Digital, Film Resolution, Diffraction, And Related

[thronobulax]

Continuing on from the thread misplaced in the Steve Sherman discussion. Here is a drilled down look at digital camera diffraction:

Diffraction Limited Photography: Pixel Size, Aperture and Airy Disks

www.cambridgeincolour.com

Digital Camera Diffraction – Resolution, Color & Micro-Contrast

www.cambridgeincolour.com

The first article introduces the topic, the second one gets to the more practical aspects of it. Note that in the diffraction calculator, it shows 24Mpix DSLRs showing some diffraction effects as early as f/5.6

 

[David M]

Interesting articles.
May I address a couple of points?
Aperture is discussed entirely in terms of f-stops. I may be mistaken, but isn’t diffraction related to the absolute size of the opening? Photons (or wavelengths) do not alter their size because you’ve bought a new camera.
Secondly, it seems to me that subjective sharpness is perceived differently in colour. At the boundary between two different patches, there is more information about that boundary if colour is present. Monochrome offers only a change in tonality, whereas a colour image offers both tonality and hue.
As the articles are concerned with digital resolution, there’s no mention of the effect of grain size on perceived sharpness. Perhaps this is a more significant issue for miniature formats, and needn’t concern us.

To answer a previous post…
Of course film is better. But it is less convenient, less quick, less easy and less expensive.

 

[thronobulax]

Interesting articles.
May I address a couple of points?
Aperture is discussed entirely in terms of f-stops. I may be mistaken, but isn’t diffraction related to the absolute size of the opening? Photons (or wavelengths) do not alter their size because you’ve bought a new camera.
Secondly, it seems to me that subjective sharpness is perceived differently in colour. At the boundary between two different patches, there is more information about that boundary if colour is present. Monochrome offers only a change in tonality, whereas a colour image offers both tonality and hue.
As the articles are concerned with digital resolution, there’s no mention of the effect of grain size on perceived sharpness. Perhaps this is a more significant issue for miniature formats, and needn’t concern us.

To answer a previous post…
Of course film is better. But it is less convenient, less quick, less easy and less expensive.

Diffraction is a function both of the wavelength of the light (not the photon size, I don't think) and the size of the aperture. We use "f/stop" as a verbal shorthand when talking about this, but it's the actual opening size that matters. f/stops are a ratio based on focal length. So f/8 on a 150mm lens will be half the size as f/8 on a 300mm lens. That's why diffraction happens at "smaller" f/stops on 5x4 lenses than, say, 35mm lenses - the larger format commands longer focal lengths hence the size of a given f/stop will be larger. But, as you say, it's the absolute opening size that matters for our purposes here. See this for more techie detail:

https://www.edmundoptics.com/knowle...ons-on-resolution-and-contrast-the-airy-disk/

I am unclear about your observations on subjective sharpness in colour because I have no spent any real time exploring it. In my observation, my colour work is less sharp than my monochrome, but that's almost certainly because the bulk of my colour work was shot on 35mm. not MF or LF, so this observation isn't useful.

Digital introduces a whole new spin on all this because of the computational power of post capture. Google et al have demonstrated very sophisticated algorithms to sharpen out of focus pictures, get rid of noise, eliminate distracting backgrounds and the like. I would suggest that this is actually moving digital photography in the direction of computer assisted painting. It's not a bad thing, it's just not photography as ordinarily construed.

Personally - and this is just a matter of taste - I find digitally produced media sterile and lacking character whether we're talking about digital capture or scanned negative or, for that matter, scanned prints. This is almost certainly my strong preference for reflective media vs. transmissive media. I don't like the way images get rendered on an LCD display, even when the image is a scan of one of my own prints. For that matter, now that I think about it, I never much cared for transparencies either ...

 

[David M]

Yes indeed. Forgive me for my attempt to cover wave/particle duality. But we are agreed, I think, that f-stops are a snare and a delusion when discussing diffraction unless other parameters are known and fixed. I'm inclined to think that quoting specific models of small-format cameras might be misleading too.

