Diffraction Test

[LeicaS2]

Advertisement (gone after registration)

We all have known that with digital sensors there is a limit of what f stop can be used. But I had yet to actually test the difference.

With the M11 I heard the limit would be f 8.  So I shot this brick building at f 8 f 11, and f 13. With extreme enlargement I could see a difference, enough to never go beyond f 11. But 11 seems OK?

ISO 64 speed 1/250, 350 and 1/500. hand held. M11. Macro-Elmar 90mm  If the full shot were 30x45, the crops would be 3x4

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

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

[Sjz]

It just me or is there turquoise fringing going on with the f11 and f13 that does not exist on the f8?

I do think the first one is sharper

[LeicaS2]

Yes I see the fringing too.  I adjusted LR to remove it at F8, didn't on the others.

[jankap]

Your crops are taken with a very long lens equivalent. Is "hand held" then still adequate?

[pippy]

19 hours ago, LeicaS2 said:

...I shot this brick building at f 8, f 11, and f 13......speed 1/250, 350 and 1/500...

Just for the purposes of clarification in which order did you shoot the images as posted? From top to bottom are they (as stated) f8, f11 and f13 or rather at 1/250, 1/350, 1/500 in which case the apertures would be reversed (f13, f11, f8)?

Philip.

[Kriehuber]

When the Diffraction kicks in, is also dependent on the focal length and the lens design itself. So it can be different for every lens, that said on such small sensors / film like a full frame sensor / 35mm ( or even smaller ) it generally kicks in faster ( 5.6 / 8 / 11 ) then on a larger sensors / film. But the biggest factor is the lens itself.

Edited March 15, 2022 by Kriehuber

[Dennis]

Advertisement (gone after registration)

25 minutes ago, Kriehuber said:

But the biggest factor is the lens itself.

Do you think there is much difference between f8 and f11, for a 28mm on an M10 for example?

On FF sensors, usually, f8 is my limit (cause the diffraction) and it's more than enough for my needs. But I don't pixel peep

[Kriehuber]

depends on the lens, on a 28mm 1.4 diffraction will start in earlier then on a 28mm 2.8 most of the time. ( 28mm 1.4 probably at 4/5.6 and n the 2.8 probably at 8/11, but that also depends on the lens design )

But generally f11 should be totally usable on a full frame sensor when you are not printing big. Most lenses made for full frame sensors will have peak sharpness between 5.6 and 8, but you have to check every lens on its own if wanting perfect sharpness. Luckily with a full frame sensor depth of field isn't a big problem, so you often can use the optimal f stop.

[laowai_]

59 minutes ago, Kriehuber said:

[...] on a 28mm 1.4 diffraction will start in earlier then on a 28mm 2.8 most of the time. ( 28mm 1.4 probably at 4/5.6 and n the 2.8 probably at 8/11, but that also depends on the lens design ) [...]

Why is that? I thought for the same f-number, same focal length and same wavelength, the diameter of the Airy disk (i.e. the diffraction) is the same. 

[LeicaS2]

1.Just for clarification, the order was in f stops.

2. The point for me is that each photographer should do their own test with their own lenses and sensors.

3. I chose to shoot handheld but braced because I rarely will use a tripod with an M. Yes I could redo the test on a tripod with remote release but I decided I had enough information for me. So yes, there remains the question of should I not only have used a tripod but held the same speed and changed the ISO. I was slightly tempted to do this test with more cameras and more lenses, but I’d rather just keep shooting photographs vs tests. But it would be great for someone else to publish some tests. What surprised me was I had not seen any tests at all, just scientific studies.

[Kriehuber]

47 minutes ago, laowai_ said:

Why is that? I thought for the same f-number, same focal length and same wavelength, the diameter of the Airy disk (i.e. the diffraction) is the same. 

I think its because of different lens designs and for what they are optimized for. My 28mm lenses are the sharpest at different apertures and diffraction kicks in on different stops. The same goes for all my lenses for 8x10.  

Obviously diffraction will get more visible with more resolution and you will see bigger difference between the difference f stops. When  shooting 8x10 film, wich captures way more detail than lets say the M11, visible diffraction will kick in earlier and with bigger differences in image quality. When shooting on a M11 and M9 it will be the same. 

 

[Kriehuber]

1 minute ago, LeicaS2 said:

But it would be great for someone else to publish some tests. What surprised me was I had not seen any tests at all, just scientific studies.

I think most of the time, diffraction just doesn't really matter. There is a reason why some Leica ( or for that matter basically all full frame / 35mm lenses ) only stop down to f 16 or f22 and lenses for bigger sensors/ films go to f32/f 45 or even f64/f128. The company's only provide the f stops on the lenses, wich the think are usable and give good results.

[Dennis]

2 hours ago, Kriehuber said:

depends on the lens, on a 28mm 1.4 diffraction will start in earlier then on a 28mm 2.8 most of the time. ( 28mm 1.4 probably at 4/5.6 and n the 2.8 probably at 8/11, but that also depends on the lens design )

My curiosity, started on another thread, was about the difference between f8 and f11 on the new 28/5.6 TT Artisan.. Just curious. Probably the difference is almost nothing

Edited March 15, 2022 by Dennis

[laowai_]

5 hours ago, Kriehuber said:

I think its because of different lens designs and for what they are optimized for. My 28mm lenses are the sharpest at different apertures and diffraction kicks in on different stops.

