Diffraction?

[Nigel Wilkins]

almanacia wrote:

Is diffraction a rule we should all stick to? Or is it simply a guideline that is meant to be crossed?

Thanks for making my signature more interesting

--
nigelwilkinsphotography.com

Everything to do with photography is a guideline. The only rules are generally enforced by the police.

 

[Gerry Winterbourne]

JulesJ wrote:

BertIverson wrote:

stevo23 wrote:

Leonard Migliore wrote:

JulesJ wrote:

I was taught that the rule of thumb was that a lens performed its best at two stops up from it's widest.

The issue here is not what stop to use to get the best performance, it's how much loss do you get when you stop down a lot. There are many times when I will use f/8 on a lens that's best at f/4, generally because of the greater depth of field that I get. If I really really need depth of field, then I'll stop down to f/16 even though I will lose sharpness in the in-focus areas. Picture looks better.

This depends on the picture for me. f/16 is easily tolerable in a f/16 is easily tolerable in a tight macro shot where the details of the scene are so prominent. But in a scenic it's not as helpful.

No one has mentioned the sensor size in this discussion ????

No one has mention rice pudding either!

No, but rice pudding has no influence on diffraction while sensor size is one of the most important things affecting visibility of diffraction.

 

[Nigel Wilkins]

Gerry Winterbourne wrote:

JulesJ wrote:

BertIverson wrote:

stevo23 wrote:

Leonard Migliore wrote:

JulesJ wrote:

I was taught that the rule of thumb was that a lens performed its best at two stops up from it's widest.

The issue here is not what stop to use to get the best performance, it's how much loss do you get when you stop down a lot. There are many times when I will use f/8 on a lens that's best at f/4, generally because of the greater depth of field that I get. If I really really need depth of field, then I'll stop down to f/16 even though I will lose sharpness in the in-focus areas. Picture looks better.

This depends on the picture for me. f/16 is easily tolerable in a f/16 is easily tolerable in a tight macro shot where the details of the scene are so prominent. But in a scenic it's not as helpful.

No one has mentioned the sensor size in this discussion ????

No one has mention rice pudding either!

No, but rice pudding has no influence on diffraction while sensor size is one of the most important things affecting visibility of diffraction.

or pixel size, to be pedantic about it.

In fact, rice pudding doesn't show diffraction up very well due to it's unusually low contrast, so you can probably stop down a little further.

--

nigelwilkinsphotography.com
.
Everything to do with photography is a guideline. The only rules are generally enforced by the police.

 

[stevo23]

BertIverson wrote:

stevo23 wrote:

Leonard Migliore wrote:

JulesJ wrote:

I was taught that the rule of thumb was that a lens performed its best at two stops up from it's widest.

The issue here is not what stop to use to get the best performance, it's how much loss do you get when you stop down a lot. There are many times when I will use f/8 on a lens that's best at f/4, generally because of the greater depth of field that I get. If I really really need depth of field, then I'll stop down to f/16 even though I will lose sharpness in the in-focus areas. Picture looks better.

This depends on the picture for me. f/16 is easily tolerable in a f/16 is easily tolerable in a tight macro shot where the details of the scene are so prominent. But in a scenic it's not as helpful.

No one has mentioned the sensor size in this discussion ????

Sounds like you're volunteering.

 

[BertIverson]

stevo23 wrote:

BertIverson wrote:

No one has mentioned the sensor size in this discussion ????

Sounds like you're volunteering.

Well not quite. But I did hope someone would bring up the subject since sensor size is quite relevant when discussing aperture and diffraction. In fact it is so important that most PS with 1/2.3 sensors do not even have a diaphragm (the lens is always wide open -- possibly with an ND filter)

Gerry hit the nail on the head in his post above:
"No, but rice pudding has no influence on diffraction while sensor size is one of the most important things affecting visibility of diffraction."

:-D
Bert

 

[stevo23]

BertIverson wrote:

stevo23 wrote:

BertIverson wrote:

No one has mentioned the sensor size in this discussion ????

Sounds like you're volunteering.

Well not quite. But I did hope someone would bring up the subject since sensor size is quite relevant when discussing aperture and diffraction. In fact it is so important that most PS with 1/2.3 sensors do not even have a diaphragm (the lens is always wide open -- possibly with an ND filter)

Gerry hit the nail on the head in his post above:
"No, but rice pudding has no influence on diffraction while sensor size is one of the most important things affecting visibility of diffraction."

