Practical diffraction

[Jim Rice]

I realize after searching "diffraction" that this is hot button issue (though for the life of me, I can't understand why). Rodenstock tells me that the optimum aperture for my 360 Caltar IIN is between 32 and 45 (a stop down from their recommendation for the 210 Sironar-N that never failed me on 4x5" at 22-32). May I just accept this? For my tastes, this issue has always been obfuscated here by either personalities (get well soon Jorge) or physics majors. I'm pretty sure that it varies with focal length, but I'm not certain.

I am also not interested in "good enough for contact printing". I'm sorry, but this seems like a fudge. OTOH, dissussions on CoC will just make my eyes glaze over.

So barring being warned off by the people i respect here, 32~45 is where I am.

The actual question: Am I full of it here?

[Eric Woodbury]

You need to wonder over to the home page HERE. Lots of good info. Start with:

http://www.largeformatphotography.info/fstop.html

and my favorite on optimum f/stop:

http://www.largeformatphotography.info/articles/hansma-dof.1.gif

plus pages 2, 3, and 4, all referenced at the end of the first link.

[Brian Ellis]

If you're not making prints larger than 16x20 from 4x5 film don't worry about it.

[Lazybones]

In my experience, anything f45 and wider seems acceptable, as far as the tradeoff between depth of field and diffraction. I've only had a problem with f64 looking a bit soft on 4x5, and usually not worth the extra depth of field. Probably better to recompose, crop, or switch lenses (if possible) in this situation.

"I am also not interested in 'good enough for contact printing'. I'm sorry, but this seems like a fudge."

I couldn't agree more. So what I'm talking about is squeezing all that nice juicy quality out of that neg/chrome! I try to shoot for flexibility in print size, even if I just want, say, 11x14 for now, I may indeed want to print big in the future. It's when you don't have it and need it, that...

[Leonard Evens]

The usual rule of thumb is that an ideal diffraction limited lens provides you with a resolution in lp/mm approximately equal to 1500 divided by the f-number. So at f/45, such a lens should yield, 1500/45 ~ 33 lp/mm. Combining the lens with a modern film reduces the resolution further, but not by too much. Let's say you end up with about 30 lp/mm. The usual rule of thumb is that the human visual system can resolve between 5 and 10 lp/mm in a print, viewed from about 10-12 inches. If your vision is at the high end, you might see a difference at roughly a 3 X enlargement, but most of us wouldn't see a difference even in a 5 X or 6 X enlargement. One school of thought says that you should consider only a 2 X enlargment since that is the only size you would view that close. People are supposed to get proportionately further back for larger prints. On the other hand, if you want to satisfy 'grain sniffers', you have to worry about what any size print looks like close up.

If you start with the the above rule and consider the various considerations I raised above, you should be able to come up with a reasonable estimate of how far you can stop down without diffraction becoming a problem for the purposes you have in mind.

[Christopher Perez]

... and then there's the chemical effect of the developer with regards to micro contrast and edge enhancement.

I have a friend, Patrick Kolb, who shoots his 19inch Artar at f/45 and soups in PyroCat or ABC and swears his images appear sharper at f/45 than they do at f/22.

It's real easy to get wrapped around an axel about this stuff. The only time I've seen aperture play an important part in overall image quality is in the Biogon design wide angle lenses. Anything below f/32 has produced nothing but mush from lenses I've tried over the years.

[Jim Jones]

. . . The actual question: Am I full of it here?

Certainly not, if what you do satisfies you. Eric's first link is a great compilation of information I've wrestled with for decades. I enjoy such a wrestling match, even if I often lose. Leonard condenses it to a form easier to digest, and which agrees with my experience. The limitations of diffraction are not scientific absolutes when applied to photography. There are important esthetic considerations, too. Edward Weston used an obsolete lens with the iris modified to permit near pinhole apertures, and could produce better images than if he had limited the aperture to the traditional f/64.

A quick look of the effect of diffraction is at hand in an enlarger. Watch a fine grain pattern through a high power focusing magnifier as you adjust the aperture. With my El-Nikkor 50mm f/2.8 the grain in the center of the image is sharpest at about f/4, and is still fairly good at my normal aperture of f/8. By f/16 it is definitely fuzzy.

