Nikon D800/E Diffraction Limits

[jvora]

Hello :

Has anyone had the opportunity to test the D800/E for diffraction limits - At which f-stop does it begin to show in the images ?


Thanks,


Jai

ps : Question cross posted

[Bart_van_der_Wolf]

Has anyone had the opportunity to test the D800/E for diffraction limits - At which f-stop does it begin to show in the images ?

Hi,

Limits are where you draw the line, but theoretically it should start to become visible (at the pixel level) at f/5.6 and gets more visible as you select narrower aperture. At f/5.6 the diameter of the diffraction pattern for green light will be 1.5x the sensel pitch. Therefore, each sensel will begin to record some of the light from multiple neighbor sensels, leading to reduced contrast.

Diffraction blur can to a certain extent be recovered from with good quality deconvolution sharpening.

Cheers,
Bart 

[jvora]

Hello Bart :

Thanks for the reply -

You bring up an interesting point of employing deconvolution sharpening - Could this mean that in the in the future perhaps Photoshop or other imaging editors may include deconvolution algorithms that automatically correct the effects of diffraction based on lenses and f-stop used - Something similar to the auto lens distortion correction we have in Adobe Camera Raw ?

Thanks,

Jai

[theguywitha645d]

At the level of the image being viewed at a normal viewing distance, you should be able to stop down to f/16 with very little impact. f/22 will be passable, but it will start to show.

[Bart_van_der_Wolf]

You bring up an interesting point of employing deconvolution sharpening - Could this mean that in the in the future perhaps Photoshop or other imaging editors may include deconvolution algorithms that automatically correct the effects of diffraction based on lenses and f-stop used - Something similar to the auto lens distortion correction we have in Adobe Camera Raw ?

Hi Jai,

We can only hope! Technically it doesn't seem too far fetched to me, but then I'm impatiently awaiting the established industry to stop dragging their feet (even more than perhaps commercially justified). Maybe the time is right for other powers to claim their ground (if only my budget would allow to ...).

Cheers,
Bart

[ErikKaffehr]

Hi,

I agree with Bart that diffraction sets in at f/5.6 and that diffraction can in part be compensated with adequate sharpening.

I also agree with the guy with the 645D that f/16 can be used without problems on prints viewed at normal distance.

So I would say that optimum aperture is normally around f/5.6 - f/8. This keeps aberrations down, above f/5.6 some deterioration of image quality occurs but is not very visible in print and can to part compensated with sharpening.

This article shows the effects of diffraction in the left column: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=1

And it also shows the effects of deconvolution sharpening, here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=2

It is interesting that the arrival of the D800/D800E triggered so much discussion. Diffraction has been with us since the black and white era, and it is the reason that f/8 was established as the norm for lens testing. Almost any decent lens is diffraction limited at f/8.

Best regards
Erik

Hello Bart :

Thanks for the reply -

You bring up an interesting point of employing deconvolution sharpening - Could this mean that in the in the future perhaps Photoshop or other imaging editors may include deconvolution algorithms that automatically correct the effects of diffraction based on lenses and f-stop used - Something similar to the auto lens distortion correction we have in Adobe Camera Raw ?

Thanks,

Jai

[shadowblade]

I would disagree with the f/5.6 figure - rather, the true figure is slightly short of f/8.

At f/5.6, the Airy disc diameter is greater than the width of a single photosite - but individual photosites do not give us pixels.

Rather, it's only at just below f/8 (around f/7.8 or so) that the Airy disc exceeds the size of a single Bayer cell, and starts to have some sort of impact on sharpness.

[ErikKaffehr]

Hi,

My impression is that my test shots are visually better at f/5.6 than at f/8 on my 4.77 micron sensor.

Best regards
Erik

[BJL]

Theory says that you will get at least some tiny degree of image degradation from diffraction at any f-stop, rather than a precise cut-off. One way to see this is that even with a very small Airy disk, some light the "should" fall near the edge of one photosite instead gets spread to a neighboring one. Another way to see it is that the MTFs from various factors, including diffraction, combine by multiplication. So as soon as there is any diffraction, it contributes a factor smaller than one to the total MTF.

