Diffraction and using a small aperture

[LF_NLD]

Hello everyone.

I have a technical question: in LF photography using really small apertures of f/64 is not uncommon. How does this affect the sharpness of the photo, thinking about diffraction?

And another question. The f-stop is related to the focal length. When you use a 210mm lens on an aperture of f/64, you have an opening equal to 210/64 = 3.3 mm. If you want the same opening on a 18mm lens (e.g. on a small format camera), you need to use 18/3.3 = f/5.4. And that's not at all a large number. So how can you possibly make sharp images using LF ... By using movements ofc!

Ok, I just answered myself. :D A 3.3 mm opening is not at all that 'closed' to cause diffraction, and movements do the rest. I still wanted to post this anyway. :D

[Bruce Barlow]

Call me blind, but I have never seen an image that was degraded in any meaningful way by diffraction. In other words, the gain in depth-of-field far outweighs the degradation from diffraction.

At least, that's my experience.

Rats! I've misplaced my white cane!

[vinny]

I recommend you search the forum and read the numerous threads on diffraction. It's a dead horse that's been well-beat.
Here's (https://www.google.com/search?q=site%3Alargeformatphotography.info+mod54&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-US:official&client=firefox-a&channel=sb#rls=org.mozilla:en-US:official&channel=sb&q=site:largeformatphotography.info+diffraction) 10+ threads to read.

[djdister]

This information from Rodenstock has some interesting conclusions - basically they just say that the "Best Aperture" to use decreases as the film size gets larger, somewhat regardless of the focal length.

Here is the table from the below article:
125519

Here is the full article:
DEPTH OF FIELD VERSUS DIFFRACTION - THE OPTIMUM WORKING APERTURES
http://www.prograf.ru/rodenstock/largeformat_en.html#table2

[jp]

Also not mentioned is that with LF, you either don't enlarge or enlarge by a pretty small multiplier compared to the small formats. A contact printed made on photo paper is going to lose some potential detail as the paper is not as detailed a medium as film.

[ic-racer]

How does this affect the sharpness of the photo, thinking about diffraction?


The effects of diffraction are quite visible. Just expose some large-format images at very small apertures and see for yourself. (eg smaller than f90 for 8x10). If you don't see the effects at the smallest settings on your lenses, be grateful that the burden of thousands of internet posts on the subject has been removed from your back.

[Ken Lee]

To see it for ourselves, we merely need to place an inexpensive loupe on the groundglass and focus on a subject having distinct edges, then gradually stop-down the lens.

Resolution and contrast reach a peak within a certain narrow range of apertures, after which they decline.

My experience matches the Rodenstock numbers exactly, which should come as no surprise: they're the optical physicists after all.

[Emmanuel BIGLER]

I like very much this executive summary about diffraction-related posts,
as stated by Leigh on this forum:

Diffraction limiting is one of those threats that mothers use to get their kids to eat Brussels sprouts, as in:
"If you don't eat your sprouts you'll be diffraction limited for the rest of your life." (http://www.largeformatphotography.info/forum/showthread.php?103382-Stopping-down-and-quot-Diffraction-quot)

P.S. and the numerical value to remember in addition to Leigh's executive summarry, is that a perfect lens stopped down to f/N cannot deliver details smaller than about N microns, and this is totally independant from the focal length.

[Jim Jones]

To quickly observe the degrading of an image due to diffraction, observe the image in an enlarger through a fairly high power grain focusing aid while varying the enlarging lens aperture. These lenses usually limit the minimum aperture so photographers can't screw up by using too small an aperture, but they still stop down enough that one can observe fine grain becoming mushy.

The effect of diffraction in photographs should be evaluated more as an artist than as a scientist. My Cole Weston print of his father's Shell, 1927, seems diffraction limited. Nowhere does it appear perfectly sharp. There is no reason why it should; that photo is about form and luminosity, not acutance. A very small aperture was required to balance depth of field with sharpness. Most other photographs demand a wider aperture as the graph Dan posted suggests.

[Drew Wiley]

Once you start making big precise enlargements it can become quite an important factor. Many LF photographers never understand this for several reasons. 1),
they never make big enlargements; 2) their film plane is never flat in the first place, because a typical filmholder doesn't even have that standard of accuracy
(at least in larger sheet sizes, certainly 5x7 up); 3) they might be contact printing or not have ideal darkroom enlarging techniques to begin with; 4) maybe they've never actually seen a really sharp print. If you've ever seen some of EW's or even AA's classic ole negs blown up, sharp they ain't - pretty much mushballs, in fact. EW intended his work to be mostly contact printed, AA's held up to 20x24 or so (approx to only a 3X enlargement from 8x10) ... after that, then you can understand why he taught that big print are "supposed to be" printed on matte paper less contrasty, otherwise his own looked like hell. ... Now in the real world, there are always tradeoff with depth of field for the sake of uneven focus planes, shutter speed question cause things don't always hold still, on
n' on. I really don't even give a damn about the math - I'm concerned with what shows up in real prints. If it's a subject that I might potentially print quite large,
then I get paranoid about this subject, if not, then I have more options. But I never like stopping 4x5 down smaller than f/32, or 8x10 lower than f/64, or even
f/45 for fussier projects.

[Mark Sawyer]

This information from Rodenstock has some interesting conclusions - basically they just say that the "Best Aperture" to use decreases as the film size gets larger, somewhat regardless of the focal length.