A rule of thumb suggests, for critical sharpness, a minimum camera lens entrance pupil of no smaller than maybe 5mm if the print is viewed from the distance that gives correct perspective. This is quite subjective. Remember, the most important rule in photography often is "@#$& the rules!"

[jim kitchen]

A few months ago, I decided to add an 8X10 to my backpack, and park my 4X5 for a while. I could not wait to see what brilliant images I could produce with this wonderful piece of equipment. I decided to set my 8x10 lenses to a middle f-stop, my normal practice with my 4X5, set the required tilts and swings, and produce the image. Unfortunately, and disappointingly, my images were extremely DOF challenged compared to my 4X5 images. My negatives were absolutely brilliant though from the tonal point of view, totally grain free, and the images were painfully sharp.

After reviewing, and enriching myself again with the archived notes on DOF and CoC to correct my issue with DOF, I understood I could probably use smaller f-stops, and I should also review my camera setup. I decided to make a few incremental changes, where the greatest change was the smaller f-stop. The smaller f-stop produced images with a better DOF wedge, but I noticed a subtle and uncharacteristic change in my image's tonality, and grain structure. I did not understand this change immediately, and I began to question the ability of my lenses at reproduce objects at infinity. I reviewed the original negative, which drew my attention to DOF, and I quickly realized I traded one problematic issue for another that being DOF for diffraction. I finally determined why my lenses performance changed at infinity, and only performed to my expectations at closer distances.

So, I decided to do an unscientific test with each lens I own on the weekend, to determine the difference in diffraction between the middle f-stop, and the smallest marked f-stop on each lens. My tests indicate diffraction positively affects each lens I own, and diffraction varies with each lens, while using smaller f-stops. How upsetting, but now the issue is understandable. Diffraction produces an incremental amount of grain across the entire 8X10 negative, and produces a softer image quality. I can improve the effects of diffraction, where I can isolate and correct this issue, to a certain degree within Photoshop. I can not however correct improper DOF and out of focus items. I must accept the use of smaller f-stops, to balance DOF, and I will continue to correct the increased grain and loss of image sharpness in Photoshop. I will however, try to improve my images using the middle f-stops, whenever possible.

Attached are two images, where the subject matter was approximately 200m away using f45 and f90 on my Rodenstock 480 Apo-Ronar, indicating the differences.

jim k

[Jeff Conrad]

The optimum f-number for the plane of focus is different from that for the limits of DoF; lens manufacturers' recommendations usually are for the former. With a large focus spread, there sometimes is a tradeoff between center and edge sharpness. Once a lens is essentially diffraction limited, stopping down further decreases sharpness in the plane of focus (though the loss of sharpness may not be apparent until you close down several more stops or examine very large prints at close distances). At the limits of DoF, closing down further reduces the defocus blur and increases sharpness, at least up to a point. Once the "maximum" f-number is reached, stopping down further decreases sharpness even at DoF limits, so there is no point in using a greater f-number.

For that reason, I consider Hansma's "optimum" f-number more of a maximum, because there is no benefit to using a greater f-number. Using a very different approach (calculated MTFs), described at http://www.largeformatphotography.info/articles/DoFinDepth.pdf (PDF) under Diffraction, I get essentially the same results as Hansma. Both Hansma's results and mine seem consistent with what most of the others have said here.

I second the recommendation of Eric's first link. It has the most relevant information on this topic that I've seen in once place.

[Brian Ellis]

Jim Kitchen said:

"Diffraction produces an incremental amount of grain across the entire 8X10 negative, and produces a softer image quality."

I've never heard of diffraction increasing grain. Which doesn't mean it doesn't, there's lots of things I've never heard of. However, FWIW diffraction is caused by the fact that as the aperture gets smaller the proportion of light rays bouncing off the edges of the aperture before they strike the film (as opposed to striking the film directly) gets greater. An oversimplification but accurate enough for present purposes. I don't see how that would increase grain though I could be missing something.

