[RioRico]

Yo, I'm not sure if this is dSLR or GenPhoto related, so I'll try the former.

I was recently pointed to Digital Camera Sensor Sizes: How it Influences Your Photography which deals with effects of sensor/film sizes, including diffraction limits (with link to much more info on diffraction). I'd previously misunderstood that diffraction was solely a function of actual aperture width, but that's not quite so - it's based on sensor/film size AND aperture size. [Briefly: when the diffraction 'airy disc' is larger than the DOF 'circle of confusion', the diffraction limit has been reached; at tighter apertures, the image only gets fuzzier.]

I ran the specs on some of my older Sony PNS's thru that site's Diffraction Limit Calculator and was gratified to find that my favorites, the 1.1mp DSC-P20 and 5mp DSC-V1 (both of which also have pretty low sensor pixel density), never reach a diffraction limit, which helps explain (besides the Zeiss optics on the V1) why their images are always crystalline - anything in focus is at maximum clarity. With their small sensors, the limits are fairly low - f/5.5 and f/4.7 respectively - but their tightest apertures are f/4.5 and f/4.0, well under those limits. Huzzah! And low pixel density means low noise (generally).

Next, the quandry. I run the beautiful K20D (with even lower pixel density) thru the same calculator - and find that for an APS 14.6mp sensor, the diffraction limit is just f/9.3 - what?!?!? (And the more megapickles in the same-size sensor, the lower the limit.) So stopping down ANY lens on the K20D to f/11 or beyond means loss of clarity?!?!? I quickly searched PentaxForums for discussions of diffraction and found some comment that "sharpening increases the limit" or whatever. But no details beyond that, that I could find.

Now, the questions. Does anyone here REALLY know the diffraction limit of the K20D at maximum resolution? How is 'sharpening' involved? As noted, that limit increases with sensor density - a 5mp APS sensor's limit is f/16 - so an older cam should give maximum clarity over a much wider DOF. Can I increase the K20D's limit by reducing its resolution to 5-10mp? When shooting macro (1:1 and beyond) must I limit the aperture (and thus suffer razor-thin DOF) for maximum clarity? Must I 'wait at 8' (never stop below f/8) unless I deliberately want fuzziness?

[Wheatfield]

I think this kind of stuff gets over analyzed by people with too much time on their hands and not enough know how to run an actual test. I shoot at very small apertures quite often and am not having issues with "pictures getting fuzzy".
My own take is that this is a non issue at best, and at worst is a complete fallacy based on someone making an incorrect assumption.
The beauty of the internet is that you can find whatever information you need, the bad part of it is people who don't know what they are talking about now have a way of pushing their pseudo-scientific crap out to the world at large with very few checks on whether there is anything factual in their blather.

Go out and take some pictures and check for yourself.

[Sean Nelson]

I'm no expert on diffraction, but the very first thing that jumps out at me is that you're talking about apertures with no mention of focal lengths. Diffraction is goverened by absolute aperture, so in terms of f/numbers the diffraction limit on, say, a 35mm lens is going to be different than one on a 50mm lens.

So the first question is: when you say the limit is f/9.3, what focal length are you referring to?

[RioRico]

All the references to diffraction I've found say that lens focal length is NOT a factor, that the diffraction limit is established ONLY by the sensor size and density. (The lens LPM resolution isn't a factor either, nor CA - those are other issues.)

They also say that diffraction effects are generally minor EXCEPT when using a tripod, and enlarging the image greatly. I'm somewhat concerned about stopped-down long teles and ultrawides, and my Kenko fisheye adapter which likes to be at f/44 or worse, and tight macro work, and any substantial crops and enlargements of tripodded photos. Some of this is pixel-peeping, but when I work at that level, I want maximum clarity. It's a technical issue.

I know that other members here, like falconeye and Ben Edict, have written knowledgibly on diffraction. I'm sorta appealing to them for clarification. Help!

[*isteve]

Yo, I'm not sure if this is dSLR or GenPhoto related, so I'll try the former.

I was recently pointed to Digital Camera Sensor Sizes: How it Influences Your Photography which deals with effects of sensor/film sizes, including diffraction limits (with link to much more info on diffraction). I'd previously misunderstood that diffraction was solely a function of actual aperture width, but that's not quite so - it's based on sensor/film size AND aperture size. [Briefly: when the diffraction 'airy disc' is larger than the DOF 'circle of confusion', the diffraction limit has been reached; at tighter apertures, the image only gets fuzzier.]