:-D
Bert

Honestly, I have no idea what it means.

But diffraction is real and visible. It's affects are only objectionable to the degree that it starts to affect the final output. If you're printing 5x7s from a point and shoot with a 2Mp sensor, diffraction may be masked by the relatively low resolution sensor. But if you're shooting with a full frame 24Mp camera and a very high resolution lens, you'll start to see it by f/8 although it doesn't usually become objectionable until after f/11. But even that depends on the subject and style. Macro shots can tolerate more.

 

[Mike Davis]

If the final print is small enough, or your desired print resolution is low enough, using f-Numbers greater than f/11 with this sensor could provide more DoF without causing diffraction to inhibit your desired print resolution. It would be a shame to always shoot at f/11 or smaller f-Numbers, for fear of diffraction that's only a problem at higher enlargement factors and/or higher desired print resolutions.

There is no single f-Number at which diffraction will begin to inhibit a desired print resolution - it varies with your anticipated enlargement factor, because the Airy disks recorded at the sensor (or film) plane with a given f-Number will suffer varying degrees of magnification for different size prints examined at an anticipated viewing distance.

If you intend to make 4x6-inch prints or even smaller wallet photos, to be viewed at a distance of 20-inches, you'll have a lot less concern for the possibility of either diffraction or defocus inhibiting whatever you personally consider to be an acceptable level of subject detail, than you would when making larger prints. You can feel free to shoot at just about any aperture available on your lens when the enlargement factor is small, the viewing distance is great, or your personal requirement for resolution in the final print is low.

If instead you intend to make 8x12-inch prints, that must survive scrutiny at a viewing distance of only 10 inches, you'll likely have a preference for greater resolution in the final print, and will therefore suffer a greater possibility of diffraction or defocus inhibiting the print resolution you desire.

Thus, it's impossible for anyone to recommend an aperture at which "diffraction begins to destroy sharpness" for any given camera, without knowing the enlargement factor at which you plan to print your images -and- your desired print resolution (or, at the very least, your anticipated viewing distance, so that we can recommend a desired print resolution.)

Again: There is no single f-Number at which diffraction "becomes a problem" for any camera at all combinations of enlargement factor and desired print resolution.

Note that an image for an enormous roadside billboard does not have to be shot with a 10,000 Megapixel camera because the associated viewing distance is typically hundreds of feet. A 24x36-inch print viewed at a distance of 20 inches can similarly appear to have every bit as much subject detail as when the same file is printed to a 12x18-inch print for viewing at half that distance.

Yet somehow, anticipated enlargement factor and specification of the resolution one personally hopes to record in the final print are, more often than not, completely ignored in discussions of aperture selection for controlling either diffraction or DoF.

A good number of DoF calculators don't even allow user-specification of a custom CoC diameter, instead allowing the user to specify only the focal length plus the near and far distances of the subject space. Countless people have been disappointed by the results had when they make use of the DoF scales engraved on their lens barrels. Why are they disappointed? Because enlargement factor and the amount of resolution they personally hoped to secure in the final image (which itself should take viewing distance into consideration) are ignored by such tools!

Have a look at this formula from Wikipedia's Circle of Confusion page, http://en.wikipedia.org/wiki/Circle_of_confusion

CoC (mm) = viewing distance (cm) / desired final-image resolution (lp/mm) for a 25 cm viewing distance / enlargement / 25

For example, to support a final-image resolution equivalent to 5 lp/mm for a 25 cm viewing distance when the anticipated viewing distance is 50 cm and the anticipated enlargement is 8:

CoC = 50 / 5 / 8 / 25 = 0.05 mm

Notice that the only variables in this equation for calculating the maximum CoC diameter to be used in DoF calculations are those for viewing distance, desired final image resolution (in lp/mm), and enlargement factor, but again, these variables are seldom proclaimed when discussing aperture selection.