I also don't understand the two thumbnails. Are they scans of an 8x10 print or a cropped section of an 8x10 print or something else? If they're a crop, what's the magnification factor of the crop? I ask because I've never seen diffraction create the degree of blurring shown in these thumbnails. Of course if any negative is enlarged by a big enough factor you'll eventually be able to see the effects of diffraction, even with a lens that's almost wide open, but with wider apertures the mag factor has to be unrealistically great. And even with smaller apertures such as f45 or even f90, in my experience (admittedly limited to my own work) the effects of diffraction don't look like these thumbnails, at least not until the mag factor gets way way up there. Perhaps you could explain what the thumbnails represent.

[jim kitchen]

Dear Brian,

My thumbnail images represent actual pixels from my 8X10 negatives, where they are presented at 100% magnification in PS. One image is at f45, the middle f-stop and f90, the smallest f-stop on my 480 Apo-Ronar. The images were taken within a minute of each other, same exposure, using different f-stops, and developed together. Unfortunately, diffraction is noticeable in every single negative I developed, since I started using a smaller f-stop to overcome my DOF issues. I am trying to demonstrate that a difference does exist in the image quality, specifically with the apparent softness of the image and the incremental grain, while using smaller f-stops with my 8X10 lenses.

Diffraction actually demonstrates that light originates as a wave, and diffraction is a physical law associated with the bending of a light wave, which causes banding and, or the creation of nodal interference points, while passing through an aperture. I believe this little law of physics is the cause of the increased grain and my noticeably softer images.

I just happened to notice a difference in my images, therefore, the test. Originally, my negatives were grain free and sharp as a tack, then I noticed an increase in the grain structure and softer images, while using the smaller f-stops. My processing technique did not change. When I say grain free, I mean absolute smooth tonality across the entire negative.

The f45 image is virtually grain free, and the f90 image is not. This condition exists with each lens I own, but the effect varies with the focal length. I should also qualify my statement by saying that this condition is not as prevalent in my 4X5 negatives, compared to my 8X10 negatives, owed to the apparent greater DOF with the smaller format, while using middle apertures.

The other lenses I tested were my Fujinon 300A, and my Schneider 240 Apo Symmar. I hope this additional information clarifies some the issues I presented.

jim k

[Brian Ellis]

Thanks for the additional information. The fact that these thumbnails represent a 100%magnification in Photoshop at least partially explains why they look the way they do. I still don't understand what you're seeing as additional grain but there's no sense in belaboring the point. You will of course potentially see the effects of diffraction with every lens you own since diffraction is a function of light rays striking the edges of the aperture. But your tests have certainly piqued my interest, I think I'll try some of my own when I have the time. Thanks again.

[Bruce Watson]

My thumbnail images represent actual pixels from my 8X10 negatives, where they are presented at 100% magnification in PS.

This, I think, is a problem. That's up to a 5x enlargement over what you'd get with a real physical print. For example, if your monitor is good for 72 ppi, and you print to your printer at 360 ppi output resolution, that's 360/72 = 5x. So of course it looks bad on the monitor.

IMHO the only way to evaluate how well your system works (that is camera, lenses, film, developer, processing, scanning, Photoshoping, printer, inks, paper, etc.) is to evaluate real prints. Trying to evaluate in Photoshop on a monitor just doesn't cut it.

Try this. Take a chunk out of each of your scans, say 100% pixels chunks that will result in a 30 x 30 cm print (about 12" sq) at optimum output resolution for your printer. Save these to files and make prints from both files. Make sure you use the same processing on both on both files - you want an apples-to-apples comparison.

Now evaluate the prints side-by-side under the same lighting. Tell us what you see. When I first did this, I was amazed. Things that I thought were show stopper artifacts were only visible using a loupe on the prints. So I'm betting in real prints this becomes a non-issue for you too.

For your work, yours is the only opinion that counts. IOW you have to do the tests yourself. Why guess when you can know?

[Leonard Evens]

Most of what Jim Kitchen reports is expected from basic optical theory. The number of usable f-stops is essentially independent of format, at least if you compare final prints of the same size and viewed the same way. If you increase the format size, for the same scene, you need to stop down more to get the same depth of field. At the same time, diffraction is less of a problem because you are going to enlarge less. Going from 4 x 5 to 8 x 10 involves a move of about two stops at both ends, but the total number of usable stops remains the same. Generally, you multiply f-numbers by the ratio of the linear format sizes.