Here is the table from the below article:
125519

Here is the full article:
DEPTH OF FIELD VERSUS DIFFRACTION - THE OPTIMUM WORKING APERTURES
http://www.prograf.ru/rodenstock/largeformat_en.html#table2

So, can somebody explain why if you're using a 360mm lens on 6x9, the "best aperture" is f/16, but if you put the same lens on an 8x10, the "best aperture" is suddenly between f/35(sic) and f/45? :confused:

[Drew Wiley]

This all has to do with degree of enlargement. The smaller the film, the greater the degree of enlargement, print size hypothetically being equal; therefore, the closer you want to be to that particular lens design's optimum f-stop per MTF, blah blah. And if you are talking about the same lens - let's imagine my Fuji 360A, which I use on all the above formats - with smaller film you need far less of the image circle, and hence can indeed use wider stops without fear of vignetting or falloff (which is damn little in this focal length anyway, but might be in a wider lens). That's the optical theory, at least. In the real world, a rollfilm back might tug at the rear standard or not hold film ideally flat, so you might want a smaller stop just for the sake of that, or for the greater potential focus error, trying to work with a smaller area on the groundglass. But let's say I am making a 30x40 print (and I like em crisp, even with people putting their noses right up to the print): an 8x10 is only about a 4X magnification (linear), while 4x5 is 8X, and 6x9 basically useless... BUT, at an equivalent perspective, the 8x10 focal length would have only half the depth of field as 4x5, so would need two stops smaller to achieve the same depth, if that were indeed requisite. There are all kinds of logistical problems and intelligent choices. You can see the effect right on the groundglass. No need for math overkill or calculators, unless you're just into that kind of geeky thing. I'm not. People underestimate the film flatness issue. Many 8x10 photographers stopped the lenses way down there because the
film sagged, and that's the only way they knew how to control it. For anything bigger than 20x24 prints I use adhesive filmholders for 8x10, which keep it flat.
Some old roll-film holders are even worse. You have to pin down all the variable, cause you're only as good as the weakest link.

[Lenny Eiger]

If you aren't shooting coke cans on a tabletop I would ignore diffraction entirely. The consensus here has been that the effect is too small to worry about with large format film. I've scanned the same film, shot from 16 to 45 and all that increased was the depth of field. Certainly when scanning it isn't worth worrying about. The amount is so small that a tiny amount of sharpening will bring it all back.

Besides, what's the big deal about sharpness anyway? I am sure there are some here that love it. However I would posit that most of us would like things pretty sharp, but it doesn't have to look like it was taken with a strobe... that's a definite look that most of us, especially those of us who go outside, don't bother with. Evidenced by the fact we don't carry strobes around with us everywhere...

Lenny

[Maris Rusis]

https://farm4.staticflickr.com/3857/14654726504_8a192e1f70_c.jpg
Snow Gum Dance, Charlotte Pass

Gelatin-silver photograph on Ultrafine Silver Eagle VC FB photographic paper, image size 24.6cm X 19.5cm, from a 8x10 Fomapan 100 negative exposed in a Tachihara 810HD triple extension field view camera fitted with an Apo-Nikkor 610mm f9 lens working at f128. Titled, signed, and stamped verso.

One hears a lot of dismay about the sharpness killing effects of image diffraction. Almost all of it emanates from digital shooters using miniature cameras trying to make big pictures. A big camera making a photograph that won't be enlarged is not so embarrassed. Even though Snow Gum Dance, Charlotte Pass was exposed at f128 it still contains detail finer than the eye can see! But diffraction still lurks. If this photograph were enlarged to 20"x16" some diffraction softening of detail would be just visible. The working aperture would be need to be f100 or bigger, say f90.

[Kirk Gittings]

One hears a lot of dismay about the sharpness killing effects of image diffraction. Almost all of it emanates from digital shooters using miniature cameras trying to make big pictures.

Boy that's not true out this way. I heard a ton about it before digital was ever invented and from LF photographers to boot. One example-I remember giving a talk at a LF conference along with Mark Citret. He casually mentioned in some images from 4x5 he was showing that they were shot at f45. A number of people in the audience took great exception with that practice like he had violated some law of nature because of "diffraction limits". A somewhat heated discussion ensued that carried over into the break. One guy pulled me aside and asked me if I agreed with that "bull". I said I rarely shot at f45 but routinely shot at f32. Oh my god I thought the guy was going to have a heart attack. Mark and and I were dumbfounded by these reactions, scratching our heads about this belief system built around the "law of diffraction limits".

In practice when shooting at f32 or smaller I catch myself thinking that as a result I will probably make a smaller print. This comes from traditional optical printing where it could IMHO soften an image slightly objectionably at 16x20 or above from a 4x5 (depends on the subject matter). But when printing from a file a very slight a very slight amount of additional sharpening overcomes this. Having said this I am a bit of a minimalist with sharpening anyway.

[Kirk Gittings]

Nice image btw Maris.

[Corran]

I have absolutely seen diffraction on really big prints from 4x5. A friend of mine shoots at f/64 pretty much no matter what. No movements, or at least it's really, really rare.

I believe that if you divide the theoretical maximum lp/mm resolution of a given aperture by the necessary lp/mm on paper for a sharp print (say, 5 lp/mm), that gives you a rough estimate for the enlargement factor you can do before noticing diffraction. Of course this assumes the whole system was diffraction limited, and not limited by other things. Work this equation out and generally at the "optimal" aperture you should have a 10x enlargement (f/22 for 4x5) as the limit before diffraction is noticed, all else being equal.

The biggest I've printed in the darkroom is 20x24, and since that's only a 5x enlargement, I don't see these effects, especially compared to other issues, even shooting at f/32 or occasionally f/45. However, negative scans definitely show some softening at extreme magnification when I dip into f/45 and such. And 8x10 shot at f/64 and higher definitely is softer at the same magnification. From my viewpoint, assuming a very good system and of course the same film/developer, a 4x5 shot at f/32 and an 8x10 shot at f/64 should technically look pretty much the same, printed to the same size, at least in terms of resolution. Contact printing of course negates that since there isn't degradation from going through another optical system.