The bottom line is that the relevant question is more subjective, like
"at what f-stop is the effect of diffraction significant"?
What I have typically seen in reports is some slight effect visible from f-stop equal to pixel pitch in microns, but most photographers not finding the effect significant until one or two stops beyond that, and even three stops beyond still not a disaster (especially with the option of deconvolution sharpening) So for the D800, "one or two stops beyond pixel pitch" means f/8 to f/11.

[theguywitha645d]

Without diffraction, photography would not be very interesting. It is because we have diffraction we have images.

It is funny, back in the silver age of photography, no one seemed too concerned about diffraction. Certainly, no one changed their aperture habit based on the resolving power of the film--diffraction was just as noticeable on a piece of film at about the same apertures as digital. It is only in the digital age where we magnify images to a much greater degree that we suddenly care. I regularly print of 44" wide printers. DoF has a much more important impact on the image than diffraction. An image with too much or not enough DoF will never be as interesting as an image with the right DoF regardless of lens aberrations or diffraction.

There is something quite not exciting about photography when everyone shoots at f/8. This is a creative pursuit, not an object lesson in resolving power.

[Bart_van_der_Wolf]

I would disagree with the f/5.6 figure - rather, the true figure is slightly short of f/8.

At f/5.6, the Airy disc diameter is greater than the width of a single photosite - but individual photosites do not give us pixels.

Rather, it's only at just below f/8 (around f/7.8 or so) that the Airy disc exceeds the size of a single Bayer cell, and starts to have some sort of impact on sharpness.

Hi,

I'm afraid there are a few problems with your assumptions.

• 1. At f/5.6, and a wavelength of 555 nm, the diameter of the first diffraction ring is 7.58 micron, which is 1.554x the sensel pitch of the D800/D800E. Therefore, at least 4 neighboring sensels will be affected, and another 4 are just starting to be hit. This first ring represents 83.9% of the diffraction, so there is already an additional 16.1% of energy going to surrounding sensels. This can cause interference with the Airy disk patterns of other sensels, so the signals can become substantially higher in the neigboring sensels at certain spatial frequencies.
• 2. The D800 has an  OLPF (AA-filter) which spread the signal for a single sensel to about 4 sensels.
• 3. Individual photosites DO give us pixels, with 2 of the 3 channels being interpolated from surrounding photosites.
• 4. You speak of a "Bayer cell", which raises the suspicion that you erroneously think that 4 photosites make 1 pixel, they don't, they make 4 pixels, all suffering from diffraction from all of their neighbors.

Cheers,
Bart

[bjanes]

Hi,

I'm afraid there are a few problems with your assumptions.

• 1. At f/5.6, and a wavelength of 555 nm, the diameter of the first diffraction ring is 7.58 micron, which is 1.554x the sensel pitch of the D800/D800E. Therefore, at least 4 neighboring sensels will be affected, and another 4 are just starting to be hit. This first ring represents 83.9% of the diffraction, so there is already an additional 16.1% of energy going to surrounding sensels. This can cause interference with the Airy disk patterns of other sensels, so the signals can become substantially higher in the neigboring sensels at certain spatial frequencies.
• 2. The D800 has an  OLPF (AA-filter) which spread the signal for a single sensel to about 4 sensels.
• 3. Individual photosites DO give us pixels, with 2 of the 3 channels being interpolated from surrounding photosites.
• 4. You speak of a "Bayer cell", which raises the suspicion that you erroneously think that 4 photosites make 1 pixel, they don't, they make 4 pixels, all suffering from diffraction from all of their neighbors.

Bart,

Marianne Oelund (a technically savvy engineer/photographer) make an interesting comment dpreview concerning the differences in how diffraction affects the D800 and D800E. I presume this difference is due to the lack of a low pass filter. When the Airy disc is already smaller than the pixel, making it even smaller by increasing the aperture has diminishing returns.