I ran the specs on some of my older Sony PNS's thru that site's Diffraction Limit Calculator and was gratified to find that my favorites, the 1.1mp DSC-P20 and 5mp DSC-V1 (both of which also have pretty low sensor pixel density), never reach a diffraction limit, which helps explain (besides the Zeiss optics on the V1) why their images are always crystalline - anything in focus is at maximum clarity. With their small sensors, the limits are fairly low - f/5.5 and f/4.7 respectively - but their tightest apertures are f/4.5 and f/4.0, well under those limits. Huzzah! And low pixel density means low noise (generally).

Next, the quandry. I run the beautiful K20D (with even lower pixel density) thru the same calculator - and find that for an APS 14.6mp sensor, the diffraction limit is just f/9.3 - what?!?!? (And the more megapickles in the same-size sensor, the lower the limit.) So stopping down ANY lens on the K20D to f/11 or beyond means loss of clarity?!?!? I quickly searched PentaxForums for discussions of diffraction and found some comment that "sharpening increases the limit" or whatever. But no details beyond that, that I could find.

Now, the questions. Does anyone here REALLY know the diffraction limit of the K20D at maximum resolution? How is 'sharpening' involved? As noted, that limit increases with sensor density - a 5mp APS sensor's limit is f/16 - so an older cam should give maximum clarity over a much wider DOF. Can I increase the K20D's limit by reducing its resolution to 5-10mp? When shooting macro (1:1 and beyond) must I limit the aperture (and thus suffer razor-thin DOF) for maximum clarity? Must I 'wait at 8' (never stop below f/8) unless I deliberately want fuzziness?

Diffraction occurs all the time in aperture limited optics. It will be degrading sharpness at all apertures, but the effect will increase as the aperture gets smaller.

However, at wide apertures, spherical aberration limits resolution far more than diffraction. The effect of this is reduced as you stop down.

Hence somewhere in the middle the lens will achieve its maximum sharpness, but the fall off from this point is quite gradual.

Even though sharpness falloff will happen earlier with more densely packed sensors, it starts from a higher peak, so a 12 MP camera at F11 may still be sharper than a 6MP camera at F11 though the differences will get less and less as you stop down.

Typically fast lenses are sharpest about 2 stops below max aperture, although corner sharpness may still improve for another stop for wide angle or very fast lenses where spherical aberation is more of a problem. After that, diffraction will start to reduce resolution slightly (even at F8) but its a gradual process. You can stop down to F11 or even F16 without seeing a major loss of resolution, and indeed if you shoot macro or landscape you may have to sometimes.

See here....
Pentax SMC-FA 35mm f/2 AL - Review / Test Report

[Wheatfield]

All the references to diffraction I've found say that lens focal length is NOT a factor, that the diffraction limit is established ONLY by the sensor size and density. (The lens LPM resolution isn't a factor either, nor CA - those are other issues.)

They also say that diffraction effects are generally minor EXCEPT when using a tripod, and enlarging the image greatly. I'm somewhat concerned about stopped-down long teles and ultrawides, and my Kenko fisheye adapter which likes to be at f/44 or worse, and tight macro work, and any substantial crops and enlargements of tripodded photos. Some of this is pixel-peeping, but when I work at that level, I want maximum clarity. It's a technical issue.

I know that other members here, like falconeye and Ben Edict, have written knowledgibly on diffraction. I'm sorta appealing to them for clarification. Help!

The best help you can give yourself is to go out and check for yourself. All the charts and graphs and pseudo-scientific verbiage isn't worth a whit compared to what you see coming off your camera.
If you really want to know what your camera is 'diffraction limited" to, then go out, put your camera on a tripod and take some pictures at medium apertures and then repeat the process at prograsively smaller apertures and then LOOK AT THE PICTURES.
If you see a loss of sharpness at smaller apertures, you've just learned something about your optical system regarding how far you can stop down.
If you don't see a loss of sharpness, you've still learned something about your optical system, and more importantly, you've learned that sometimes theoretical crap is just that.

At the end of the day, a chart on a computer screen is not a picture, and more often than not, has no relevance to a picture.

[ftpaddict]

All the references to diffraction I've found say that lens focal length is NOT a factor, that the diffraction limit is established ONLY by the sensor size and density. (The lens LPM resolution isn't a factor either, nor CA - those are other issues.)

Different "formulas" are the basis of a shaky theory. I know that my pictures get really fuzzy at f/22. Everything above that is fine. My DSLR uses a 6 million pixel sensor.