All that being said, I don't really pay that much attention to resolution anymore. Recently another friend told me a print from 4x5 on textured inkjet paper looked really soft to her. Even up close, I thought the print looked really dang sharp, so I guess there's also a lot of personal subjectivity there as well.

[Maris Rusis]

Not sure how you can know that it "...contains detail finer than the eye can see!"

If you can't see it, how can you know that it's there?

How? I look at the photograph from a distance of 25cm which is the optical convention for close vision in search of maximum visual acuity. Then I look at the photograph with a 4x loupe and I see detail I could not see before. Then I call a colleague over and ask them to repeat the experiment. Yup, they see details with the loupe they couldn't see without it.

There is a nice question whether having detail in a photograph finer than unaided eye can see enhances the experience of looking at that photograph. Audiophiles have long claimed that the presence of frequencies beyond hearing range make music sound better. Maybe a photograph with "ïnfinite" detail offers that vague but exhalted quality that's been referred to as "magic realism".

[kintatsu]

I'm shooting mostly 4x5 when it's film, and have noticed nothing. I usually shoot between f/32 and f/64. Even with a 10x loupe on the lightbox, I can't seem to see any difference. It may be there, but being blind-ish and not having an enlarger to do my own work means I can't see it.

I do notice it on digital, though. I often take shots on my digital at the same exposure as my LF and find the loss of detail at f/22 noticeable, while at f/45 on the film, everything seems right.

It's also what causes "starburst" streaks, and is more noticeable with longer exposures than shorter ones in my experience, as the lower light levels further from the source get more exposure.

[Andrew Plume]

Nice image btw Maris.

it certainly is

andrew

[ic-racer]

I do notice it on digital, though. I often take shots on my digital at the same exposure as my LF and find the loss of detail at f/22 noticeable, while at f/45 on the film, everything seems right.
.

One possibility is that when you mount the 4x5" scan back it is at the same plane as your ground glass, but when you mount your film holder, the distance is off.

[kintatsu]

One possibility is that when you mount the 4x5" scan back it is at the same plane as your ground glass, but when you mount your film holder, the distance is off.

Sorry for the confusion. I meant with my DSLR. I wish I could afford an LF digital back!

[Maris Rusis]

Thanks Kirk and Andrew for your kudos on the Snow Gum photograph. I went for f128 so I could get a 2 second exposure. This is as fast as I can do consistently with a lens cap.

[Zndrson]

I have more experience with digital diffraction than I do with diffraction results on film. When I worked as a tech support specialist for a medium format digital retailer I regularly dealt with clients working with 60+ resolution backs on 40.4x53.9mm sensors wondering why their images weren't sharp when they stopped down to F/22.

With such high resolutions using Schneider and especially Rodenstock HR glass, diffraction started at around f/11 and got much much worse afterwards. Most modern lenses being made currently are designed to be shot at 5.6-8 and not much smaller.

As others have said, enlargement size makes all the difference. I haven't tested it personally but I don't plan on shooting at f/45 or smaller on 4x5 and making enlargements. 8x10 I'd be fine with and 11x14 would be no sweat I'd imagine. God I miss shooting 11x14...

[Bill Burk]

Aren't all pinhole shots "diffraction limited"?

Yet the infinite depth of field in these prints are extremely satisfying. Even though we can tell it's all diffracted.

[Dan Fromm]

Aren't all pinhole shots "diffraction limited"?

Yet the infinite depth of field in these prints are extremely satisfying. Even though we can tell it's all diffracted.

Right you are. Fuzzy everywhere.

[Jim Jones]

Pinhole image sharpness is limited by diffraction when the pinhole is smaller than ideal, but by geometric optics when the pinhole is larger than ideal.

[Dan Fromm]

Pinhole image sharpness is limited by diffraction when the pinhole is smaller than ideal, but by geometric optics when the pinhole is larger than ideal.

Huh? Right you are, fuzzy everywhere.

[Bill Burk]

Huh? Right you are, fuzzy everywhere.

In other words, Jim Jones is saying a big hole isn't fuzzy because of "diffraction" but... yes it's fuzzy.

I refer to the graph provided in "Way Beyond Monochrome" when deciding on f/stops to use "in general" when I think I want to avoid diffraction. I assume it's easy enough to find and familiar to everyone here, but if you want a link I'll hunt it down.

[djdister]

I refer to the graph provided in "Way Beyond Monochrome" when deciding on f/stops to use "in general" when I think I want to avoid diffraction. I assume it's easy enough to find and familiar to everyone here, but if you want a link I'll hunt it down.

Does the graph in that book agree with the "Best Aperture" as recommended in the article from Rodenstock, posted earlier?
125616

[timparkin]

One of my old jobs was taking photographs of Army and Air Force Battalions which they would get enlarged to 3 meters wide sometimes and stick at the end of the barracks. I did a bunch of tests on my 4x5 using a Super Symmar 150mm lens at f/11 f/16 f/22 and f/32. The results were scanned at 4000dpi. The f/16 were clearly sharper than the f/11 and f/22. The f/32 was significantly blurry in comparison. We're talking very large enlargements though

I also did a big camera comparison where I looked at the effective reduction in quality of image at various apertures.

at f/16 2/3 4x5 had an equivalent of of 205mp

f/16 2/3 = 205mp
f/22 2/3 = 157mp
f/32 2/3 = 115mp

So you've got a ball park for what stopping down can do to your results. Not many people scan 4x5 at 4000dpi though. I would guess you won't see much difference between f/22 and f/32 if you're scanning at 2000dpi but f/45 will degrade quite a bit. I have in my head that f/22 2/3 is the maximum I would prefer to go to.. f/16 2/3 is optimal. Obviously I'll shoot at whatever aperture gets the shot tough... 115mp is enough for me!!!!