The difference in perceived sharpness between colour and monochrome is based firstly on my own observation and secondly on the fact that sharpness is rarely mentioned by people discussing colour work, whereas sharpness is mentioned for monochrome. There may be other explanations; it's possible that a visual interest in colour fidelity overrides an interest in sharpness or resolution. This all dates back to before digital imaging raised it's hyper-intelligent head. We have yet to see what AI can do and it's just when we thought it was safe to come out again...

Interestingly, you mention digitally-produced media and your personal distaste for them. But your examples are of digitally-displayed material. An important difference, I suggest.

 

[thronobulax]

Yes indeed. Forgive me for my attempt to cover wave/particle duality. But we are agreed, I think, that f-stops are a snare and a delusion when discussing diffraction unless other parameters are known and fixed. I'm inclined to think that quoting specific models of small-format cameras might be misleading too.

The difference in perceived sharpness between colour and monochrome is based firstly on my own observation and secondly on the fact that sharpness is rarely mentioned by people discussing colour work, whereas sharpness is mentioned for monochrome. There may be other explanations; it's possible that a visual interest in colour fidelity overrides an interest in sharpness or resolution. This all dates back to before digital imaging raised it's hyper-intelligent head. We have yet to see what AI can do and it's just when we thought it was safe to come out again...

Interestingly, you mention digitally-produced media and your personal distaste for them. But your examples are of digitally-displayed material. An important difference, I suggest.

Point taken. In my own mind, a negative scan, a print scan, or an image straight off digital capture are "produced" in the sense that they go through some sort of digital production mechanism on the way to being displayed. But it's entirely right that my objection is manifested in the display mechanisms.

 

[David M]

Many "alternative" producers, platinum printers and the like, go through an intermediate process by making a digital negative – a negative on a clear medium – which they can crop, resize and adjust with greater ease. Then they use this negative to make their final print, by perfectly traditional means. They may use a UV light box rather that God's Own Light, but I think we can forgive them that.
The results are indistinguishable from prints made directly from a camera-negative, at least to my eyes.

 

[thronobulax]

Many "alternative" producers, platinum printers and the like, go through in intermediate process by making a digital negative – a negative on a clear medium – which they can crop, resize and adjust with greater ease. Then they use this negative to make their final print, by perfectly traditional means. They may use a UV light box rather that God's Own Light, but I think we can forgive them that.
The results are indistinguishable from prints made directly from a camera-negative, at least to my eyes.

I'm sure that's right. And, again, the final artefact is reflective which is what my old eyes much prefer.

I was working in the recording studio business when the Moog synth and its ilk first hit. There was all manner of wailing and rending of garments about how this would destroy "real" music, be the death knell of Steinway and Bosendorfer, and so on. Of course, this didn't happen. Digital music became a genre' in its own right distinct from both amplified and acoustic instruments that had preceded it. It was just another fork in the road.

So, in much the same way, I see digital photography as another branch much like romantic photography spun out of painting, and then forked again into the modernist movement. Digital photography has established itself well as a separate artistic vein. But ... I just don't like the output much, most of the time. That said, it's hard to know whether it really is all about it not being reflective, or whether I am writhing under the gross overproduction I see being applied to digital media (if I see one more overly manipulated HDR image with 10,000% saturation I may become clinically insane), or ... maybe I am just a grumpy old fellow stuck in the past...

 

[Reginald S.]

f/stops are a ratio based on focal length.
So f/8 on a 150mm lens will be half the size as f/8 on a 300mm lens.

I believe that's not correct.

A ratio needs two values.
f/stops are an optophysical, a mathematically result coming from the focal length AND the diameter of the lenses entrance pupil.
The entrance pupil herself isn't any fixed number - it's diameter depends on the lens design but not on the FL.
This means that two different 300mm lenses could have very different diameters from the same aperture numbers.
As you said the diameters come from a ratio - but it is a ratio from two variables.

