Pages: [1]   Bottom of Page
nma
Full Member

Offline

Posts: 162

 « on: January 19, 2007, 08:39:14 AM » Reply

If I have understood Prof. Johnson's article, he is suggesting f21 as the aperture for maximum DOF with FF sensors while maintaining adequate suppression od diffraction effects.  My best recollection is that this is the highest f-number that I have seen recommended. Other writers are suggesting values betweem f11-f16 for FF.  How do we evaluate his assertion? Put another way, I am seeking max DOF with sufficient IQ for a 16x24 in print with my 5D, what apertuure should I choose?
 Logged
Roy
Full Member

Offline

Posts: 196

Quote
If I have understood Prof. Johnson's article, he is suggesting f21 as the aperture for maximum DOF with FF sensors while maintaining adequate suppression od diffraction effects.  My best recollection is that this is the highest f-number that I have seen recommended. Other writers are suggesting values betweem f11-f16 for FF.  How do we evaluate his assertion? Put another way, I am seeking max DOF with sufficient IQ for a 16x24 in print with my 5D, what apertuure should I choose?
[a href=\"index.php?act=findpost&pid=96567\"][{POST_SNAPBACK}][/a]

That depends on your lens. Johnson's article was about diffraction and DOF, and his numbers assume a perfect lens. If you have a very sharp lens that remains sharp when stopped down, f/21 is your number. But most lenses are "best" at around f/8 and diffraction doesn't suddenly start at f/21, it gradually increases as you stop down. It is at f/21 with a full frame and a perfect lens that diffraction becomes worse than the limit (the circle of confusion) that Johnson has chosen for a sharp image. However, as a real world non-ideal lens is stopped down past its "best" aperture, its contrast, sharpness and other qualities decrease. So maximum DOF may happen well before f/21.

Note that maximum sharpness (not DOF) will happen at the "best" aperture, typically f/8, but different for every lens, and for a zoom, often changing over the zoom range. If you don't need DOF, shoot at the "best" aperture of your lens.

For maximum DOF in the real world with full-frame, f/11 to f/16 is a good recommendation. For 16 by 24 prints, buy the best lenses you can afford. Primes are usually sharper and have more contrast than zooms.
 Logged

Roy
nma
Full Member

Offline

Posts: 162

Quote
That depends on your lens. Johnson's article was about diffraction and DOF, and his numbers assume a perfect lens. If you have a very sharp lens that remains sharp when stopped down, f/21 is your number. But most lenses are "best" at around f/8 and diffraction doesn't suddenly start at f/21, it gradually increases as you stop down. It is at f/21 with a full frame and a perfect lens that diffraction becomes worse than the limit (the circle of confusion) that Johnson has chosen for a sharp image. However, as a real world non-ideal lens is stopped down past its "best" aperture, its contrast, sharpness and other qualities decrease. So maximum DOF may happen well before f/21.

Note that maximum sharpness (not DOF) will happen at the "best" aperture, typically f/8, but different for every lens, and for a zoom, often changing over the zoom range. If you don't need DOF, shoot at the "best" aperture of your lens.

For maximum DOF in the real world with full-frame, f/11 to f/16 is a good recommendation. For 16 by 24 prints, buy the best lenses you can afford. Primes are usually sharper and have more contrast than zooms.
[a href=\"index.php?act=findpost&pid=96609\"][{POST_SNAPBACK}][/a]

Roy and others.

I am still perplexed. A perfect lens would not be subject to diffraction. Johnson's calculations include the effect of diffrection.  If we are left with the f range of 11-16 that I quoted in my initial post as optimal, what is the value of his article? The well known rule of thumb about closing down  two stops to get the "best" a lens has to offer is always quoted. But as a landscape photographer, f8 may not be the best trade off of sharpness and contrast, when I need near to infinity sharpness.
 Logged
EricV
Full Member

Offline

Posts: 137

The original article has all the information you need to answer your question, but it was not presented very clearly.  The author jumps to calculating CoC at the sensor, without considering print size, then later mentions the calculation might have to be adjusted depending on print size.  Here is how you can calculate the answer to your question more logically, starting from print size:

1) Decide on print size, viewing distance, and assumed visual acuity.  Example: 16x24" print, viewed from 18" distance, eyesight angular resolution 1/1500.
2) Calculate allowable CoC on print consistent with eyesight and viewing distance.  Example: CoC = 18" / 1500 = 0.3mm = 300um.  The eye can resolve features this small on the final print.
3) Calculate magnification needed to make desired print given sensor size.  Example: 16x24" print / 1x1.5" sensor --> magnificaiton = 16.
4) Translate CoC from print to sensor by dividing by magnification.  Example: CoC at sensor = 300umm / 16 = 19um.
5) Calculate lens aperture which will give diffraction spot equal to this CoC at sensor, using standard diffraction formula.  Example: 2.44 * lambda * N = 19um (lambda=0.555um) --> N = 14.