[mdarnton]

Your experience with diffraction is similar to mine, in that there's no arguing that quality changes with f-stop are there, and visible, if one chooses to look, and then you have to decide if you care. I ran all of my 35mm Nikon lenses through my digital Nikon to see how they behaved (this is beautifully easy to check, nearly instantly). Now I know what they can do, and usually, for most pictures, in the most likely printing sizes for me, it doesn't really matter and I don't think about it. But if I'm doing something special, I can consider that aspect, now that I know how it works.

[Drew Wiley]

You're only as good as your weakest link. How many damn times have I said that? Now I know that hard-sharp, relatively high-contrast glossy prints are anathema to Lenny, but I happen to like em! And for me, diffraction is a real and distinctly apparent issue, visible in prints. Now if your film isn't flat in the holder to begin with, or flat in the enlarger carrier, you've got bigger problems to contend with first. But this is a no-brainer to me. Diffraction is a potential issue. And
you do need to print or look at the magnified image with a good focus scope. Judging merely with a lightbox has quite a few pitfalls until you learn the ropes, though it's a start.

[Lenny Eiger]

I say phooey! I don't believe it of a minute. Everyone has to lose all this great depth of field they can get - and for what - a little sharpening applied and its gone!

Yes, I scan my film these days, with a high resolution scanner, at close to 8,000 optical resolution. Diffraction is such as small factor that it should be ignored. I mean jeez, most of the people that scan here are doing it on Epson scanners. That is a far more serious loss of sharpness vs a drum scanner. Yet people muddle thru, learn how to sharpen, and get the results that they feel ok with. Some choose to buy a drum scanner...

Others are using film that isn't TMax or Delta, say FP4+, worse, the 400 speed films like Tri-X or HP5. High solvent developers vs high definition. Or color film in the 400 range. This is also much more of an effect that diffraction is.

I'm not saying diffraction doesn't exist. I'm saying ignore it, and enjoy your depth of field. If you do scan your film, apply a tiny bit of sharpness to it (we all do) and it will go away.

I scan so much film that people shot at 22, and the sharpness falloff is much more serious... I am constantly being asked why their images aren't sharp, and they are, just not exactly all over the image like they want it to be... This diffraction thing is a bad joke, perpetrated on photographers. I'm sure there are lens combinations where its more severe, and digital of course, but not as severe as the ubiquitous loss of depth of field that I am seeing everywhere.

Ignore it!

Lenny

[Bob Salomon]

I say phooey! I don't believe it of a minute. Everyone has to lose all this great depth of field they can get - and for what - a little sharpening applied and its gone!

Yes, I scan my film these days, with a high resolution scanner, at close to 8,000 optical resolution. Diffraction is such as small factor that it should be ignored. I mean jeez, most of the people that scan here are doing it on Epson scanners. That is a far more serious loss of sharpness vs a drum scanner. Yet people muddle thru, learn how to sharpen, and get the results that they feel ok with. Some choose to buy a drum scanner...

Others are using film that isn't TMax or Delta, say FP4+, worse, the 400 speed films like Tri-X or HP5. High solvent developers vs high definition. Or color film in the 400 range. This is also much more of an effect that diffraction is.

I'm not saying diffraction doesn't exist. I'm saying ignore it, and enjoy your depth of field. If you do scan your film, apply a tiny bit of sharpness to it (we all do) and it will go away.

I scan so much film that people shot at 22, and the sharpness falloff is much more serious... I am constantly being asked why their images aren't sharp, and they are, just not exactly all over the image like they want it to be... This diffraction thing is a bad joke, perpetrated on photographers. I'm sure there are lens combinations where its more severe, and digital of course, but not as severe as the ubiquitous loss of depth of field that I am seeing everywhere.

Ignore it!

Lenny

My experience has been that most of that loss of depth of field occurred when people focused on the wrong spot and thus minimized the available depth of field.

[Lenny Eiger]

My experience has been that most of that loss of depth of field occurred when people focused on the wrong spot and thus minimized the available depth of field.

I wouldn't disagree. Lots of people fail to understand how far into the scene they should go - and they also misunderstand when a tilt moves the plane of focus more than they expect, and cuts off the focus at a higher elevation.

Years ago, I got a really good Rodenstock loupe from you - and its been wonderful to really see what's going on.

However, depth of field can help in all of these things. It doesn't fix everything, but having very little depth of field makes all those mistakes worsen.

Lenny

[Bob Salomon]

I wouldn't disagree. Lots of people fail to understand how far into the scene they should go - and they also misunderstand when a tilt moves the plane of focus more than they expect, and cuts off the focus at a higher elevation.

Years ago, I got a really good Rodenstock loupe from you - and its been wonderful to really see what's going on.

However, depth of field can help in all of these things. It doesn't fix everything, but having very little depth of field makes all those mistakes worsen.

Lenny

Enjoy that loupe and treat it well. Rodenstock, as well as Schneider, no longer are making loupes.

[Drew Wiley]

Lenny ... even a sharpness freak like myself also happens to take numerous images where selective focus in the name of the game ... where one arrives at the
correct "look" on the GG, where critical details are put in critical focus, and I don't necessarily want more depth of field in other areas of the scene. Optical facts
and aesthetic choices come into play - so "phooey" on your "phooey" (but not too loud a phooey - don't want to make an enemy of you ... might need your scanning services one of these days!)

[8x10 user]

Diffraction is a big issue in MFD. The highest resolution backs require large apertures and have less depth of field.

A lot of the large format lenses need to be stopped down to F/22 or more for maximum sharpness. Only the best Large format lenses are close to being diffraction limited at F/16. F/22 is a good aperture for most LF work. The theoretical limit for a lens at F/22 is about 3600 DPI

Sharpening can increase the appearance of sharpness but cannot increase the DOF.

