it would be worth showing the results of a simple experiment. It should be noted that it is universally accepted that determining DOF is best done on a stuffed shark.

 

It is clearly visible from images 2 and 3 that the DOF using the 90 mm is substantially less than with the cropped 28 mm. Note that indeed the perspective is identical.

 

Image 1: 28mm summicron at f/2.8, 1/30

 

Image 2: image 1 cropped to the 90 mm perspective

 

Image 3: 90mm tele-elmarit at f/2.8, 1/30

 

To make it 'quantitative' the 28-crop DOF is from just in front of the eye until just before the tail fin, the 90 DOF is from just behind the eye until just behind the dorsal fin. Using the span of my hand we have 3 span with the 28mm crop and 1 span with the 90mm. The agreement with the calculation is good enough for me.