So in this example f/14 is the aperture at which diffraction will blur everything by a just barely perceptible amount, given the print size and assumed viewing distance.  Stopping down further than this will increase DoF, but at the expense of overall print sharpness.
 Logged
nma
Full Member

Offline

Posts: 162

Quote
The original article has all the information you need to answer your question, but it was not presented very clearly.  The author jumps to calculating CoC at the sensor, without considering print size, then later mentions the calculation might have to be adjusted depending on print size.  Here is how you can calculate the answer to your question more logically, starting from print size:

1) Decide on print size, viewing distance, and assumed visual acuity.  Example: 16x24" print, viewed from 18" distance, eyesight angular resolution 1/1500.
2) Calculate allowable CoC on print consistent with eyesight and viewing distance.  Example: CoC = 18" / 1500 = 0.3mm = 300um.  The eye can resolve features this small on the final print.
3) Calculate magnification needed to make desired print given sensor size.  Example: 16x24" print / 1x1.5" sensor --> magnificaiton = 16.
4) Translate CoC from print to sensor by dividing by magnification.  Example: CoC at sensor = 300umm / 16 = 19um.
5) Calculate lens aperture which will give diffraction spot equal to this CoC at sensor, using standard diffraction formula.  Example: 2.44 * lambda * N = 19um (lambda=0.555um) --> N = 14.

So in this example f/14 is the aperture at which diffraction will blur everything by a just barely perceptible amount, given the print size and assumed viewing distance.  Stopping down further than this will increase DoF, but at the expense of overall print sharpness.
[a href=\"index.php?act=findpost&pid=96618\"][{POST_SNAPBACK}][/a]

Eric,

I think you have made a good contribution. Thank you.
 Logged
Roy
Full Member

Offline

Posts: 196

Quote
I am still perplexed. A perfect lens would not be subject to diffraction.
[a href=\"index.php?act=findpost&pid=96614\"][{POST_SNAPBACK}][/a]

Every lens is subject to diffraction.

Diffraction is not a property of the lens such that better lenses have less diffraction and poor lenses have more. Diffraction happens when light passes by an opaque object that causes a shadow. Some of the light that passes near the object changes direction and bends around the object causing a fuzzy sahdow. In this case the object is the iris diaphragm in the lens. At a large aperture most of the light passing through the lens is well away from the edge of the diaphragm and goes to the right spot on the film plane (sensor). At a small aperture much more of the light is near the edge of the iris and this is when diffraction becomes a problem.

The optimum aperture is the one that best balances the effect of diffraction and the imperfections of the lens. That varies from lens to lens and also depends on what you are trying to  achieve.

As a rule of thumb, if you want maximum DoF with a full-frame image, try f/11 to f/16. If you want maximum sharpness rather than maximum DoF, use f/8.

If you want better than rules of thumb, test your lenses.

Note that the calculation Eric suggests assumes a perfect lens. Doing the math tells you the smallest aperture you can use on a perfect lens before diffraction effects on their own exceed your chosen CoC, but you may have to back off with a real lens which adds its own distortions. There is no substitute for testing.
 Logged

Roy
Ray
Sr. Member

Offline

Posts: 8947

Quote
If I have understood Prof. Johnson's article, he is suggesting f21 as the aperture for maximum DOF with FF sensors while maintaining adequate suppression od diffraction effects.  My best recollection is that this is the highest f-number that I have seen recommended. Other writers are suggesting values betweem f11-f16 for FF.  How do we evaluate his assertion? Put another way, I am seeking max DOF with sufficient IQ for a 16x24 in print with my 5D, what apertuure should I choose?
[a href=\"index.php?act=findpost&pid=96567\"][{POST_SNAPBACK}][/a]

There will come a point, as you stop down, when the increase in blur at the precise point of focus will equal the increased sharpness at the extremes of the DoF range (in the close foreground for example). At such a point you will have maximum DoF consistent with 'good' resolution. It then becomes a personal choice whether to trade off maximum DoF for a subtle increase in sharpness at the point of focus, and that choice may also be influenced by lens quality.

For example, taking 2 shots of the same scene with my 24-105 zoom and 5D, at f8 and f16, I would expect to find that the point of focus in the f8 shot would be very marginally sharper than in the f16 shot (but nothing to worry about). However, the extremes of the DoF range, in front of and behind the point of focus, would be significantly sharper in the f16 shot.