[philbaum]

Article Title: Diffraction: Resolution taxed to its limits by Lloyd L. Chambers

This magazine had an excellent article on Diffraction effects on DSLRs. Same conclusion you mentioned that it's dependent on the aperture and the size of the sensor photosites. Once the Airy's disc overlaps more than half way onto the adjacent site, than contrast first suffers, followed by resolution. (the only bad thing about this article is that one is unable to read it online without buying a reprint of the article)

Some of your replies indicated to go out and take pics. Well this article did that, they compared the 21 megapixel Canon EOS 1D Mark III with the 12 megapixel Nikon D3 at apertures f2.8 through f45 using an outstanding Nikon 85mm f2.8D PC-Micro Nikkor lens on both cameras. "The two cameras are esentially equal in resolving power by f/32"

Other quotes from this magazine article:

"When dept of field is a priority, "tilt lenses should be used in order to evade the diffration/depth of field conflict. To paraphrase an old maxim: f/8 and stop there. That simple rule will maintain optimal or near-optimal lens performance and image contrast resolution with today's DSLRs while offering reasonable depth of field for many subjects. Stopping down to f/11 or f/16 is warranted with some subjects, but the contrast compromise should be kept in mind."

There is way too much information in this article to put into this post.

As a last comment, i was out in the Skagit County tulip fields a coupla days ago. I brought all my lenses along, including 3 DA's, 21mm, FA50 1.4, 35mm Macro, 300mm, as well as a M 135mm f3.5(which did quite well i would mention). The lens that produced the most remarkable shots was my newest lens, the DA 35mm ltd Macro lens. It produced images that had lovely creamy colors with that 3-d effect, consistently. When i looked back on one remarkable tulip image from this lens, i shot it at a f21 aperture and noone would ever accuse this image with a lack of contrast, just the opposite. So i don't know how to reconcile that experience with the article i just quoted from, except that the magazine did mention that one way to keep diffraction partially at bay was better lenses.

I remember talking to a photographic professional who was selling large images at the local craft fairs last summer. He was shooting a Pentax 6 by 7cm film camera that he had brought with him. Told me that it had that remarkable 3-d effect which he attributed to film layers. I'm now wondering if the most inexpensive way of producing large diffraction free images is to get a medium format film camera (photosites - whats that?)

[Marc Sabatella]

You talk about reaching a diffraction limit as if it were some of brick wall beyond which pictures cannot be taken. It's not like that all. It's just part of the normal variation we expect to see in the performance of any lens - a bit soft wide open, gradually getting sharper as you stop down, until some point (where f/9.3 or elsewhere) at which point it starts getting a bit softer again. So what?

[Ben_Edict]

BE WARNED: The following post is long, but I have not found another way to explain this diffraction/resolution issue...

We had a very long (and sometimes a bit heated) discussion not so long ago about diffraction, resolution and diffraction limits. Though Wheatfield's opinion, that the truth only lies in really doing it, with which I symphathize, there is some easy to understand theory, which is anything but new.

The smallest detail an optical system can image is defined by the Airy disc, a small disc with a central (light) intensity peak surrounded by alternating dark and brighter rings. (This is true for round photographic lenses). The intensity in the outer rings decreases rapidly and that's the reason, why in applied optics the size of the Airy disc is usually given as the radius r of the first ring.

Now comes the easy equation: r = 1.22 Lambda N (Lambda is the wavelength and N the relative aperture, calculated as the focal length divided through effective lens diameter).
What you see immediately, thet the radius of the Airy disc decreases, when the wavelength gets shorter and/or the aperture N decreases.

A smaller Airy disc means, that details will be imaged smaller and thus more details can be imaged on a given film/sensor area = aka the resolution increases. This is the case, because resolution is usually (not always!) defined by the Rayleight criterion, which very simply states, that two different points can be resolved, if the central intensity peaks of the Airy discs (which form the points) are just as far apart as the first dark gap in the Airy disc. That is: Resolving power = 1/r and given in mm*-1

That's the theory. Then comes realisty: Most lenses will have different abberations, which are worst, when the lens is fully open = the N is smallest. These aberrations reduce the resolution, as they will "disfigure" the Airy disc. Sometimes the Airy disc will loose its shape completely and you get a smeared out bright area, instead of the clearly defined central brightness peak with rings. Sometimes the pouter rings are brighter or a t least as bright, as the first inner ring etc. You get the idea. This all leads to reduced resolution, as the first dark gab is not defined. Two or even more Airy discs smear together to form a "blob".