In the same thread, Bobn2 posted a graph showing how the f-number at which diffraction starts to become visible has nothing at all to do with the pixel size, which is somewhat at variance with Marianne's post.

Your insights would be appreciated.

Regards,

Bill

[GuzziRob]

Without diffraction, photography would not be very interesting. It is because we have diffraction we have images.

It is funny, back in the silver age of photography, no one seemed too concerned about diffraction. Certainly, no one changed their aperture habit based on the resolving power of the film--diffraction was just as noticeable on a piece of film at about the same apertures as digital. It is only in the digital age where we magnify images to a much greater degree that we suddenly care. I regularly print of 44" wide printers. DoF has a much more important impact on the image than diffraction. An image with too much or not enough DoF will never be as interesting as an image with the right DoF regardless of lens aberrations or diffraction.

There is something quite not exciting about photography when everyone shoots at f/8. This is a creative pursuit, not an object lesson in resolving power.

Absolutely - it is a rare voice of sense that is heard on the internet!

Does the pixel level image really matter at the end of the day?  It is all about the final print surely.

[Bart_van_der_Wolf]

Absolutely - it is a rare voice of sense that is heard on the internet!

Does the pixel level image really matter at the end of the day?  It is all about the final print surely.

Hi,

Does you mean that people buy all these megapixels, only to downsample for print? Surely there must be some who use these pixels to produce proportionally larger output than they could, and thus the per-pixel diffraction matters.

Of course the diffraction pattern doesn't get bigger or smaller when we use a different sensor but the denser sampling of the diffraction pattern does mean that the pixel contrast is reduced.

Cheers,
Bart

[Bart_van_der_Wolf]

Marianne Oelund (a technically savvy engineer/photographer) make an interesting comment dpreview concerning the differences in how diffraction affects the D800 and D800E. I presume this difference is due to the lack of a low pass filter. When the Airy disc is already smaller than the pixel, making it even smaller by increasing the aperture has diminishing returns.

Hi Bill,

Assuming a perfect aberration free lens, diffraction will increase proportionally with the F-number as the aperture gets narrower. However, wide open the diffraction pattern is very small. In fact it is smaller than a single sensel. Therefore it cannot be resolved and has no meaningfull impact on resolution, because the sampling aperture of the sensel is the limiting factor.

When the diffraction pattern gets large enough by using narrower apertures, it's effect on the per-pixel contrast will become visible. The micro-detail gets lost because of the reduced contrast between pixels (they are contaminated with optical signals related to its neighboring pixels).

When we compare to a different sensor with smaller sensels, the per-pixel contrast will be affected earlier when we use ever smaller apertures on the smaller sensel pitch version, however, we do get a more accurate (over-)sampling of the diffraction pattern. So, we lose contrast and detail per pixel (we can produce larger output at the same PPI, but lose detail), but for same size output we gain some resolution because we can use a higher PPI (and because the diffraction pattern is oversampled, we may be able to restore even more resolution by deconvolution sharpening).

So the whole debate is between larger size output (with lower per capture pixel contrast), and same size output (with higher per output size pixel contrast), which isn't much of a debate since they are different things.

Marianne Oelund is correct, when the diffraction pattern gets smaller than a single sensel (by using a wider aperture, and in the absence of residual lens aberrations) then there is little resolution gain to be expected.

In the same thread, Bobn2 posted a graph showing how the f-number at which diffraction starts to become visible has nothing at all to do with the pixel size, which is somewhat at variance with Marianne's post.

I would have to read about the source data for that graph, before I can interpret what it tells. Maybe it becomes clear when I've had the time to read that whole thread.

Cheers,
Bart

[BJL]

In the same thread, Bobn2 posted a graph showing how the f-number at which diffraction starts to become visible has nothing at all to do with the pixel size ...