If I were to use a high quality prime, I would expect the plane of focus in the f8 shot to be slightly better than 'very marginally sharper'. As one stops down into the region dominated by diffraction, all lenses tend to be equal.

Generally, I find when maximum DoF is desired, using my 5D at f16 will deliver the goods without any significant sacrifice of resolution at the plane of focus, but clearly this is not a 'cut and dried' situation. After all, people pay a lot for high quality, razor sharp primes and understandably such lenses are not at their sharpest at f16.
 Logged
larsrc
Full Member

Offline

Posts: 173

Quote
Every lens is subject to diffraction.

Diffraction is not a property of the lens such that better lenses have less diffraction and poor lenses have more. Diffraction happens when light passes by an opaque object that causes a shadow. Some of the light that passes near the object changes direction and bends around the object causing a fuzzy sahdow. In this case the object is the iris diaphragm in the lens. At a large aperture most of the light passing through the lens is well away from the edge of the diaphragm and goes to the right spot on the film plane (sensor). At a small aperture much more of the light is near the edge of the iris and this is when diffraction becomes a problem.

The optimum aperture is the one that best balances the effect of diffraction and the imperfections of the lens. That varies from lens to lens and also depends on what you are trying to  achieve.

As a rule of thumb, if you want maximum DoF with a full-frame image, try f/11 to f/16. If you want maximum sharpness rather than maximum DoF, use f/8.

If you want better than rules of thumb, test your lenses.

Note that the calculation Eric suggests assumes a perfect lens. Doing the math tells you the smallest aperture you can use on a perfect lens before diffraction effects on their own exceed your chosen CoC, but you may have to back off with a real lens which adds its own distortions. There is no substitute for testing.
[{POST_SNAPBACK}][/a]

If you want to see the difference graphically illustrated, SLRGear.com has done a few lens tests on both APS-C and FF, for instance [a href=\"http://www.slrgear.com/reviews/showproduct.php/product/157/cat/10]http://www.slrgear.com/reviews/showproduct...duct/157/cat/10[/url] -  the FF doesn't show significant diffraction at f/22, while the APS-C is getting noticably softer at f/16 and pretty bad at f/22.

-Lars
 Logged

nma
Full Member

Offline

Posts: 162

Quote
s.

Note that the calculation Eric suggests assumes a perfect lens. Doing the math tells you the smallest aperture you can use on a perfect lens before diffraction effects on their own exceed your chosen CoC, but you may have to back off with a real lens which adds its own distortions. There is no substitute for testing.
[a href=\"index.php?act=findpost&pid=96632\"][{POST_SNAPBACK}][/a]

Roy,

Thanks for your help. Just a point of clarification. Eric's calculation merely extends Johnson's theory. Both include the effect of diffraction explicitly and that is what makes them potentially moe valuable then the conventional formulas.  Your point about testing on a real lens is cedrtainly well taken. We need some tests to see if there is general agreement with the theory.
 Logged
bjanes
Sr. Member

Offline

Posts: 2882

Quote
The original article has all the information you need to answer your question, but it was not presented very clearly.  The author jumps to calculating CoC at the sensor, without considering print size, then later mentions the calculation might have to be adjusted depending on print size.  Here is how you can calculate the answer to your question more logically, starting from print size:

1) Decide on print size, viewing distance, and assumed visual acuity.  Example: 16x24" print, viewed from 18" distance, eyesight angular resolution 1/1500.
2) Calculate allowable CoC on print consistent with eyesight and viewing distance.  Example: CoC = 18" / 1500 = 0.3mm = 300um.  The eye can resolve features this small on the final print.
3) Calculate magnification needed to make desired print given sensor size.  Example: 16x24" print / 1x1.5" sensor --> magnificaiton = 16.
4) Translate CoC from print to sensor by dividing by magnification.  Example: CoC at sensor = 300umm / 16 = 19um.
5) Calculate lens aperture which will give diffraction spot equal to this CoC at sensor, using standard diffraction formula.  Example: 2.44 * lambda * N = 19um (lambda=0.555um) --> N = 14.

So in this example f/14 is the aperture at which diffraction will blur everything by a just barely perceptible amount, given the print size and assumed viewing distance.  Stopping down further than this will increase DoF, but at the expense of overall print sharpness.
[{POST_SNAPBACK}][/a]

Eric has stated the issues very clearly. For those who do not want to carry out these calculations manually, there are excellent calculators and further explanations on Sean McHugh's web site:

[a href=\"http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm]Diffraction Calculator[/url]

DOF Calculator
 Logged
 Pages: [1]   Top of Page