Most lenses will get better, when stopped down 1, 2 or 2.5 f-stopps (that's very different). So, when abberations get eliminated the aperture N is already getting bigger, thus increasing the radiu r of the Airy disc and reducing resolving power.

These are simple facts.

Now, the question, in how far the sensor resolution and the pixel density and the size of each photo site influences resolution is a completely different thing. It has, in principal, nothing to do with the Airy disc. BUT it is clear, that a lens, that has not enough resolving power, to make use of the sensor resolution is a poor tool. On the other hand a lens, which delivers much more resolution, than the sensor has, is overkill (because one pays for a kind of quality, that is not useable).

The hard limit of sensor resolution is given by the pixel count. If a Sensor has 4000 pixels distributed over 24mm, each pixel would have a max. size 6µm. In fact it is a bit smaller, as there are small structures between the rows and columns of photo sites. So, the sensor cannot resolve a single point smaller than 6 µm, as it can hardly resolve two points, that are imaged on to the same single photo site.

The funny thing about these digital sensors with their strict rectangular matrix is, that they can indeed resolve single neighbouring points, even when there is no dark pixel in between, as long as the point don't have the exact same brightness.

The 6…m size for a single pixel on the sensor is an average for APS-C cameras. So, now we can do some rough calculations and see, what lens would meet this resolution limit or falls short; we use a medium visible wavelength of 550nm = 0.00055mm (or 0.55 µm)
We simply use our Airy disc based diffraction limit r = 1.22 * 0.00055 * N

let's take a lens of f/1.4: r = 0.00094mm = 0.94 µm (easily within the size of the 6 µm sensor pixel)

for f/2.8 we get: r = 0.0019mm = 1.9µm (easily withion the size of the sensor pixel)

for f/5.6 we get: r = 0.0038mm = 3.8µm ...

for f/11 we get: r = 0.0075mm = 7.5µm - AND we are already exceeding the size of the sensor pixel and not using the potential resolution.

Let's see whether f/8 is better: r = 0.0054mm = 5.4µm - that's near optimal - because we still make use of the sensor's full potential and at the same time meet or around the "sweet spot" of many lenses, at which abberations are corrected as good as possible.

---

So, is the conclusion, that apertures beyond f/11 are unuseable? In my opinion not. Or perhaps only for pixel peepers. There is a clear loss of resolution beyond that and I would call it the optimum aperture for 6µm sensors (provided the lens delivers according to theory!)

But If you look at the final image (and not the single pixels at 100% on screen) I doubt you will notice any quality loss at f/11 or f/16 unless you are photographing strong graphic structures (for what purpose ever). This last sentence is also, why I deeply mistrust lens tests made with rectangular black and white patterns (the USAF test chart) or Siemens stars etc: these simply do not represent 99% of real world images and their outcome limits the quality evaluation of lenses in a (for me) too simplistic way. A good lens is constuted by more, than sheer resolving power.

Ben

P.S.: anybody intersted in the lens side of the theory will find it in most any physics text book or in a more enjoyable format in "Applied Photographic Optics" by Sidney Ray.

[Sean Nelson]

All the references to diffraction I've found say that lens focal length is NOT a factor, that the diffraction limit is established ONLY by the sensor size and density.

I did a little more research and stand corrected. My understanding was that the airy disc size depends on the absolute aperture, but that refers not to the SIZE of the disk but to the ANGLE of the disc. What this means is that a longer focal length projects a larger airy disc than a shorter one for a given absolute aperture, and using a relative f/number adjusts for it.

So you're quite correct, it's only f/number that's important, at least in determining the effects of diffraction in the image that falls on the sensor.

[Digitalis]

additionally I'll point out that different wavelengths of light suffer from diffraction at different rates than others, I think it is safe to assume that the green channel is the least likely to be affected,considering the fact that most lens manufacturers optimise performance at that wavelength. (The human eye's sensitivity peaks in the green part of spectrum) so for instance if you took a photo at f/16 on a K20D you could single out the green channel and sharpen that channel.

but to be honest, that would only be of use if you are printing a very large print,say, 24X36" and over. diffraction's effects are so hard to detect that your average person would not notice the effect of diffraction unless you were significantly past the limit.