 

Oh, I see. You are saying that at high f-stops (small opening of the aperture) modern lenses are essentially diffraction limited (i.e. ideal, very close to the theoretical limit). At low f-stops lens performance is limited by aberrations. Somewhere at medium f-stops there is an optimal performance point and that point depends on the lens design, manufacturing tolerances of the various lens elements, accuracy of alignment, quality control, etc. in other words all the many details that Leica is good at. I understand this now, thank you, Kriehuber. 

[Kriehuber]

8 hours ago, Dennis said:

Probably the difference is almost nothing

I agree, for most situations it just doesn't matter. But there are a lot of reasons to test this, its always good to know where your lenses perform best. No matter if you really need it or not.

Edited March 16, 2022 by Kriehuber

[adan]

Decent shirt-cuff experiment.

My main comment would be that linear resolution (line pairs per mm) and thus details of edges and linear textures (screens, bricks, brick seams, textured stone) only go up as the square-root of the megapixels.

People keep thinking "OMG! - 60 Mpixels, 50% more than the M10-R!" - but the M11 can really only count 21% more "pickets in the fence" than the M10-R. (9528 divided by 7864).

It's just not that much sharper to begin with. And the same math applies for any other comparison of resolution (including the dreaded "camera shake" and so on).

Still, 21% is not zero, so there likely will be a small difference in sensitivity to diffraction, subject to all the other caveats others have mentioned.

I tested my M10(vanilla) in the studio, with electronic flash (no shake issue) and the resolution did not really decrease much between f/8 and f/11 - it just sort of flat-lined to about f/13, and then blurred a tiny bit at f/16. At 100% pixel-peeping (but not in A2/21"x14" prints - where either image looked equally "sharp as nails" (75 APO-Summicron)). That was a tiny bit better than the math predicts for 24 Mpixels, but mostly in line with it.

Sometimes I think Leica should include with each camera a large fold-out poster that says (in 200-point type)

BEWARE THE
NARCISSISM OF SMALL DIFFERENCES!

Although I'm not sure it would have any actual effect on Leica users - it's bred in the bone (and too much fun!)

[shirubadanieru]

Is diffraction different on digital full frame VS 35mm film? I tend to stick to F8 anyway but was just curious as I mainly shoot film with a 35 lux pre-asph and a 50 rigid cron.

[adan]

The technical diffraction blurring (the physics) for a lens and aperture is fixed - doesn't depend on whether the image is projected on film, or a sensor, or a piece of paper, or just on air (a virtual image). The light rays do what they do, regardless.

https://www.edmundoptics.com/knowledge-center/application-notes/imaging/limitations-on-resolution-and-contrast-the-airy-disk/

But the higher the resolution/MTF and the lower the "noise" of the light-sensitive medium, the more obvious and limiting a factor the diffraction will be. Digital's MTF and clarity will reveal diffraction that film's natural diffusion and grain may disguise. The medium may blur the picture more than diffraction does.

As always with film, the question is which film? Adox 20, (Kodachrome 25 R.I.P.), Velvia 50, Ektar 100, Superia 200, TX400, Delta 3200 - in order of decreasing signficance.

And as always with digital, which digital? With or without a thick refracting cover glass, or an image-blurring anti-alias filter?

Edited March 16, 2022 by adan

[Jean-Michel]

The OP's images make for an interesting view of diffraction. And I have studied optics in my first year of my photography program eons ago – still have a copy of  McKenzie's "A second course of light"

On my screen, the full size image measures 10 inches wide; the crop area measures 6 inches wide. That area measure 0.5 inches on the full size image.

For a print that would have the cropped area measure 6 inches, the full size of the print would have to be 120 inches wide. Since a general rule for the viewing distance is twice the diagonal of the print, the ideal viewing distance of the resulting 80 by 120 inches print would be 24 feet. Diffraction remains real, but rather moot.

[Advertisement]

Hallo,

Look here: Leica M

[LeicaS2]

Jean-Michel,

When I uploaded the images, the forum requires pixel limits and therefore resolution adjustments that may have contributed to a miscalculation.

The original print was dimension 26" x 39" at 240 resolution. The crop panel would have been 3"x4". the ratio is 8.7 to 1 (26" vs 3") not  20 to 1 ( 10" vs 0.5").

If you were viewing the crop on your screen as 6" vs  3" it would have been the size equivalent of a 52"x78" enlargement at 240 res. For an over 4 foot by 6 foot print the viewing distance would probably be say as much as twelve feet. But in todays world a viewer often then walks right up to the print and views it at as little a foot away just to examine detail.

Separate from a discussion of diffraction, the ability to enlarge a M11 image is impressive. But this is architecture. My experience with landscapes is more challenging; which is why I chose to crop to an area that had vegetation, which is more challenging to resolve in these extreme enlargements.

I resubmitted the images again to the forum. The crop detail to a 6"x8" area that could be uploaded without exceeding the forum limits as is the equivalent of the full image blown up to 52"x78" at 240 res.

 For a 4 foot by 6 foot enlargement the detail looks very good. But back to the original discussion, you are correct; the diffraction with higher f stops is only a factor when one is printing big enlargements. For many photographers, camera shake is going to me a bigger problem, but handled with higher shutter speeds available with higher ISO sensors.

 

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

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