"f/stops" isn't just a term.
It's a formula mostly written and seen on old lenses.
In germany you will see "f:4,5" or f/4,5 and the hint f=300mm, just as an example.

This formula means:
Dividing the focal length "f" of 300mm through a figure of 4,5 will give the diameter of the entrance pupil, which my have the same diameter as the aperture but often it is not the case.

On the other hand with a measurement of the entrance pupil in diameter ( not an easy nor a precise job), with a given focal length, and with changing the formula (dividing 300mm FL through 66,66 EP ) you approximatively can find the wide-open f/stop for an unknown lens.
That's important because from now on and with some logarithmics one can build the aperture row in diameter up to the critical point showing diffraction.
Which should be something about 0,3mm if I remember correct.

 

[thronobulax]

I believe that's not correct.

A ratio needs two values.
f/stops are an optophysical, a mathematically result coming from the focal length AND the diameter of the lenses entrance pupil.
The entrance pupil herself isn't any fixed number - it's diameter depends on the lens design but not on the FL.
This means that two different 300mm lenses could have very different diameters from the same aperture numbers.
As you said the diameters come from a ratio - but it is a ratio from two variables.

"f/stops" isn't just a term.
It's a formula mostly written and seen on old lenses.
In germany you will see "f:4,5" or f/4,5 and the hint f=300mm, just as an example.

This formula means:
Dividing the focal length "f" of 300mm through a figure of 4,5 will give the diameter of the entrance pupil, which my have the same diameter as the aperture but often it is not the case.

On the other hand with a measurement of the entrance pupil in diameter ( not an easy nor a precise job), with a given focal length, and with changing the formula (dividing 300mm FL through 66,66 EP ) you approximatively can find the wide-open f/stop for an unknown lens.
That's important because from now on and with some logarithmics one can build the aperture row in diameter up to the critical point showing diffraction.
Which should be something about 0,3mm if I remember correct.

OK so that's interesting. I'd be obliged if you could provide a link that described this in detail. I was under the apparently wrong impression that f/stop literally meant focal length/f number = diameter.

 

[David M]

That’s the formula they teach us. We can shuffle the elements to derive a missing term.
Then you come across advice to measure the pertinent aperture’s diameter while looking through the elements of the lens, rather than applying your callipers to the opening between the blades. A tricky undertaking, I would have thought.
For practical use in the field, the simpler equation, like the Zone System, might be enough.
I’m looking forward to the more complicated explanation. No knowledge is ever wasted.

A note: I didn’t think anyone was confusing focal length with lens diameter. I wonder if we might be conflating the lens’s specification (eg: For sale - Symmar 300mm 5.6) with working aperture (eg: I shot this with my Symmar 300 at f22.)

…and an addendum.
I believe the simple formula applies only to a single-element lens with no thickness and the aperture in the same plane as the lens.

 

[Ian Grant]

Sometimes lenses have unmarked apertures, For instance I have a Wollensak 159mm f12.5 lens, however I also have a f9 version of the lens, both open to f9 for focussing. In this case the f9 version is slight better optimised during manufacture usually by better matching of lens cells/element, and found by optical bench testing.

My Dallmeyer 4¼" wide angle lens was sold as f6.3 - 11, again the maximum marked aperture is f11, but it opens to f6 for focussing.

Lens design can also skew the usual simple f stop = Focal Length/Diameter. First with Wide angle lenses for SLRs which need a retro focus design to allow clearance for the mirrors, thy don't affect LF users as non were ever made for LF SLRs, unfortunately. More important though are the LF telephoto designs, the challenge here is marked "effective" Focal length is comparative, the actual lens flange distance very significantly less, the lens flange distance for a 360mm Tele-Xenar is 214mm.