Lets say the sharpest chrome films have an average dye cloud of about 8-9 microns. If put into a grid this would be equivalent to ~2800-~3175 DPI. However the dye clouds on film do no appear in a grid but rather are stochastic. There are a few ways to obtain the best possible digital representation of the film. One way would involve scanning the film at an extreme resolution with a sample size much smaller then the film grain. This would yield a file with an extreme amount of noise from "grain aliasing" however it could be possible for the image to be processed (essentially digitally de-screened) in a way that produces a good image. Another way would be to scan the image with an aperture size that is closer to the film grain and then to subsample in order to find the grain edge with more accuracy. This files will also benefit from digital processing only with more of a focus on sharpening rather than noise reduction.

So 3600 DPI might be the limit of resolution at F/22 but a higher resolution scan is helpful to get all of the data.

3600 DPI or ~ a 10x enlargement printed at 360DPI should be close to the limit of the best color films, some films will not hold that much. Some black and white films are able hold more.

At F/32 the limit would be ~ 2540 DPI (Rockwell's (http://www.kenrockwell.com/tech/diffraction.htm) numbers x 2 x 25.4), this is still a good number for 5x7 and 8x10s. Large formats are about more then just resolution. There would be less grain and better tonality in a 2540 PPI 8x10 then a 3600 PPI 4x5.

At F/45 (~1780) is where I would think someone would start to notice the difference.

Images from the F/64 club are limited to ~1270 DPI so they would only be good for maybe a 4x enlargement. However that 4x enlargement is going to look very nice with its high DOF and low grain (due to low reproduction ratio). I was talking to a retired commercial photographer who shot 8x10 and he said he used F/64 (for DOF) for everything and it wasn't an issue because the images were only printed to magazine size.

[8x10 user]

I once saw test results from a Leica and ultra high resolution surveillance film that indicated an obtainable resolution of 16,000 DPI. Of course a 2000 DPI 8x10 will have a much different look.

[Drew Wiley]

What is DPI ? Didn't that go out of style with Seurat? Too much work with that little pointy brush. Retouching dust spots is bad enough.

[8x10 user]

You can replace DPI with PPI. Many talk about DPI when they are really talking about PPI.

[Bob Salomon]

You can replace DPI with PPI. Many talk about DPI when they are really talking about PPI.

Not to be confused with lp/mm that is used on MTF charts.

[8x10 user]

Line pairs per MM x 2 x 25.4 = ~PPI

Getting into the % of transfer makes the numbers more complicated.

[Lenny Eiger]

Everyone keeps coming back to engineering and ppi/spi/dpi, lp/mm and everything else. I do not speak to anything but large format film and lenses, as I don't do digital capture, and haven't done the tests over there myself. FWIW, the issue for me isn't resolution but bit depth, or the ability of the capture medium to represent the largest number of tones.

I would say this. I think (not I know) that most large format photographers are expecting a certain amount of sharpness from their systems. Damn the numbers, and do a test for yourselves. Waste a whole sheet of film, take the shame shot at f22 and f45, do whatever else you do to it, print it in the darkroom or scan it and whatever... and see if you can tell the difference in your print.

My only point is that the difference is so small that most photographers can be happy about shooting with plenty of depth of field.

I have heard the adage "shoot 2 stops down from wide open" and I declare it a falsehood, or meaningless, or whatever you want to call it. I like depth of field and anyone else that also likes it, should be able to have plenty without degrading their image in any significant way. If its true at all, its splitting hairs...


Lenny

[Drew Wiley]

Lenny ... YOU are the one making up artificial rules by stating that this is categorically a non-issue. Well, I cut my teeth printing large Cibachromes from large format chromes, and I absolutely CAN see the effect. So can my viewers. Does it matter to everyone? Of course not. But each of us needs our own set of standards, relative to our own equipment, workflow, and intended kinds of prints. One man's medicine is another man's poison.

[sanking]

Diffraction is neither an opinion nor a religion. It is an optical fact/phenomenon that occurs when light rays are bent around opaque objects, the result of which is that as the aperture is stopped down resolution decreases. Diffraction may or may not impact your work with LF, but it could, so best to understand what it is so so you can use the knowledge to your advantage, rather than just ignoring it.

The effects of diffraction on resolution can be approximated by the formula R=1800/f-N, where R= resolution, 1800 is a constant, and f-N is the f-number. The formula applies to lenses of all focal length. A perfect 20mm lens used at f/16 has the same *maximum possible* resolution, i.e. about 110 lpm, as a 2000 mm lens used at f/16. The formula is not absolute in that resolution is also dependent on wavelength of the light, highest for blue, lowest for red.

The limits of diffraction, which apply to both film and digital capture, depend on format size, print size, and the smallest "blurry circle" acceptable used to calculate the minimum circle of confusion. Ralph Lambrecht, in his book Way Beyond Monochrome, has an interesting chapter called Sharpness and Depth of Field, which treats the issue of diffraction, and other resolution determining issues.

Bearing in mind that the human eye has resolution of about 10-15 lpm at a standard viewing distance of 10" one can fairly quickly calculate from format size and print size required the diffraction limited resolution. For example, if we want to enlarge a 4X5" negative to make a 16X20 print we need to use an aperture of f/32 or less. A perfect lens at f/32 can resolve a maximum of about 55 lpm (1800/16), which at 4X becomes only about 14 lpm. From 8X10 user's post you know that lpm (Line pairs per MM) x 2 x 25.4 = ~PPI, so optimum file size of native resolution for printing with a digital device would be about 720 ppi.

Sandy

[Old-N-Feeble]

Everyone keeps coming back to engineering and ppi/spi/dpi, lp/mm and everything else. I do not speak to anything but large format film and lenses, as I don't do digital capture, and haven't done the tests over there myself. FWIW, the issue for me isn't resolution but bit depth, or the ability of the capture medium to represent the largest number of tones.

