Diffraction effect (originally titled: Reciprocal Failure) quick test MF

[David A. Goldfarb]

hi Bart,

in any case, when I was shooting on a 4x5", with film, the diffraction effect was MUCH amplified with the reproduction scale (magnification). At 1:1, e.g., one could visibly see diffraction with f45+, when at infinity, this would not influence f-stop would not have any influence. And of course, at higher reproduction scales, e.g. 2:1 this was much worse.

How much the reproduction scale plays on a digital sensor like we are used to have them currently, I couldn't say for not having tested it, but certainly it does from a certain scale, respectively from a certain lens to image plane length.

Kind regards,
Thierry

This is my sense as well. When I'm at 1:2 magnification or greater, I tend to be much more careful of avoiding small apertures than I might be with greater subject distances.

 

[Ken Tanaka]

hi Bart,

in any case, when I was shooting on a 4x5", with film, the diffraction effect was MUCH amplified with the reproduction scale (magnification). At 1:1, e.g., one could visibly see diffraction with f45+, when at infinity, such a closed f-stop would not have any influence. And of course, at higher reproduction scales, e.g. 2:1 this was much worse.

How much the reproduction scale plays on a digital sensor like we are used to have them currently, I couldn't say for not having tested it, but certainly it does from a certain scale, respectively from a certain lens to image plane length.

Kind regards,
Thierry

That's exactly the observation to which I clumsily referred earlier, Thierry. Big image circles + small subject areas = PF. ("Potential Funkiness")

PF is also more likely in macrophotography, something with which I'm now dabbling with an MP-E 65mm lens.

But I still maintain that in common styles of photography, particularly 35mm-format, diffraction distortion is not a phenomenon worth spending any time worrying about. Amateur photographers, and even most professionals, would profit far more by devoting attentions toward many more substantial issues both in front of, and behind, their lenses.

 

[Thierry Hagenauer]

... and/or use a view camera, when possible, to adjust the plane of sharpness in order to reduce the stopping down and bring it to the optimal quality the lens can give.

Best regards,
Thierry

Amateur photographers, and even most professionals, would profit far more by devoting attentions toward many more substantial issues both in front of, and behind, their lenses.

 

[leonardobarreto.com]

"diffraction distortion is not a phenomenon worth spending any time worrying about"

For some reason I see the phenomenon come up a lot this days when people talk about the fine bouquet of a given lens, or how some cameras are reaching resolving frontiers of certain optic lenses. Also it may be because people may spend $8,000 good US dollars on a new camera where they have managed to squeeze 24 million pixels ... (D3 and D3x are identical cameras except for the price and pixel population)..

 

[David A. Goldfarb]

I think what Ken is suggesting, and what I find to be true as well, is that diffraction is usually the least of one's problems. Except in cases like macrophotography or when you know you'll be making a really big enlargement, having adequate DOF is usually a more important issue than fuzziness due to diffraction.

 

[leonardobarreto.com]

Not being argumentative (just a little LOL) but, why would you buy a Canon 1DsMk3 or D3X if you don't want to be "be making a really big enlargement" your are better of with a D3 or 1Ds and save $8k. It is as if you get a Ferrari that you know will begin shaking after 200km/h and you say: "just drive at 100km/h"

 

[David A. Goldfarb]

You might as well ask why anyone buys a high end camera or lens and uses it without a tripod in continuous light. A tripod will almost always make more of a difference in sharpness than diffraction or even choice of lens, if one is concerned about sharpness, which may not always be the case. It just seem to me that people seem to get hung up on diffraction before addressing other issues that are more obvious.

 

[John Sheehy]

In the banknote example, the effect of diffraction is exaggerated, because the effective aperture is considerably smaller than the f:number set on the lens, isn't it? I don't recall offhand, but you remember these things better than I do, Bart. Is the effective aperture relevant for calculating the maximum resolution limited by diffraction? It seems like it should, and if it is, then the effect of diffraction at a given f:stop should be less in a landscape than in a macro shot.

