Pages: [1]   Go Down

Author Topic: Are ApoSironar HR really diffraction limited?  (Read 5282 times)

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Are ApoSironar HR really diffraction limited?
« on: May 03, 2006, 12:36:55 pm »

There is no doubt that these new Rodenstock Sironar Digital HR lenses are very fine lenses as Michael has reported, outshining the Zeiss glass he used on his Contax. However, are they truly diffraction limited at maximum aperture?

MTF curves for these lenses are published. I will be considering only the sagittal figures, but the same considerations would apply for the meridional ones.

http://www.phootos.com/Library/rodenstock_...ses_digital.pdf

Now a diffraction limited lens at MTF of 80% should resolve 58 lp/mm with green light, 500 nm.

http://www.clarkvision.com/imagedetail/sca...tml#diffraction

However, if we look at the MTF chart for the 35 mm lens at f/5.6 for 60 lp/mm (close enough to 58 mm), we see that the MTF is 65%, not the predicted 80%. It does rise to about 70% about 13 mm off axis. Furthermore, the MTF for the same parameters 35 mm off axis drops to 40%. On stopping down to f/8 the off axis MTF for 60 lp/mm rises above 50%, and f/8 may be a better aperture for sharpness across the entire field, and I would conclude that the lens is not really diffraction limited at f/5.6.

An ideal lens would have high MTF above the Nyquist limit of the camera (73 lp/mm for the p45 back) and zero MTF above the Nyquist limit so as to avoid image degradation from alaising. The resolution of the lens above Nyquist is not merely useless, it degrades image quality and many cameras incorporate a low pass filter to get around this difficulty. Unfortunately, it is not possible to design a lens with a sharp dropoff in MTF above a certain resolution.

A comparison with the MTF of film is of interest and shows why digital has better perceived sharpness than film even though film may have higher resolution. Consider the MTF of Fujichrome Velvia 100:

http://www.fujifilm.com/products/professio...f/Velvia100.pdf

The MTF begins to drop off at 20 cycles/mm (20 lp/mm) and reaches MTF50 at about 48 cycles/mm (48 lp/mm). Above that, MTF drops off sharply.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Are ApoSironar HR really diffraction limited?
« Reply #1 on: May 03, 2006, 10:46:03 pm »

I think we have to accept, bjanes that the very term itself, 'diffraction limited', is too imprecise to enable a precise answer to your question. You have already agreed that there is a transition over a number of stops from the first visible effects of diffraction to a state where nothing but diffraction limits resolution.

Is the typical 35mm lens diffraction limited at f11 or f32? Is it meaningful to say, resolution is very slightly diffraction limited at f11, moderately diffraction limited at f16, seriously diffraction limited at f22 and totally diffraction limited at f32? I prefer to think about it this way.
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Are ApoSironar HR really diffraction limited?
« Reply #2 on: May 04, 2006, 09:29:49 am »

Quote
I think we have to accept, bjanes that the very term itself, 'diffraction limited', is too imprecise to enable a precise answer to your question. You have already agreed that there is a transition over a number of stops from the first visible effects of diffraction to a state where nothing but diffraction limits resolution.
[{POST_SNAPBACK}][/a]

Contrarian as always, Ray. With ordinary lenses there is a "sweet spot" of several stops where one balances off diffraction and aberrations, and the situation you mention above occurs. However, with a lens that is diffraction limited at full aperture, there is no sweet spot and MTF falls off with any stopping down. That was the whole point of Michael's article--with a diffraction limited lens, stopping down degrades image quality as shown in the equation below:

Dawes limit = 1/(Fw) in line pairs per mm, where F = the focal ratio, and w = the wavelength of light in mm; MTF50 ~ 0.39 * Dawes limit.