A viewing distance of 25 cm is about 9.84 inches - about the closest that a person with healthy vision can focus with the naked eyes. If you are willing to confine your specification of a desired final image resolution to always satisfy a viewing distance of 25cm, you can reduce the Max. CoC calculation to this equation:

CoC (mm) = 1 / desired final-image resolution (lp/mm) for a 25 cm viewing distance / enlargement factor

Here's a similar formula for determining the aperture, f/N, at which diffraction's Airy disks will begin to inhibit your desired final image resolution (in lp/mm) at the anticipated enlargement factor for a 25cm viewing distance:

N = 1 / desired print resolution (lp/mm) / anticipated enlargement factor / 0.00135383

Notice that it's just the calculated CoC value divided by the constant 0.00135383.

See David M. Jacobson's Lens Tutorial for the origin of this constant: http://graflex.org/lenses/photographic-lenses-tutorial.html .

This same math is employed by Sean McHugh's diffraction tutorial at http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm .

(Continued below...)

--

Mike Davis

Home

 

[Mike Davis]

(...Continued from above)

For calculating either the CoC diameter that you can specify in your choice of DoF calculator (of those that permit user-specification of max. CoC) or for calculating the f-Number at which diffraction will begin to inhibit a desired final image resolution at an anticipated enlargement factor, here are some definitions of the two variables used in each equation:

Enlargement factor is simply the ratio of the final image dimensions to your sensor or film dimensions. (Divide print diagonal by sensor diagonal, or print width by sensor width, for example.)

For your desired final image resolution, I recommend a value of 4- to 8-lp/mm when you want the print to survive scrutiny at a viewing distance of 10 inches. 4 lp/mm will satisfy most people at that viewing distance and anything greater than 8 lp/mm would be overkill (despite Ctein saying that some people can appreciate resolutions as high as 25 lp/mm in a print viewed at 25cm). If you only want to satisfy a viewing distance of 20 inches, you can cut my recommendation for desired resolution (at 10 inches) in half.

A desired final image resolution of 2 lp/mm requires an unresampled image file resolution of 100 dpi (for a sensor that lacks an AA filter) or 144 dpi for a typical CMOS sensor, due to a 30% loss of resolution typically imposed by the RGBG Bayer algorithm and AA filter.

A desired final image resolution of 3 lp/mm requires an unresampled image file resolution of 150 dpi (no AA filter) or 216 dpi (with 30% Bayer and AA losses).

A desired final image resolution of 4 lp/mm requires an unresampled image file resolution of 200 dpi (no AA filter) or 288 dpi (with 30% Bayer and AA losses).

A desired final image resolution of 5 lp/mm requires an unresampled image file resolution of 250 dpi (no AA filter) or 360 dpi (with 30% Bayer and AA losses).

A desired final image resolution of 6 lp/mm requires an unresampled image file resolution of 300 dpi (no AA filter) or 432 dpi (with 30% Bayer and AA losses).

A desired final image resolution of 7 lp/mm requires an unresampled image file resolution of 350 dpi (no AA filter) or 504 dpi (with 30% Bayer and AA losses).

A desired final image resolution of 8 lp/mm requires an unresampled image file resolution of 400 dpi (no AA filter) or 576 dpi (with 30% Bayer and AA losses).

Before you can select a "desired" final image resolution, you have to be realistic when selecting an enlargement factor.

For example, the Nikon D300's 15.7 x 23.7mm sensor captures 4288 x 2848 pixels (12.21 MP) at a moderate pixel density of 181 pixels/mm.

Taking into account the loss of resolution caused by the RGBG Bayer algorithm and AA filter, typically a 30% loss relative to actual pixel count, if you "desire" to record subject detail in the final image at a resolution of 5 lp/mm, you'll be limited by pixel count to the following print dimensions:

Max. 5 lp/mm Width: 4288 pixels / 360 dpi = 11.91 inches

Max. 5 lp/mm Height: 2848 pixels / 360 dpi = 7.91 inches

Assuming you plan to make 7.91 x 11.91-inch prints (where the Pixel Count will support a desired print resolution of 5 lp/mm), your enlargement factor (without cropping) would be 11.91 inches / 23.7mm = 302.54 mm / 23.7mm = 12.8x

Now let's run the two equations for an anticipated viewing distance of 25cm (9.84-inches):

FOR CONTROLLING DEFOCUS:

CoC (mm) = 1 / desired final-image resolution (lp/mm) for a 25 cm viewing distance / enlargement factor

CoC (mm) = 1 / 5 / 12.8 = 0.0156 mm

Your largest CoCs, which will occur at the near and far limits of DoF, must not exceed 0.0156 mm at the sensor, before enlargement. Plug this CoC diameter into your choice of DoF calculators (www.dofmaster.com, for example), then follow the calculator's recommendations for setting both your working aperture and your focus distance.