I've heard many people claim this. I guess a good test would be to take the same macro lens, and do a varying f-stop test for distant, and for highly magnified subjects separately.

 

[John Sheehy]

Not being argumentative (just a little LOL) but, why would you buy a Canon 1DsMk3 or D3X if you don't want to be "be making a really big enlargement" your are better of with a D3 or 1Ds and save $8k. It is as if you get a Ferrari that you know will begin shaking after 200km/h and you say: "just drive at 100km/h"

The images from the higher-res cameras are better, whether you take care to get the most out of them or not. The difference is between better and a lot better; not between nothing (or even worse, as many claim) and a lot better.

Higher-MP cameras are not more demanding; they simply have more potential. Higher-MP images are more easy to edit away problems to satisfaction. They rotate better. They CA correct better. They perspective correct better. They resize better.

If you think a high-MP image shows problems more, compare it to a similar lower-MP image upsized to equal the higher-MP one in image magnification. It turns to relative mush.

Any notion that higher pixel density worsens image diffraction, camera shake, subject motion, lens aberrations, etc, is groundless. All these things are worse when combined with the blur of lower pixel density.

 

[leonardobarreto.com]

So we should look forward to the 1DsMk7 with 60megapixels of "improved resolution" ?

Or is there a point where we hit the shoulder of the curve and get less and less return from squeezing more millions of sensels in a given real state?

 

[John Sheehy]

So we should look forward to the 1DsMk7 with 60megapixels of "improved resolution" ?

Or is there a point where we hit the shoulder of the curve and get less and less return from squeezing more millions of sensels in a given real state?

The point of "almost no" return depends on the optics. The central areas of the sharpest lenses with sharp focus and camera/subject stability, in their sweet spots, should facilitate pixel pitches in DSLRs smaller than in current 14MP P&S cameras; over 250MP for full-frame, 100 for DX.

There may be more of a limit for full-frame than DX, because of the corner qualities. Lenses that are very sharp in the FF corners are very hard and expensive to make, and sensor MFRs may use the corners as a reason/excuse to limit FF pixel densities. However, a lossy RAW could zone the sensor based on potential resolution for each area based on aperture and lens, and keep the output to a reasonable size.

Of course, it is not known at this point in time if it is practical to get low high-ISO noise the way they do now with very small pixel pitches, so an extremely hi-res camera may have excellent resolution and image-level (what DXO calls "Print") DR at base ISO, but fall a bit behind lower pixel densities at high ISOs in shadow noise.

 

[Ken Tanaka]

We have reached the "point of almost no return"...and it is entirely unrelated to optics. It turns out to be most directly related to global economics.

 

[leonardobarreto.com]

Economics is another factor, but not necessarily stopping this process. The first digital camera I got was an Olympus 2mp that cost me $1k now you can almost get a 5DMk2 for that ...

I remember how long it took Mamiya to go from RZ to RZ II and there was almost no change at all. 35mm cameras changed much faster, first with AE, then AF. and the digital transition has been really fast, but now we may be getting back to the point where we have what we want?

I have two (and a half, considering my NIKONOS) systems: DX and MF. With my D300 I have smaller lenses, a very fast and responsive camera -I don't do sports- and files perfect for sending to my stock collection. Then I have a 22mp 6x 4.5 P25 digital back that I can use on my SINAR, super clean (9x9 micron) and Capture 1 etc etc. so, I really don't aspire to more.

... but a 250mp Nikon (or Canon) sounds interesting ...

We have reached the "point of almost no return"...and it is entirely unrelated to optics. It turns out to be most directly related to global economics.

 

[John Sheehy]

... but a 250mp Nikon (or Canon) sounds interesting ...

The 100MP DX is probably more practical.

We've been hearing rumors of crop cameras being phased out, especially with Canon, but FF seems to be running into more issues. FF definitely provides for less noise in an image, but it's going to be a lot harder to get the corners and edges of the frame sharp enough to take advantage at the edges as can be done in the center of the frame, at the wider angle FOVs that most people shoot at (super teles and macros are much less of a problem).