Quote
Is the typical 35mm lens diffraction limited at f11 or f32? Is it meaningful to say, resolution is very slightly diffraction limited at f11, moderately diffraction limited at f16, seriously diffraction limited at f22 and totally diffraction limited at f32? I prefer to think about it this way.
[a href=\"index.php?act=findpost&pid=64423\"][{POST_SNAPBACK}][/a]

That is a reasonable way to look at it. If you want to measure when the lens becomes diffraction lilmited, then one would determine at what aperture the lens resolves at the level predicted by the above eqation. There is little ambiguity there. However, the diffraction limits on the system (lens + sensor) must also take into account the pixel size. The D2X becomes diffraction limited at a larger aperture than the D70. If you want to take into account the entire imaging chain including the lens, sensor, size of the final print, and the viewers visual acuity, then the diffraction calculator below is a useful tool.

[a href=\"http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm]http://www.cambridgeincolour.com/tutorials...photography.htm[/url]
« Last Edit: May 04, 2006, 09:31:27 am by bjanes »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Are ApoSironar HR really diffraction limited?
« Reply #3 on: May 04, 2006, 10:12:32 pm »

Quote
Ray. With ordinary lenses there is a "sweet spot" of several stops where one balances off diffraction and aberrations, and the situation you mention above occurs. However, with a lens that is diffraction limited at full aperture, there is no sweet spot and MTF falls off with any stopping down. That was the whole point of Michael's article--with a diffraction limited lens, stopping down degrades image quality as shown in the equation below:


bjanes,
I would imagine that Michael's Rodenstock lens is diffraction limited to the same degree that a 'typical' 35mm lens might be said to be diffraction limited at f8. In other words, there's probably a partial overlap of Airy Disks rather than a full overlap at f5.6 and the next stop down (f8) begins to show increasing effects of diffraction just as the typical 35mm lens at f11 begins to show increasing effects of diffraction with no looking back.

I agree that in practice, given the resolution limits of the P45 sensor, we cannot know at what stage of transition the lens is at f5.6 (without a lot of calculation beyond my capabilities and without a set of line charts of known and varying contrast ratio). Is the Rodenstock at the (typical) 35mm f8 equivalent or the f11 equivalent? Your MTF data for diffraction limited lenses would appear to be for lenses that are fully diffraction limited and those figures are higher than the MTF response in Rodenstock's charts, so there is at least a suggestion the lens is not diffraction limited in the way Michael has defined diffraction limitation.

But surely this is only of academic interest. A working photographer needs to know at what aperture his 'system', including lens, is sharpest and what the trade-off in resolution might be when stopping down, offset against the advantages of increased DoF.

My style of shooting does not lend itself to heavy calculation or use of DoF calculators of any sort. I rely completely on guesstimates, experience and trial and error. I find that in order not to miss the shot, I sometimes have to make rapid decisions because some of my shots are from precarious positions like a fast moving boat on the Tongle Sap or from an elephant's back. (Ever tried taking a photo whilst riding on an elephant?   )

In situations where I'm on firm ground and have the time, I'm far more concerned about getting an interesting angle and an unusual or at least pleasing composition. If there's any doubt about DoF issues, I'll take a second, or third shot at a different f stop, as I do if there's doubt about ETTR.

It needs to be mentioned also that DoF calculators are useless in the field unless you have an accurate means of measuring distances.
Logged

Anon E. Mouse

  • Full Member
  • ***
  • Offline Offline
  • Posts: 197
    • http://
Are ApoSironar HR really diffraction limited?
« Reply #4 on: May 05, 2006, 12:05:11 am »

First, it is "diffraction limited" meaning that resolving power is limited by diffraction only and not by the lens quality. Diffraction limited does not just refer to the maxium aperture either. This is shown by Micheal's experiment where stopping down the aperture lowers resolving power. Diffraction limited criteria is also related to the image circle which would increase in size as the Rodenstock lenses are stopped down. But I would not worry. It would be hard to find a lens today that is not diffraction limited as it is not that difficult today to exceed that standard compared to the time the the standard was made.

Secondly, if the MTF is using white light, then the performance will not equal to what can be acheived at only 500nm. Also, are you taking transmission, abberations, and flare into account as that would impact the MTF results and deviate from the "perfect" lens, but not contradict the diffraction limited criteria. (You did notice that your definition of diffraction limited would not include chromatic abberations because it is being measured at one wavelength of light.)