FOR CONTROLLING DIFFRACTION:

N = 1 / desired print resolution (lp/mm) / anticipated enlargement factor / 0.00135383

N = 1 / 5 / 12.8 / 0.00135383 = 11.54 (or f/11.54)

Thus, even when your DoF requirements indicate that you should use an f-Number larger than f/11.54, doing so would cause diffraction to inhibit your desired goal of 5 lp/mm in prints that require an enlargement factor of 12.8x.

(Continued below...)

 

[Mike Davis]

(...Continued from above)

Unfortunately, the loss of subject detail caused by diffraction impacts the entire image uniformly, where loss of subject detail caused by defocus occurs only at the near and far sharps of DoF. Thus, it's more difficult to discern a loss of resolution that impacts the entire print vs. a loss of resolution that occurs only at the nearest or farthest distances in the subject space, unless you make a side-by-side comparison of two prints.

Any print can be made "sharp" by enhancing its acutance (edge sharpness), through the use of tools like unsharp mask in PS, but rendering actual subject detail at resolutions that cause people to gasp in appreciation, requires that you control both defocus and diffraction, to ensure that both the circles of confusion caused by defocus -and- diffraction's Airy disks are kept small enough, after enlargement, to satisfy your desired print resolution (which should have been chosen with consideration for an anticipated viewing distance).

For a better understanding of the difference between acutance and resolution, see Sean McHugh's tutorial on "sharpness" at http://www.cambridgeincolour.com/tutorials/sharpness.htm

Remember that if the DoF calculator calls for an f-Number that you can't use (because the corresponding shutter speed would be too slow, or because that f-Number is larger than the f-Number at which diffraction will become visible, or because your lens simply doesn't offer that f-Number), all you have to do is....

Back away from the nearest subject until you're far enough away to use a viable f-Number (you'll have to recalculate DoF for this new range of subject distances).

-OR-

Go to a shorter focal length without moving the camera (you'll have to recalculate DoF for this new focal length).

-OR-

Resign yourself to making a smaller print: Reducing the enlargement factor by 1.414x (making a 10-inch print instead of a 14-inch print) allows you to open up one stop and get the same apparent resolution. Reducing the enlargement factor by 2x (making a 7-inch print instead of a 14-inch print) allows you to open up two stops and get the same apparent resolution).

-OR-

Resign yourself to hanging the print in a location (over a piano?) where people can't examine it at your originally anticipated viewing distance: Increasing the viewing distance by a factor of 1.414x allows you to open up one stop. Increasing the viewing distance by a factor of 2x allows you to open up two stops. (This last solution is difficult to enforce, so it's not practical.)

Sometimes, in the interest of maintaining your intended composition (as defined by your original choice of camera position and focal length), it's best to just go for a smaller print (an enlargement factor lower than that specified when you calculated the Circle of Confusion diameter used to produce your DoF calculators) rather than backing away from the nearest subject or selecting a shorter focal length - both of which can change the composition drastically. But you've got to remember that you made this choice, in the field, to go with a smaller print. Don't cop out later and produce the full sized print only to suffer visible degradation caused by defocus and/or diffraction. Stay the course.

Lastly, please realize that there are many factors other than available pixel count, defocus and diffraction that can prevent you from achieving a "desired" final image resolution. What I've written here only describes an approach to controlling defocus and diffraction, with no attention given to lens resolving power at various apertures, the smearing caused by subject or camera motion at inadequate shutter speeds, film resolution and in-camera flatness (for those who are still shooting film), etc. Defocus and diffraction are, however, among the most controllable of factors affecting final image resolution, if you're willing to exercise that control instead of just rolling the dice every time you make an exposure.