 

[Bart_van_der_Wolf]

I'll digg up some aperture bracketing images to demonstrate the principle at close to infinity focus.

As promised, an aperture bracketing series (crops at 100% zoom) taken with a TS-E 90mm f/2.8 lens, on the same camera body as the macro series:

The result is consistent with what was more clearly shown with the macro. There are unfortunately some fluctuations due to changing light, but I suggest to focus on low contrast detail, like the thatched roof, or even the bricks, or one of the ropes that is actually a chain. The moment you go from f/8 to f/11 into diffraction affected territory for this sensel pitch, the loss is visible. Mind you, I wouldn't hesitate to use f/11, or even reluctantly use f/16 if the situation calls for it, but I'll also know that subsequent deconvolution sharpening is needed to recover some of the losses.

The images were not sharpened beyond the Capture One Raw conversion process with all sharpening turned off. Capture One is the Raw converter that extracts the most detail from the 1Ds3 files.

The important thing to realise is that there is more to diffraction than a loss of sharpness. The MTF will suffer, and that will start to affect the overall cripness of the image in addition to things near the limiting resolution. Of course the MTF is the result of all (optical) components in the chain, so individual lenses may react a bit differently on stopping down.

Of course there is more to an image than its technical quality alone. I'm just trying to make sure that the implications of our choices are know, so we can make informed choices when we can choose different settings. Again, the technical quality mostly plays a role with significant enlargements when we want the best that our gear can provide. That also includes the use of tripods if possible.

Bart

 

[leonardobarreto.com]

Bart:

This is another fantastic demonstration on the importance of Diffraction -and I have learned more than the proper name with this thread--. I just spent the weekend in the Bolivian semi tropics with my D300 and was looking at my apertures all the time. F.8 F.11, tomorrow I will be looking at all the data (photos) at 100%..

Thanks, you have singlehandedly saved this thread, LoL,

Regards,

Leonardo

 

[Bart_van_der_Wolf]

Thanks, you have singlehandedly saved this thread, LoL,

Hi Leonardo,

You're too kind. It took someone like you to start the thread in the first place, then the exchange of ideas followed ...

I also agree with Ken and David that there is more to a good photograph than technique alone, but that wasn't the topic of the thread.

It is interesting what Thierry has to say about his observations, that with macro photography magnifications it seems like there is more of a diffraction effect visible. I think it is mostly caused by the shallow DOF which makes it look like the little sharpness that remains suffers more. I cannot demonstrate such an objective difference, so it must be subjective. Aperture size/shape and wavelength are the only contributors to diffraction, and the sensel size/pitch determines how accurate we can sample the pattern and thus what the per pixel sharpness will be (if we require that for enlargements).

I would be interested to see if the MF cameras without AA-filter and microlenses also show visible diffraction effects when the diffraction pattern diameter (= 2.44 x 0.555 x aperture#) exceeds 1.5x the sensel pitch, or whether that threshold can be pinpointed to another multiplier. The Raw converter used (and its secret default minimum sharpening) can also make a bit of a difference.

Bart

 

[David A. Goldfarb]

Interesting demonstration. I'm wondering if what Thierry and I have observed shooting large format may just have to do with the way diffraction looks on film or the groundglass without the added factor of sensel pitch.

 

[Cem_Usakligil]

I have edited the title of this thread to make searching for the diffraction effect possible, which is what this thread is really about.

 

[leonardobarreto.com]

Thank you for changing the topic name, there is much to assimilate and incorporate in our daily work flow.

In Bart example the pre-capture work flow that could be presented to the photographer could now be:

If increased depth of field is not strictly needed, and since D of F increases with distance, (or wide angles), it would be a good choice to "assign" one or two f. stops to faster shutter speeds and keep the aperture at bellow f.16/f.11

For photographers doing art reproduction work. They should test their own equipment and then use the optimum f.stop to get the must out of the images.