The problem then with larger formats is not basically any difference in depth of field or diffraction. It is that for the same DOF, you are forced to use slower shutter speeds since the required exposure doesn't change.

Of course, there can also be subtler effects such as interaction of diffraction and grain or when considering digital capture, sensor size. Grain and other factors can limit film resolution to something like 60 lp/mm. There are various rules of thumb for combining resolutions, but they all result in a lower resolution than either component alone. For roughly equal resolutions the combination can result in between half and 70 percent the resultion of either compoenent. If one compnent already has a significantly lower resolution than the other, the combination will have resolution smaller than the smaller component resolution, but perhaps not by too much. Thus film resolution will have more of an effect in the middle of the range of usable f-stops than at either end. However, at the largest aperture set by DOF considerations, the effect will be only at the limts of the DOF, while at aperture limited by diffraction, the effect will be an overall softening of the subject. Whether this explains Jim's observations or not is anyone's guess.

Scanning of course has something of the same effect in lowering overall resolution, but since effective scanning resolution is often lower than film resolution, the effect may be more pronounced at either end of the of the f-stop range.

[jim kitchen]

Dear Group,

Thank you for your replies and suggestions...

I completed the differential 16X20 printing tests a few days ago prior to my original post, and the smaller f-stop images are softer and do contain more grain, and therefore require additional adjustments in Photoshop to compete with an image exposed at a middle f-stop. If the uncorrected image is printed, incremental grain is quite visible in an open sky, but not uniquely discernable in congested subject matter within the final print.

While working in Photoshop with an image generated by a drum scan, you witness every imperfection buried in the negative, and although you can visit every pixel on your screen, I try to address and correct the obvious imperfections based upon the final print size, before a print is generated. As I mentioned earlier, this observation became prominent while moving from scanned 4X5 negatives to 8X10 negatives, and although my tests are very unscientific, I now accept my images are softer and contain incremental grain using the smaller apertures with the larger 8X10 format. I will adjust my images in Photoshop; accordingly, to produce a sharp, yet smooth tonal image.

Scanned images are inherently softer for several reasons, but Photoshop allows you several luxuries, which include the ability to sharpen and to add smooth continuous tonality to an image, when an uncharacteristic blemish is present. The larger the image file, the more pixels Photoshop can work with, and any change made to the original file becomes very subtle. My USM corrections very rarely migrate past 80, 0.3, 0 while printing my 24"X30" images.

Anyway, I really appreciate everyone's feedback, and I will adapt my image making process as required. My issues overtook this thread, and should not have.

Thank you again,

jim k

[Chuck Pere]

Can anyone explain the interaction of diffraction and grain. I would think that grain would be more a function of the film and developer used. Or is this something related to the digital darkroom process.

[Leonard Evens]

Chuck,

I don't know if anything more subtle is involved, but whenever you combine two photographic systems, the resolution of the combination is reduced so that it is less than the resolution provided by either alone. This is clear if you use MTFs (modulation transfer functions). That gives you the response in terms of reduced contrast for each spatial frequency in lp/mm. So, say, film provides a 50 percent reduction in resolution at 60 lp/mm, and the lens provides a 40 percent reduction in contrast at that spatial frequency, then the combined system provides a 50 x40 = 20 percent reduction in contrast at 60 lp/mm. You do this for each frequency and you get an MTF curve for the combination. This tells you all you want to know. In practice, resolution of lenses and films is usually given by specifying a spatial frequency in lp/mm at which the reduction in contrast is a certain acceptable amount. This gives a single number, which is easier ot use but gives incomplete information. There is no satisfactory theoretical justification for any specific way to combine two such numbers, but there are two rules of thumb which are in common use. The first says to take the reciprocals of the two resolutions, add them and then take the reciprocal of the result. So, suppose you had a film which, using a single number, resolved 60 lp/mm and a lens which, again using a single number, resolved 50 lp/mm. To get a number for the combination, you would find
1/50 + 1/60 = 110/3000 and take the reciprocal of that to get about 27 lp/mm. The second commonly used rule is similar except that you square each reciprocal before adding, and then take the square root of the answer. In the above example that would give 1/Sqrt(1/2500 + 1/3600) which is about 38 lp/mm. Note that the second method gives a larger number than the first, but given the crudness of trying to specify resolution by a single number, neither gives an entirely accurate idea of what is really happening.