For more on how I approach landscape photography (only one of many ways to boil the proverbial egg), read these five posts (or the entire thread) from the Canon Digital Photography Forum:

http://photography-on-the.net/forum/showpost.php?p=2756604&postcount=25

http://photography-on-the.net/forum/showpost.php?p=2780643&postcount=39

http://photography-on-the.net/forum/showpost.php?p=2784205&postcount=45

http://photography-on-the.net/forum/showpost.php?p=2786373&postcount=54

http://photography-on-the.net/forum/showpost.php?p=2788383&postcount=57

The complete thread starts here:

http://photography-on-the.net/forum/showthread.php?t=279951

 

[Mike Davis]

Here you go - an Excel spreadsheet inspired by DPReview members jpr2 and shadowhumper:

http://www.accessz.com/tools/CoC&f-NumToSupportDesiredPrintRes.xls

This will negate the need for doing the math with a four-function calculator, if you have Excel.

Enjoy!

 

[Chris Malcolm]

Shape of iris is a factor often ignored in diffraction discussions. For example you get a lot of extra diffraction from corners, which is exploited to get the starburst rays effect when overexposing a bright light source a lot. Then there's the extra diffraction you get from the combined inner and outer edges of the ring shaped fixed aperture of a reflex lens.

 

[_sem_]

Diffraction is real, of course.

But mind that pure diffraction blur responds relatively well to sharpening. So the sharpness loss with state-of-the-art sharpening applied may not be as bad as it may seem from straight-out-of-the-camera images. With more megapixels and improved deconvolution sharpening methods this may improve further.

 

[Mike Davis]

Points well taken, Chris! I wish I knew how to quantify those factors, but I'm hoping their effect on aperture selection is negligible.

Thanks!

 

[Glen Barrington]

Shape of iris is a factor often ignored in diffraction discussions. For example you get a lot of extra diffraction from corners, which is exploited to get the starburst rays effect when overexposing a bright light source a lot. Then there's the extra diffraction you get from the combined inner and outer edges of the ring shaped fixed aperture of a reflex lens.

 

[stevo23]

Diffraction is real, of course.

But mind that pure diffraction blur responds relatively well to sharpening. So the sharpness loss with state-of-the-art sharpening applied may not be as bad as it may seem from straight-out-of-the-camera images. With more megapixels and improved deconvolution sharpening methods this may improve further.

I don't think it does in a positive way personally. I've tried some of the demo software and many methods to bring blur back from the brink and none have been satisfactory to me.

 

[wfektar]

Fourier transform of the aperture function, modulus square to get to irradiance level.

 

[_sem_]

Diffraction is real, of course.

But mind that pure diffraction blur responds relatively well to sharpening. So the sharpness loss with state-of-the-art sharpening applied may not be as bad as it may seem from straight-out-of-the-camera images. With more megapixels and improved deconvolution sharpening methods this may improve further.

I don't think it does in a positive way personally. I've tried some of the demo software and many methods to bring blur back from the brink and none have been satisfactory to me.

I've seen a some quite convincing examples on this forum. But surely one can't expect to get perfect results with the lens stopped down all the way.

One problem is that the best examples invariably come from test charts and currency as objects 2D deconvolution is relatively straightforward (yet more complex than USM), but real-life images are 3D. Depth info and 3D processing would be needed for 3D objects.

 

[Mike Davis]

Fourier transform of the aperture function, modulus square to get to irradiance level.

Unfortunately, I'm not a mathematician - I'm a consumer of algebra, at best.

:-D

 

[Mike Davis]

Here you go - an Excel spreadsheet inspired by DPReview members jpr2 and shadowhumper:

http://www.accessz.com/tools/CoC&f-NumToSupportDesiredPrintRes.xls

This will negate the need for doing the math with a four-function calculator, if you have Excel.

Enjoy!

 

[stevo23]

Diffraction is real, of course.

But mind that pure diffraction blur responds relatively well to sharpening. So the sharpness loss with state-of-the-art sharpening applied may not be as bad as it may seem from straight-out-of-the-camera images. With more megapixels and improved deconvolution sharpening methods this may improve further.

I don't think it does in a positive way personally. I've tried some of the demo software and many methods to bring blur back from the brink and none have been satisfactory to me.

I've seen a some quite convincing examples on this forum. But surely one can't expect to get perfect results with the lens stopped down all the way.

One problem is that the best examples invariably come from test charts and currency as objects 2D deconvolution is relatively straightforward (yet more complex than USM), but real-life images are 3D. Depth info and 3D processing would be needed for 3D objects.

Me thinks the best of it will be the optical equivalent of "auto tune" on voices. It can be subtle and nearly undetectable, but the sense what you're not hearing the real thing is irritating. A lot like breast implants...