Bart, one more observation. It is interesting to compare results at f.32 as opposed to f.2.8 and how this effect would (or should) change the notion we all carry on from the film era that closing the lens is a good way to improve IQ. Remember f. 64 from Ansel Adams?

 

[David A. Goldfarb]

Adams was shooting mostly large format, and in reality didn't use f:64 that often, so the reduced enlargement factor more than makes up for any loss due to diffraction.

For 4x5", f:22 is a nice middle range aperture, not unlike f:8 on a 35mm camera. For 8x10", it's not so unusual to be using f:32 or f:45.

 

[Bart_van_der_Wolf]

Bart, one more observation. It is interesting to compare results at f.32 as opposed to f.2.8 and how this effect would (or should) change the notion we all carry on from the film era that closing the lens is a good way to improve IQ. Remember f. 64 from Ansel Adams?

Well, in a way it can be a good thing because it reduces other lens aberrations, but one just has to be careful how far one closes the aperture. Some of these aberrations are also responsible for the creamy bokeh at wider apertures so on has to balance things to fit the subject and intended recording of it.

Adams was shooting mostly large format, and in reality didn't use f:64 that often, so the reduced enlargement factor more than makes up for any loss due to diffraction.

Exactly, the diffraction blur has a predictable effect on sharpness. As long as we don't magnify the finest detail enough to resolve by eye, we will hardly notice it (besides the effect it has on lowering the MTF). Another reason he closed down the aperture is because the image quality with lenses with such a large image circle 'suffers' when used wide open.

By the way, the effect of slight defocus on killing detail is much stronger because it has a larger blurring effect than diffraction has. The (more subtle) degradation by diffraction can also be used to reduce/avoid aliasing artifacts, so it is important to know your equipment if you want to master the challenges we're faced with.

For 4x5", f:22 is a nice middle range aperture, not unlike f:8 on a 35mm camera. For 8x10", it's not so unusual to be using f:32 or f:45.

Yes, diffraction scales linearly with output magnification, so one can get away with a lot more of it on larger recording media that require less magnification. Diffraction also scales linearly with aperture number, so whatever threshold one uses it allows to scale the aperture with the linear size increase of the capture medium.

The 'only' thing that sets digital aside from analog with regards to this subject, is that digital has an abrupt lower limit on resolution (the sensel size), where analog/film just keeps on going until the MTF drowns in graininess.

Bart

 

[Thierry Hagenauer]

hi Bart,

for what it is worth, I believe that the diffraction is highly dependent on the reproduction scale, respectively the distance lens-image plane.

To illustrate it I have here a little drawing: one can understand, that if the distance lens to image plane becomes longer, the broken light rays at the edge of the aperture blades do "spread" wider when they hit the image plane, thus causing more fuzziness.

That's my only explanation, and it becomes really dramatic when working at around 2:1 and then closing at 45 on a 4x5" camera, when f45 or even 64 at infinity (shortest possible lens to image plane distance) there is nothing noticeable (there is a famous club, don't know if it still exists, the "f64" club, suggesting to shoot at f64: they are/were "landscapers", thus shooting at infinity or close to, without noticeable diffraction).

Best regards,
Thierry

Hi Leonardo,

It is interesting what Thierry has to say about his observations, that with macro photography magnifications it seems like there is more of a diffraction effect visible. I think it is mostly caused by the shallow DOF which makes it look like the little sharpness that remains suffers more. I cannot demonstrate such an objective difference, so it must be subjective. Aperture size/shape and wavelength are the only contributors to diffraction, and the sensel size/pitch determines how accurate we can sample the pattern and thus what the per pixel sharpness will be (if we require that for enlargements).


Bart

 

[Jack_Flesher]

Stated aperture of any lens is only accurate when the lens is at infinity focus. With conventional lenses, as one focuses closer, focal length increases and thus aperture decreases proportionately. (Not necessarily so for IF or Internal Focus lenses though.) This effect is so small as to be insignificant for exposures at less than 1:5 reproduction ratios. However, as you focus closer, the need to compensate for the added lens extension kicks in for this very reason.