Some years ago I bought an ex-Military 17" (432nn) f5.6 Telephoto, almost certainly a Dallmeyer, but no f-stop markings. In fact there are no markings of any kind on the lens, only the lens cap is engraved with the lens details. Knowing the maximum f-stop was f5.6 I measured the diaphragm wide open, from this I was able to back calculate the actual focal length (in terms of focus at Infinity and aperture position relative to the film plane, which is close to the flange distance. Then it was easy to calculate the other aperture diameters, then create an f-stop scale.

For critical work f-stops aren't used, T-stops are used instead, these take into account light transmission as well as the f-stop.

When it comes to diffraction, theory is far less important than practice and knowing how your lens actually behave at minimum apertures is more meaningful. Similarly, we should know how our lenses behave at wider apertures, how edge and corner sharpness drops off as we open them up from f22, then we can exploit lenses to their full. Making use of sharpness fall off can be beneficial for portraiture etc.

Ian

 

[David M]

Yes indeed. In the field, we trade degradation due to diffraction with degradation due to shallow depth of field. (My own choice is with more DoF.) The size of the negative influences our choice here. A small loss of cutting sharpness from diffraction may well be better than a blurry foreground.
Ben Horne (qv) regularly shoots at f64, well into diffractionland, but he’s using 10x8 and for all reasonable enlargements his images appear sharp. He does get some failures which he discusses in his current film-reveal blog.

 

[Ian Grant]

John Sexton is known to shoot at f32, sometimes f45, and with excellent results on 5x4. Someone analysed the apertures he used in "Listen to the trees" and he used f32 for 50% of his images and f45 for 20%

In relative terms f64 with a 300mm lens is similar to f32 and a 15omm lens and is not going to bean issue. Diffraction is really far more of an issue with smaller formats 35mm or Digital (full frame & smaller) where the negative or file are enlarged by a far larger factor than 5x4 or 10x8 negatives.

f16 with a 50mm lens on 35mm is a smaller diameter aperture than f45 with a 150mm lens on a 5x4 camera, or f90 with a 300mm lens on 10x8, that puts some perspective on an issue that gets over-hyped.

Ian

 

[David M]

Ian’s right. Despite its fascination it belongs in the same corner as the Sacred 0.01 above B+F and the number of inversions that can dance on the head of a pin.
Despite that, I’d still like to know a bit more if possible. The inner geek is clamouring.

 

[thronobulax]

Ian’s right. Despite its fascination it belongs in the same corner as the Sacred 0.01 above B+F and the number of inversions that can dance on the head of a pin.
Despite that, I’d still like to know a bit more if possible. The inner geek is clamouring.

My Inner Engineer wants to understand the nuances. My Inner Photographer doesn't much care.

 

[Reginald S.]

I'd be obliged if you could provide a link that described this in detail .
I was under the apparently wrong impression that f/stop literally meant focal length/f number = diameter.

Sorry, there is no link.
I'm referring to ancient literature (coming from anno 1900) from where I studied some articles years ago.
These articles describe the precise determination of focal length and luminous intensity/amount of light passing a wide open lens system.
David, this articles very physically describe the science for every lens, not only a single lens system which always makes things more easy than systems with 4, 6 or 8 single lenses or different lensgroups.

For a deeper look please make a research for "Compendium der practischen Photographie" from Professor F. Schmidt, Docent der Photographie an der Technischen Hochschule Karlsruhe Baden / Wiesbaden Verlag from Otto Nemmich 1899

Yes, we are talking about diameters and focal length.

Focal length is clear, but "diameter" can say different things.
The most important thing is the wide opening /without any aperture) of a lens from where engineers start to determine the luminous intensity.
This is the effective wide opening.
You can bring an aperture with the same diameter to the lens but then the effective W.O. changes to a relative wide opening, concerning the behaviour of light rays passing a convex front lens in a convex way.

The entrance pupil is the visable diameter from wide opening and a base for measurements, and the aperture is just a tool with it's diameter depending on the entrance pupil.
To my understanding the entrance pupil gives the description of the wide opening diameter.