I'll include my personal table of lens resolution limits, and you will most likely find conflicting statements in regards to them but I have come to these conclusions through my own personal testing and lens evaluations and they have been confirmed through the purchase and my personal use of the 180mm f/5.6 Rodenstock APO Sironar HR - which is a very highly regarded diffraction limited lens for 4X5 view cameras.

these numbers correspont to the maximum theoretical limit of lens resolution at each apeture.

f/number: Lines per millimeter

1: 1400

2: 700

4: 350

8: 175

16: 88

32: 44

64: 22

[Wheatfield]

You talk about reching a diffraction limit as if it were some or brick wall beyond which pictures cannot be taken. It's not like that all. It's just part of the normal variation we expect to see in the performance of any lens - a bit soft wide open, gradually getting sharper as you stop down, until some point (where f/9.3 or elsewhere) at which point it starts getting a bit softer again. So what?

A friend of mine who hangs out on some Canon forum or another recently told me of a thread on the forum he is on where a guy pretty much trashed Canon for putting apertures on their lenses that were unusable because of the diffraction limit wall.
Such is the group intelligence of the internet. People who publish irresponsibly cannot be held to account, since the internet allows for so much anonimity and very little accountability.
If they use big words and have a smart layout, the credibility is instant, even if the message is crap.

[Ben_Edict]

additionally I'll point out that different wavelengths of light suffer from diffraction at different rates than others, I think it is safe to assume that the green channel is the least likely to be affected,considering the fact that most lens manufacturers optimise performance at that wavelength. (The human eye's sensitivity peaks in the green part of spectrum) so for instance if you took a photo at f/16 on a K20D you could single out the green channel and sharpen that channel.
...
I'll include my personal table of lens resolution limits, and you will most likely find conflicting statements in regards to them but I have come to these conclusions through my own personal testing and lens evaluations and they have been confirmed through the purchase and my personal use of the 180mm f/5.6 Rodenstock APO Sironar HR - which is a very highly regarded diffraction limited lens for 4X5 view cameras.

these numbers correspont to the maximum theoretical limit of lens resolution at each apeture.

f/number: Lines per millimeter

1: 1400

2: 700

4: 350

8: 175

16: 88

32: 44

64: 22

Your list is quite interesting, haven't calculated that, but should be easy from the numbers I have given in my previous post.

Nevertheless: lenses are NOT optimized to give least diffraction at green wavelengthes. That would be great, but is not possible, as diffraction simply decreases (well, the size of the Airy disc decreases) with frequency, aka: the shorter the wavelength (green - bule- violet - UV) the higher the resolution gets. Simple physical law and no space for optimization.

Another issue is, how the overall performance of a lens is optimized!

Ben

[jeffkrol]

YI'd previously misunderstood that diffraction was solely a function of actual aperture width, but that's not quite so - it's based on sensor/film size AND aperture size.



Now, the questions. Does anyone here REALLY know the diffraction limit of the K20D at maximum resolution? How is 'sharpening' involved? ?

The diffraction"black arts" are widely discussed on the open forum in DP review...
Main consensus seems to be don't worry about it for now ..
Diffraction and sharpening: Open Talk Forum: Digital Photography Review
I explain that the pixel spacing to recover all the available information with diffraction after using sharpening is half the usually cited value. When using sharpening the table below shows the diffraction limited pixel pitch and the resulting full frame mega-pixels based on the f-stop used. Note thart in the real world other sources of blurring like the anti-aliasing filter and lens aberrations exist that would suggest that fewer megapixels are useable; but with Bayer sensors only half the sensor mega-pixels (the green ones) yield the luminance information anyway so a Bayer sensor with the full number of listed pixels would still be reasonable.
fstop microns FF-Mpixels
4.00 : : 1.3 : : 517
5.66 : : 1.8 : : 258
8.00 : : 2.6 : : 129
11.31 : : 3.7 : : 65
16.00 : : 5.2 : : 32
22.63 : : 7.3 : : 16
32.00 : : 10.3 : : 8

The bottom line is that even with a 16 Megapixel full frame sensor f/22 is not past the sensor diffraction limit with sharpening.


Over 500 Megapixels: Open Talk Forum: Digital Photography Review
As others have pointed out the system response is the product of the responses of each component. There is still a point called the diffraction cutoff frequency where the lens does set an absolute limit to the system resolving power but it is at a much finer spacing than typically cited "diffraction limits". When we consider that we can sharpen the lens output to recover information where the MTF is less than 50% we find that we can get useful information with about half the pixel spacing (which is four times the pixels) as the usually cited limit. Here is a post where I calculate this limit:

sharpening decreases effects of diffraction (well not really decreases but your formulas can use much lower contrast constant) since you can use a lower MTF then 50.. even down to 10