I would say this. I think (not I know) that most large format photographers are expecting a certain amount of sharpness from their systems. Damn the numbers, and do a test for yourselves. Waste a whole sheet of film, take the shame shot at f22 and f45, do whatever else you do to it, print it in the darkroom or scan it and whatever... and see if you can tell the difference in your print.

My only point is that the difference is so small that most photographers can be happy about shooting with plenty of depth of field.

I have heard the adage "shoot 2 stops down from wide open" and I declare it a falsehood, or meaningless, or whatever you want to call it. I like depth of field and anyone else that also likes it, should be able to have plenty without degrading their image in any significant way. If its true at all, its splitting hairs...


Lenny

Pinhole should fit your needs quite well.:)

[C. D. Keth]

I'd rather have a picture that's theoretically not as sharp as it could be than no picture at all.

[Lenny Eiger]

Bearing in mind that the human eye has resolution of about 10-15 lpm at a standard viewing distance of 10" one can fairly quickly calculate from format size and print size required the diffraction limited resolution. For example, if we want to enlarge a 4X5" negative to make a 16X20 print we need to use an aperture of f/32 or less. A perfect lens at f/32 can resolve a maximum of about 55 lpm (1800/16), which at 4X becomes only about 14 lpm. From 8X10 user's post you know that lpm (Line pairs per MM) x 2 x 25.4 = ~PPI, so optimum file size of native resolution for printing with a digital device would be about 720 ppi.

Sandy

Sandy, this is exactly the point. If I understand you correctly, your last line suggests an optimum size of 720 ppi. While I can see this difference, most people state they can't see much beyond 300 or 360. Even if they can see past these numbers, they indicate that its just fine for them.

I have made prints much larger than 16x20 from 4x5's.

There is critical sharpness that is available in controlled situation like a studio. Personally, I don't care for this. In fact, I know very few people that do. I'm not a fashion photographer, that's certain.

I am far more interested in the tonality of a print, and how the tones match to each other vs the sharpness. I like translucency, I like images that appear to glow a bit, I like luminescence, and I like a highly tactile quality to a print. These come from working with the various tones.

I think a person's response, or at least my intended response, occurs when there is an abundance of textural information throughout the image. Someone else's aesthetic might be different. The paper matters, the ink set, the way one sets up the inkset in the RIP, the type of scanner, the type of film and development, etc.

I've done a comparison between a Mamiya 6x7, a 4x5 with Sironar S and an 8x10 with Sironar S, and printed them all up to 40 inches, as close as I could get them. If I was after sharpness, the Mamiya 7 wins, hands down. However, the larger format is much smoother when enlarged (with scanner and photoshop) to that size. I am interested in this smoothness.

To suggest that you can't make a good print larger than 16x20 from a 4x5 is, well, let's just say it doesn't match to my experience.

You say it's easy to calculate the numbers exactly. However, when you do so you come up with conclusions I can't agree with. To suggest that a 4x5 at more than f32 can't make a 20 inch print we consider sharp just doesn't square with my experience. The only reason I can come up with is there is some optimal number.

For example, on my scanner, one can Autotrack, as I'm sure you know. If one want's to scan at 10 microns, one can assume that you will get no more than 2540 ppi of resolution. When I scan at 10 microns and 4000, there is a tiny bit of falloff that I can barely see. It isn't anywhere near close to half the resolution. There is also more at 8,000 than there is at 4,000, even tho' most would say there can't be. It's there. There is something wrong with the numbers, they are apparently too absolute for real life. I would say the same is true here. My real world experience goes against these numbers you are presenting. I can not, in all good conscience, having made these prints myself, agree with this.

I remember talking to Phil Lippincott about Nyquist and his theory. Phil could be a curmudgeon, as you know. However, he dismissed Nyqvist's theorem out of hand. I don't know either way. All I know is that something doesn't square with my experience, and as a result, I prefer to do real-world testing rather than rely on the numbers. That's why I suggest hat people should waste a sheet of film once in a while and find out for themselves, with their own materials and equipment.

Lenny

[Dan Fromm]

Oh, dear, not real world testing. The real world testing I've done makes it clear that getting a print that's acceptably sharp -- looks good when viewed with the naked eye from ~ 10" --is very difficult when the print is > 10x the size of the negative, much easier with less enlargement. That's why diffraction, whose effects can be seen on the negs, isn't much of a problem for larger formats printed at relatively small enlargements.

In other words, Lenny's right and so are the people who don't agree with him. You're all talking past each other.

[Corran]

You can't argue the math. 1+1 = 2, you can't say "but I don't see that."

However, the nature of light is quite different than 1+1. I've studied physics and even read Feynman's primer on quantum electrodynamics, and while I by no means purport to know everything about physics/light/QED, I think I do know a bit about it. I think the reason diffraction has become such a hot topic is that it's easily observable in digital devices. Personally, I think the non-random pixel grid of a digital sensor creates an environment where the standard math, Nyquist theory, and all of that gives a very accurate model of the results. In contrast, film and the somewhat random grain along with other factors may make diffraction impact the image differently and perhaps less overall.

I thought about this all last night and this morning and I have some ideas for visual descriptors of this idea. I might have to whip those up and post them.

[Jim Jones]

To avoid wrestling with math or tables in the field, stop down a lens with normal coverage so the entrance pupil is about 5 or 6mm in diameter. The hyperfocal distance for critically sharp negatives will be maybe 2000 times that diameter. For lenses with less or greater coverage, a good entrance pupil and hyperfocal distance will be less or greater in proportion to the focal length. Some mathematical masochist can verify this like I did decades ago with pencil, paper, and slide rule. Since then it has worked well in practice.

The optimum values cited above will vary widely with different subjects, different presentations, and different photographers. It is these differences that cause so much dissention, not the underlying physics. These differences also free photography to be more of an art than a science.