Generally resolution provided by film is a function of more than one factor. One of them is grain.

There may be some further subtle factors, such as how grain affects scanning, for example.

[paulr]

Beware of trying to derive anything meaningful from figures like diffraction limited resolution. A print can have visible detail at extremely high resolutions and still look soft. These numbers tell you nothing about the effect of diffraction on the contrast of fine and medium details (which is what you perceive as sharpness and clarity). The best instrument for measuring that is your eyes, unless you have an optics lab.

[jim kitchen]

Dear Group,

While cruising the net last night, I came across a web page that discusses a few relevant items, such as DOF, Diffraction, Sharpness and CoC. Although the page is skewed to the digital side of image making, it rather illustrates a few of my issues quite nicely.

Anyway, for those of you that might be interested, the web page is here: http://www.cambridgeincolour.com/

I found some interesting items in the tutorial section, specifically under the heading of "Advanced Topics: Understanding Diffraction..." I think their illustration of Airy Disks is rather well done. Place your cursor over the f-stop and watch what happens to the image. There are a few other items of interest also, such as the calculators for DOF, Diffraction Limitations, Hyperfocal Distance, and Noise among other things. Someone spent many hours working on this site...

Thank you again for your replies,

jim k

[Armin Seeholzer]

Hi Jim
Your test is a bit tricky if I see it right then you focused on a plant is this right?
So if there is only a tiny bit of wind it makes sence to me the huge difference.
I do not say it is not possible do to diffraction but more stopped down gets longer times and more chance to movements of the plant!
If you really want to be save do the test on a very stable Rock or building with fine structure!
And then you are shure its diffraction and no motion blur!
Just my 2 cents from Sinarland!

[Leonard Evens]

Jim,

I haven't studied all the tutorials at the website you gave, but I did glance at the one on depth of field. It repeats one of the standard mistakes about depth of field and focal length. For a lot of small format photography, this error may not be too serious in common picture taking situations, but for large format photographers, it is bound to lead to confusion.

The error is in saying that, for constant magnification in the plane of exact focus, the total depth of field for a given relative aperture is essentially independent of focal length. This is an approximation which is valid as long as the subject magnification in that plane is not too small. That usually means you are focusing relatively close to the lens, although you need not actually be in the close-up range. It is definitely not true if you are focusing at, or close to, the hyperfocal distance, for, in that case, the far depth of field in either very large compared to the near depth of field or actually infinite. It is easy to focus well within the hyperfocal distance for a long lens but at or beyond it for a short focal length lens , even if the magnification in the exact plane of focus is the same for both lens. And this happens regularly in large format photography. For example, with my 75 mm lens, the hyperfocal distance at f/22 is about 2.56 meters. If I focus at that distance, everything from 1.28 meters to infinity will be in focus. The magnification would be about 0.03. If I were to maintain the same magnification with my 300 mm lens, I would have to get further back and focus at about 10.2 meters. This is well within the hyperfocal distance for the 300 mm lens at f/22, which is about 41 meters. Everything from roughly 8.2 meters to 13.6 meters would be in focus. In one case, the DOF is infinite and in the other case it is bit over 5 meters. (Warning. I haven't double checked all the arithmetic, so it is possible I got something wrong, but I have done similar calculations many times, and I am confident of the conclusion.)

The truth of the matter is that there are exact formulas which tell you depth of field limits, and it is not possible to find a simple rule which summarizes what those formlas tell you and which works in all common situations.

In any case, when I find someone repeating a blooper like this, I lose confidence that he is right about other things he may say.

[jim kitchen]

Dear Armin,

Thank you for your suggestion, but the images presented, were not influenced by the wind. It is pure unadulterated diffraction. I should have shown the tree stumps and rocks.

Thank you again,

jim k

[jim kitchen]

Dear Leonard,

Out of curiosity, does this DOF example located here: http://www.luminous-landscape.com/tutorials/dof2.shtml illustrate your comment, or am I misunderstanding what you wrote?

jim k