What that graph for the Nikon 50/1.8G reminds us is that resolution is at least a three way contest between diffraction, aberrations, and sensor resolution, and that when the aperture is big enough for diffraction to stop dominating over sensor resolution (about f/8 to f/5.6), lens abberations are also a major factor. The peak of resolution does seem to shift slightly, from f/5.6 with the D3 to f/4 with the D3x, but the main effect of increased sensor resolution is to raise the resolution curve at all apertures, just raising it a bit more at f/4 than at f/5.6.

What we need is a 3D plot with both aperture and pixel size axes!

So maybe we should shift from worrying about image sharpess being limited by diffraction (unavoidable optical physics) at small apertures to it being limited by lens aberrations (and OOF effects, some more fundamental optics) at large apertures. Resolution obsessives are stuck between a rock and a hard place!

[whats-his-name]

To see if the D800 can produce large prints in accordance to the sensor size one has to do real world tests with a sensor of similar size but larger photosite pitch, i.e., a MF back, to see if diffraction affects resolution at lower apertures in the Nikon.

I would suggest adapting one of the Schneider Apo-digitars or Rodenstock (or similar high resolution lens) to fit on both cameras to keep the playing field level. Process the image equally to allow only the difference in diffraction to show, print at large size, and compare.

I'm guessing there won't be much difference because ultimately other factors have a bearing on how large you can print (the printer itself being one). If people expect the D800 to have twice the resolution of the D4 then they'll be disappointed. It will have higher resolution but the law of diminishing returns (caused by those 'other factors') will step in to lower the increase in resolution.

[arlon]

I don't know or care about the physics of the situation. I shoot something then look at the picture and see if it's useable. I've been shooting a lot of macro with the D800E at f11-32 (using several vintage manual focus lenses, shot hand held in a windy West Texas) and haven't had anything yet that wasn't better than I could have gotten from my D700. F32 just turns a D800E into a D800... Most of the macro I'm shooting are tiny bugs and flowers and end up being used as 100% crops. Even at 100% and F22, the pictures are sharper than anything I've gotten with any other camera I own.

I know macro isn't landscape but it should still have some merit. Also I'd assume the imag quality is going to vary from lens to lens, especially at infinfity and stopped all the way down..

Here is an F11 version that's a 100% crop. Click the image for the full size version




This is one shot at f22..

[EricV]

Of course the diffraction pattern doesn't get bigger or smaller when we use a different sensor but the denser sampling of the diffraction pattern does mean that the pixel contrast is reduced.

And that is a good thing, a definite advantage for the denser sensor.

[Bart_van_der_Wolf]

I don't know or care about the physics of the situation.

That's unfortunate, because it would allow you make a better decision when you can choose between options. Narrower apertures will not only show more sensor dust but also produce more background clutter in Macros, and microcontrast suffers, so if you can avoid it it would improve your overall image quality (you also can use shorter shutterspeeds which helps to reduce handheld motion blur).

Most of the macro I'm shooting are tiny bugs and flowers and end up being used as 100% crops. Even at 100% and F22, the pictures are sharper than anything I've gotten with any other camera I own.

That is unfortunately not plausible. At 100% the smaller sensels will have lower contrast than larger sensel cameras at the same aperture, or even lose all resolution at such apertures, and micro contrast detail will be lost whether we care or not. To illustrate I'll show a Macro crop from a larger sensel camera (sorry, I don't have a D800 to demonstrate it with) which demonstrates at f/8.0 what will occur at f/5.6 and narrower on the D800:

The 'Dsd' mentioned is the diffraction spot diameter in sensel widths.

Here is an F11 version that's a 100% crop.

As the theory predicts, until you reach f/16, there is still a lot that can be salvaged.

This is one shot at f22..

It's a pitty you can't compare it to an f/16 version, it would have been much sharper, but then maybe you don't care ... It's just too bad that people spend such an amount of money and then throw quality away by stopping down too far. Beyond f/16 there is virtually no fine detail left even in the plane of focus. At f/22 you only get, at best, some 75% of the maximum resolution that the camera is capable of as if shooting with a camera with 56% of the mega pixels, which you could put to good use. And as I said, you'll reduce your risk of motion blur if you shoot a f/16 instead.

Cheers,
Bart