[sanking]

....
I think the reason diffraction has become such a hot topic is that it's easily observable in digital devices. Personally, I think the non-random pixel grid of a digital sensor creates an environment where the standard math, Nyquist theory, and all of that gives a very accurate model of the results. In contrast, film and the somewhat random grain along with other factors may make diffraction impact the image differently and perhaps less overall.
...


This makes sense to me. I do a lot of system testing, i.e. resolution targets, to give an understanding of what the camera system is capable of rendering. There is a strong "relative" component to this since I have been doing this kind of testing for a long time, which provides a basis for understanding the limits of how tests translate into real prints.

I vividly remember the first time I compared a resolution target using a film and digital camera with the same lens. I was quite blown away by the gradual nature of film results compared to the very sudden and absolute nature of digital sensors. Scanning introduces other issues so the bottom line is that optical printing directly from negatives, printing digitally from files of scanned negatives, and printing directly with digital files all have very specific differences and requirements that are of more importance than the mere difference in format of magnification.

Sandy

[Drew Wiley]

Forget the *&**%%$ math and manifestos about how things are "supposed to be done", and above all, trash that useless religion about "normal viewing distance". Try standing for four hours in front of a genuine Vermeer painting like I once did - what is in focus, what is not, and why. He didn't own a damn calculator, yet nobody in history has understood the implications of translating human vision onto a flat surface better. If we are doing our job intelligently, we can make any
damn rules we want in this respect, and apply an aspect of the science we wish, including diffraction, which is indeed real. But we pick and choose our tools as we
need them. This is just another tool.

[Old-N-Feeble]

Forget the *&**%%$ math and manifestos about how things are "supposed to be done", and above all, trash that useless religion about "normal viewing distance". Try standing for four hours in front of a genuine Vermeer painting like I once did - what is in focus, what is not, and why. He didn't own a damn calculator, yet nobody in history has understood the implications of translating human vision onto a flat surface better. If we are doing our job intelligently, we can make any
damn rules we want in this respect, and apply an aspect of the science we wish, including diffraction, which is indeed real. But we pick and choose our tools as we
need them. This is just another tool.

Vermeer was all about the "light" and he was a genius at capturing and/or modifying it. THAT is the most important aspect of art. It's the LIGHT. This has absolutely nothing to do with optical diffraction but it is SOMETHING TO BE PONDERED. HOWEVER... Vermeer's subjects were all IN FOCUS... and per his eyes... not diffraction-limited.

The point is we can't always do this with analog optics and film... no matter how much we want to.

[Sal Santamaura]

...The paper matters, the ink set, the way one sets up the inkset in the RIP, the type of scanner, the type of film and development...Not necessarily. It might be that the paper, type of film / film development, along with enlarging lens / enlarger / enlarger alignment, degree of enlargement and paper developer matter. :D Unless one is making contact prints. :D:D

[Drew Wiley]

Vermeer painted as he actually saw things - imperfectly - and there are all kinds of optical flaws deliberately present in his paintings, because our own eyes are not perfect, nor the quality of window light, etc. Everything is about "light" anyway, otherwise nobody would see anything, so that remark doesn't say anything in
itself. Artificially sharp would be something more like Van Eyck or the various Dutch Minaturists, certainly not Vermeer. Very little is truly crisply painted in his
works - on those specific details which truly count in that respect. Any good monograph on his work will point out these strategic features.

[8x10 user]

I've done a comparison between a Mamiya 6x7, a 4x5 with Sironar S and an 8x10 with Sironar S, and printed them all up to 40 inches, as close as I could get them. If I was after sharpness, the Mamiya 7 wins, hands down. Lenny

Really?



For example, on my scanner, one can Autotrack, as I'm sure you know. If one want's to scan at 10 microns, one can assume that you will get no more than 2540 ppi of resolution. When I scan at 10 microns and 4000, there is a tiny bit of falloff that I can barely see. It isn't anywhere near close to half the resolution. There is also more at 8,000 than there is at 4,000, even tho' most would say there can't be. It's there. There is something wrong with the numbers, they are apparently too absolute for real life. I would say the same is true here. My real world experience goes against these numbers you are presenting. I can not, in all good conscience, having made these prints myself, agree with this.

I remember talking to Phil Lippincott about Nyquist and his theory. Phil could be a curmudgeon, as you know. However, he dismissed Nyqvist's theorem out of hand. I don't know either way. All I know is that something doesn't square with my experience, and as a result, I prefer to do real-world testing rather than rely on the numbers. That's why I suggest hat people should waste a sheet of film once in a while and find out for themselves, with their own materials and equipment. Lenny

The reason why you are getting a better scan with 8000 10 micron samples versus 2540 10 micron samples is due to the stochastic nature of film as I explained in my earlier post.

[Bob Salomon]

"I've done a comparison between a Mamiya 6x7, a 4x5 with Sironar S and an 8x10 with Sironar S, and printed them all up to 40 inches, as close as I could get them. If I was after sharpness, the Mamiya 7 wins, hands down. However, the larger format is much smoother when enlarged (with scanner and photoshop) to that size. I am interested in this smoothness."

But there are lots of other variables that can account for this.

Vibration in the system
Aperture - Optimal or something else?
Magnification - optimal or something else?
Enlarging lens. All designed for the magnification used?
All used at optimal aperture?
Vibration in the enlarging system
Exposure and development
Film grain and fim resolution.

[8x10 user]

DOF versus diffraction is a trade off. Often the amount of DOF desired is an artistic choice. It is generally recommend that a photographer uses the largest aperture while maintaining focus of the area desired. There are a number of DOF calculators to help a photographer determine what the optimal aperture is for each shot. There is one built right into my Sinar camera. You focus at one end of the area that is to be in focus, you turn the gauge until it matches your viewing aperture and then you focus at the other end. The gauge will tell you the optimal aperture for the shot.