Thus, diffraction effects will be greater for any given lens and aperture combo when used for close subjects relative to subjects at infinity...

Aperture formula, f = F/D where f is aperture number, F is the lens' focal length and D is the aperture diameter.

Exposure formula (one of them), E = (I/O+1)^2 where E is the Exposure factor, I is the Image dimension at the sensor plane and O is the Object dimension actual. Note that an exposure factor of 4 is equal to a 4x shutter speed duration increase or a 2-stop aperture increase.

Cheers,

 

[Thierry Hagenauer]

yes, or in other words, f45 at infinity is the same as f22 at 1:1 (double the distance), etc ...

At 1:5, the difference is exactly 1/3 of an f-stop, at the limit of being neglectable.

Stated aperture of any lens is only accurate when the lens is at infinity focus. As one focuses closer, focal length increases and thus aperture decreases proportionately. This effect is so small as to be insignificant for exposures at less than 1:5 reproduction ratios. However, as you focus closer, the need to compensate for the added lens extension kicks in for this very reason.

Yes, that's what I am saying since the begin.

Thus, diffraction effects will be greater for any given lens and aperture combo when used for close subjects relative to subjects at infinity...

Best regards,
Thierry

 

[Bart_van_der_Wolf]

P45 has 6.8 Micron pitch

hi Bart,

for what it is worth, I believe that the diffraction is highly dependent on the reproduction scale, respectively the distance lens-image plane.

To illustrate it I have here a little drawing: one can understand, that if the distance lens to image plane becomes longer, the broken light rays at the edge of the aperture blades do "spread" wider when they hit the image plane, thus causing more fuzziness.

Hi Thierry,

As I've tried to demonstrate, the onset of visible degradation happens at exactly the same aperture number (relative to sensel pitch) regardless of the projection magnification (5:1 macro and 1:1000 far focus). The effect of diffraction is affected by the magnification because (also) degraded image detail is magnified, which makes its effect more visible.

The proof is in the eating of the pudding. Leonardo showed a bracketing series on a P25 (if I'm not mistaken). To my eyes, the images start to degrade past f/11. The P25 has a sensel pitch of 9 micron (just like a Sinarback 54). According to my rule of thumb, 1.5 x 9 micron = 13.5 micron as onset of visible diffraction. At f/11 the diameter of the diffraction pattern is 2.44 x .555 x 11 = 14.9 micron, so slightly better quality than predicted, maybe due to the absence of an AA-filter?

If someone with a MF back could do another aperture series with a (distant) subject that also has some low contract detail in it, we can determine if that is consistent with theory. Then a series of close ups to see if there's any difference.

BTW, a P45 (6.8 micron pitch) would have almost the same sensel pitch as my 1Ds3 (6.4 micron), so would probably also degrade beyond f/7.1 (or f/8 due to no AA-filter), and a Sinarback 75 (with 7.2 micron pitch), would degrade beyond (1.5 x 7.2 =10.8 micron) or f/8 (2.44 x .555 x 8 = 10.8336) or a third stop narrower.

Let's settle this. Who's in for the challenge ...

Cheers,
Bart

 

[leonardobarreto.com]

I have done a test with a far away subject on my D300 and the Mamiya/PhaseOne. I used 50mm and 80mm respectively.

I also learned how to do the trick with layers, but may have done too big an image...

This is the one for the Nikon... I will work on the other one tomorrow (should be smaller)

 

[leonardobarreto.com]

I would choose f.4 f.8 or f 8 and 1/2 ?

 

[Bart_van_der_Wolf]

This is the one for the Nikon...

To me, f/8 looks slightly better than f/11 (look at the drapes behind the windows).
Theoretically f/6.3 would probably be optimal for a 5.5 micron sensel pitch sensor.

Bart

 

[leonardobarreto.com]

I was looking at the drapes too... and I realize that I left out the f. 5.6 layer.

I would agree that by f. 11 you left the sweet spot....

Close up or infinity is not a factor, is it?