I find it interesting that Micheal loves making these technical articles when he keeps commenting that the science behind imaging does not really reflect the reality. Actually, the science is excellent and produces very good imaging products. The problem is that people don't really understand the numbers.

Since the inverse of the resolving power of the componants of a system are additive and resolving power is related to target contrast, you would naturally want to design a lens to exceed the Nyquist frequency. And as you pointed out, sharpness (and that can only be a percieved quality) is not related to resolving power. I would be very careful is singling out a piece of data and applying it to a single cause. It tends to be too simplistic. Naturally, "perfect" lenses do not exist and and are just models used to teach a specific point about optical theory. They are not really a standard to which a real lens can be measured. To understand how lenses work (or don't work), you need to think systemically. Fortunately, the manufacturers do a lot of that for you and so provide good products.
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Are ApoSironar HR really diffraction limited?
« Reply #5 on: May 05, 2006, 07:20:34 am »

Quote
Secondly, if the MTF is using white light, then the performance will not equal to what can be acheived at only 500nm. Also, are you taking transmission, abberations, and flare into account as that would impact the MTF results and deviate from the "perfect" lens, but not contradict the diffraction limited criteria. (You did notice that your definition of diffraction limited would not include chromatic abberations because it is being measured at one wavelength of light.)

[{POST_SNAPBACK}][/a]

Obviously, flare would affect the MTF but transmission would not. Light absorbed by the lens elements would merely reduce the luminous flux by a proportional amount and would not affect the MTF. Since resolving power is related to wavelength, one obviously needs to take wavelength into account in this analysis. Green was used because the eye is most sensitive to this color. Since the lens is presumed to be diffraction limited, chromatic aberration and other aberrations need not be considered.

Quote
Since the inverse of the resolving power of the componants of a system are additive and resolving power is related to target contrast, you would naturally want to design a lens to exceed the Nyquist frequency.

[a href=\"index.php?act=findpost&pid=64560\"][{POST_SNAPBACK}][/a]

Actually, resolving power beyond Nyquist is not desired since it leads to alaising and degraded image quality. That is why low pass filters are used. The additive properties of the reciprocals of MTF are only approximate and are used for MTFs below 10%. Since perceived image sharpness correlates best with MTF50, one needs Fourier transforms for a more precise analysis.

[a href=\"http://www.normankoren.com/Tutorials/MTF.html]http://www.normankoren.com/Tutorials/MTF.html[/url]
Logged

michael

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5084
Are ApoSironar HR really diffraction limited?
« Reply #6 on: May 05, 2006, 08:30:22 am »

And now we know why the rest of us should stick to just taking pictures.  

Michael
Logged

Anon E. Mouse

  • Full Member
  • ***
  • Offline Offline
  • Posts: 197
    • http://
Are ApoSironar HR really diffraction limited?
« Reply #7 on: May 06, 2006, 12:51:32 am »

Quote
Actually, resolving power beyond Nyquist is not desired since it leads to alaising and degraded image quality. That is why low pass filters are used.

Is that why manufacturers make lenses for their cameras that exceed the Nyquist frequency? By using a low-pass filter with high-resolution lenses, the manufacturer can keep the final resolving power of the image close to the Nyquist frequency over the aperture range. Actually, the final resolving power should be below the Nyquist frequency as it is at the Nyquist frequency that aliasing is apparent. Your solution presents a problem - at which aperture should the resolving power of the lens equal the Nyquist frequency?

Lens transmission can impact contrast and so influence the MTF curve.