If everything else is perfect and you use the right lens/ film then the difference in sharpness between F/45 and F/22 is noticeable. F/45 will be more forgiving of focus mistakes, film sag, and various aberrations that may or may not be present in the lens design. Some lenses are sharper at F/32 then F/16. Most LF lenses are designed for working apertures of F/16 (ED glass) or F/22 (no ED glass)-F/45.

The diffraction limit of F/22 might be 2x that of F/45 however the quality per pixel will be higher from a F/45 chrome scanned at 1700 PPI versus a F/22 image scanned at 3400 PPI scan. Any short falling of the scanner and lens will be more noticeable with the higher resolution scan and since the film resolveablity versus grainalty of the film remains a constant there are diminishing returns when you increase resolution. So even if the theoretical limits of diffraction state that the limit of resolution is 4 times higher (by total number of pixels) at F/22 versus F/45 the actual level of obtainable "quality" is less then 4 times higher. A 1700 PPI scan on 8x10 is still ~206 MP at at F/45 the DOF should be higher then the DOF obtained from a 200MP MFD (Sinar, Hasselblad multishot) back because diffraction forces one to use a very large aperture with such a back.

Some use a larger aperture and focus stack to produce images of the maximum technical quality. I've even heard of this being done with 8x10.


Everyone keeps coming back to engineering and ppi/spi/dpi, lp/mm and everything else. I do not speak to anything but large format film and lenses, as I don't do digital capture, and haven't done the tests over there myself. FWIW, the issue for me isn't resolution but bit depth, or the ability of the capture medium to represent the largest number of tones.

I would say this. I think (not I know) that most large format photographers are expecting a certain amount of sharpness from their systems. Damn the numbers, and do a test for yourselves. Waste a whole sheet of film, take the shame shot at f22 and f45, do whatever else you do to it, print it in the darkroom or scan it and whatever... and see if you can tell the difference in your print.

My only point is that the difference is so small that most photographers can be happy about shooting with plenty of depth of field.

I have heard the adage "shoot 2 stops down from wide open" and I declare it a falsehood, or meaningless, or whatever you want to call it. I like depth of field and anyone else that also likes it, should be able to have plenty without degrading their image in any significant way. If its true at all, its splitting hairs...


Lenny

[Drew Wiley]

You've got to enlarge a 6x7 image over three times as much, so even a garden-variety 4x5 lens will record far more detail than any M7 lens. It's better to compare
apples to apples. I reprinting an exceptional 6x7 neg right now, and the problem is not about getting the same general look in a 16x20 print as with 4x5, but the
sheer headache of dealing with the fact that miniscule film flaws or bit of dust that would be almost imperceptible when enlarging from large format film mean a
lot more nitpickiness in prep and inevitable spotting when working with MF film. Per tonality, sure I'd be rather working with 8x10 film rather than either 4x5 or
6x7; but in this case, I barely had time to set the camera up on a tripod and get the shot. The light would have been totally gone if I had tried using the view camera (which I did have along). But given a lens like an Apo Rodagon N 105 in a very precise carrier, with a premium paper, the tonality itself is pretty darn acceptable.

[8x10 user]

My personal experiences with with Apo Sironar S and the 8x10 image format indicated that somewhere around 3000 PPI is possible under ideal circumstances. The chromes were evaluated with a 12x Schneider loupe, 2 different drum scanners, and an Eversmart Supreme.

[Jac@stafford.net]

My personal experiences with with Apo Sironar S and the 8x10 image format indicated that somewhere around 3000 PPI is possible under ideal circumstances.

I admit to innumeracy. I cannot wrap my head around the "PPI" metric. How big is a "P", and how does it relate to film resolution?

How about good, old "LP/mm"?
.

[Lenny Eiger]

Really?

The difference was huge. Undeniable.

As to Bob's question I did my best to match as many variables as I could. Same film, developer, exposure, etc. The cameras were all set up right next to each other and the shutter clicked within moments of each other - to make sure the light didn't vary. Everything was on a tripod, in my garage, no wind, etc.

There is one more thing I want to say here. I've been doing this as long as any of you. There are times when I was super technical, making graphs and everything else, and other times when I focused on the shooting and just did what I knew worked, adjusting slightly here and there, as needed. Most people that look at my work appreciate the excellence of it. I have the ability to be quite consistent, and I get the results I want. I am not stupid.

I get so tired, it seems any time we talk about technical issues, that some people feel the need to slip in a "well if you knew anything, you'd know that: whatever their point is."

All I am suggesting is that there are times when the numbers don't make sense. Maybe its the stochastic nature of film that yields a certain result, but there are plenty who will argue that "the numbers don't lie". They haven't taken that extra factor into consideration. As long as these conversations can stay mutually respectful I'm happy to participate. I have an opinion, and everyone is welcome to disagree. However, I'd like to have it toned down a bit.

Lenny

[jp]

I admit to innumeracy. I cannot wrap my head around the "PPI" metric. How big is a "P", and how does it relate to film resolution?

How about good, old "LP/mm"?
.

PPI is pretty simple, if it's 3000 ppi, then a P is 1/3000". It's comparable but a little more specific than dots per inch. If you scan film in, you base resizing resolution on pixels, as output detail is measured in PPI for displays or DPI for inkjet or some minimal fraction of DPI if you making a line screen. An image scanned or captured 6000 pixel wide and displayed or printed 20" wide is going to have 300 ppi of information maximum in the file, your monitor or printer will have to resample from that.

[8x10 user]

^ Exactly

DPI (dots per inch) and PPI (pixels per inch) are common terms in dot based printing (inkjet, screen, and printing presses). For example Epson's newest large format inkjet printer can output 360 pixels per inch using a 2880 dots per inch halftone pattern.

[BetterSense]

LP/mm at the film plane is also meaningless unless one specifies the degree of enlargement and viewing distance. If photography were a real field we would express resolution in minutes of arc or steradians.