The visable entrance pupil either lays directly behind the aperture or somewhere within the lens system, depending on the lens design.
By the way there also exists an outgoing or exit pupil which can live her live widely outside of the lens what often is the case to real Tele constructions described by Ian..

Now, to calculate the correct diameter of the physical aperture you need the pupil magnification of the system.
To calculate the pupil magnification, you need the correct focal length (which often isn't the seen number on a lens) and the position of the main planes of the part of the lens in front of the aperture.
Such data are rather difficult to obtain but you can also try to measure or estimate with calipers and things as described before.

Mr. Belitski ( Deutsche Photogr. Zeitung Nr. 7 und 8, from 1889 ) described an experiment for determining the effective wide opening by making visable the focal point of a lens .
This visable focal point will give a perfect measurement of the entrance pupil.

I am aware about the logical f/stop system, and to all of us should be clear that an f/8 will give the same amount of passing light to every lens .
But as said before I can't imagine "f/8 on a 150mm lens will be half the size as f/8 on a 300mm lens" ,
because an f/stop term tells about the amount of light instead of the size of aperture diameters so I believe.

So next I want to determine some values from my big 300mm Tessar and my smaller 305mm G-Claron and probably from my 360mm Heliar (huge) and a 360mm Tele Xenar (small in direct comparism to the Heliar).
I would be surprised seeing results which could lead to "f/8 on a 150mm lens will be half the size as f/8 on a 300mm lens".

Best,
Reginald

 

[thronobulax]

As it happens, I have a variety of lens in many focal lengths here. I will attempt measurement as well. Now I am really curious.

 

[Ian Grant]

But as said before I can't imagine "f/8 on a 150mm lens will be half the size as f/8 on a 300mm lens"

With the same type of basic lens design the diameter of the f8 aperture of a 150mm lens is 18.75mm, a 300mm at f8 has a diameter of 37.50mm, so the 150mm is exactly half the diameter. No imagination needed it's pure mathematics, however in terms of area the 150mm f8 is 1104.5 sq mm, the 300mm f8 is considerably larger at 4418.0 sq mm, four times the area, but it's the diameter that is relevant to aperture.

Looking at it another way put a 150mm Symmar set to f8, shutter open, alongside a 300mm Symmar at f16 and the aperture diameters will be the same 18,75mm, it really is that simple.

Ian

 

[Reginald S.]

With the same type of basic lens design

So here we have the first restriction of the Formula Thronobulax ?
But I can see the point said from myself that using the typical formula based on a given aperture and the focal length leeds to the wanted result.
I am still looking for " f/8 on a 150mm lens will be half the size as f/8 on a 300mm lens" with my own lenses because T's formula sounds valid to all lenses.
If I'm ending wrong, I probably have to train my mathematic capabilities

 

[Ian Grant]

So here we have the first restriction of the Formula Thronobulax ?

It's not a restriction, as I mentioned earlier certain lens designs like retro focus wide angles and telephoto lenses, and I could add zooms don't conform. For simplicity I was comparing like to like with a symmetrical lens design.

I am still looking for " f/8 on a 150mm lens will be half the size as f/8 on a 300mm lens" with my own lenses because T's formula sounds valid to all lenses.
If I'm ending wrong, I probably have to train my mathematic capabilities

Do the maths yourself: f-stop= Focal length / diameter of the aperture, or rather start with: Focal Length / f-stop = Diameter of aperture.

Remember it is a basic formula, more complex lens designs will vary slightly, and as mentioned above some types very significantly, the important factor is the back focus of a lens rather than the marked Focal Length, at Infinity.

It's also important to remember that lenses for LF cameras, and also 35mm and 120 rangefinder cameras (I could add Mirror less Digital), use purer designs, they don't have to compromise design to accommodate a mirror system it's also why many of our lenses are essentially symmetrical, particularly wide angles.

Ian