But I am afraid we will need to disagree. Like Micheal said, it is a rather academic subject and does not really have a lot to do with the problem of taking pictures. After all, the contents of the picture are far more important than the contents of an MTF plot. As far as the quality of modern optics, there is very little to worry about. They certainly exceed the standard of diffraction limited, which was your question.
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Are ApoSironar HR really diffraction limited?
« Reply #8 on: May 06, 2006, 10:10:20 am »

Quote
Is that why manufacturers make lenses for their cameras that exceed the Nyquist frequency? By using a low-pass filter with high-resolution lenses, the manufacturer can keep the final resolving power of the image close to the Nyquist frequency over the aperture range. Actually, the final resolving power should be below the Nyquist frequency as it is at the Nyquist frequency that aliasing is apparent. Your solution presents a problem - at which aperture should the resolving power of the lens equal the Nyquist frequency?

[{POST_SNAPBACK}][/a]

Most manufacturers rightly make their lenses to have the highest MTF possible at all frequencies, but with digital it is more important to have high MTF below Nyquist than high MTF above Nyquist. Also, with digital, high MTF at lower resolution is often more important than low MTF at high resolution. It's Leitz vs. Zeiss all over again.

Alaising does not occur at Nyquist as you state, but rather above Nyquist. For further explanation of these topics, refer to this paper from Schneider. Look at figure 3; which lens would you prefer? Also, your use of resolving power without specifying contrast is incomplete. Alaising with MTF of 10% or less usually is not apparent to the eye. Point and shoot digitals do not use low pass filters, because they are not needed. The lens acts as the low pass filter, since the MTF above Nyquist is low.

[a href=\"http://www.schneider-kreuznach.com/knowhow/digfoto_e.htm]http://www.schneider-kreuznach.com/knowhow/digfoto_e.htm[/url]

You misinterpreted my comment about resolution above Nyquist. It is not currently possible to design a lens with high MTF right up to Nyquist and no response beyond Nyquist and that was not my suggested solution. Rather, the solution is a low-pass filter as we both suggested.

Alaising that produces colored Moiré is particularly bothersome, but with natural subjects without repetetive patterns, alaising may not be that apparent and the false detail that it adds may even appear to increase image detail. The ill fated Kodak full frame 35 mm style camera used no low pass filter and was noted for its sharpness.

Quote
Lens transmission can impact contrast and so influence the MTF curve.

But I am afraid we will need to disagree. Like Micheal said, it is a rather academic subject and does not really have a lot to do with the problem of taking pictures. After all, the contents of the picture are far more important than the contents of an MTF plot. As far as the quality of modern optics, there is very little to worry about. They certainly exceed the standard of diffraction limited, which was your question.
[a href=\"index.php?act=findpost&pid=64639\"][{POST_SNAPBACK}][/a]

Really? If you put a neutral density filter over your lens, does MTF suffer? What is the difference if the light is absorbed by the lens elements rather than the filter?

The subject has a lot to do with taking pictures, whether or not you go into the physics. The whole point of Michael's article was that with a lens that is diffraction limited at full aperture, stoping down degrades image quality. Unfortunately, contrary to you assertion, relatively few lenses exceed the standard of diffraction wide open, especially fast zooms.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Are ApoSironar HR really diffraction limited?
« Reply #9 on: May 07, 2006, 05:09:17 am »

Quote
The whole point of Michael's article was that with a lens that is diffraction limited at full aperture, stoping down degrades image quality. Unfortunately, contrary to you assertion, relatively few lenses exceed the standard of diffraction wide open, especially fast zooms.
[a href=\"index.php?act=findpost&pid=64654\"][{POST_SNAPBACK}][/a]

bjanes,
It's still not clear to me at what aperture a 'system' should be described as being 'diffraction limited'. Is it when merging Airy disks completely obscure any other aberrations such as coma? Or is it at an aperture, smaller than which, resolution begins to fall off.

If it's the latter, then it's clear that most 35mm 'systems' are diffraction limited at f8, with a few at f5.6 and fewer still at f4 and f2.8.

If we are discussing the lens itself, then the camera is an inadequate tool to determine diffraction limitation. The former definition would apply and the lens MTF would need to be measured and compared with the MTF/resolution definitions for diffraction limitation which you have referred to. Is this not the case?
Logged
Pages: [1]   Go Up