Lens Diffraction vs Image Quality

[Edward1064]

While recently attempting some panoramic photography, I was using f-numbers of f/11 to f/16 to increase the DOF. Later I wondered whether light diffraction was spoiling some of the image quality. My equipment was a 7DII and Canon 16-35 mm IS USM f/4 lens.

Using some basic optics, even with a perfect lens, each point on the object is not focused to a point on the sensor, but to a blurred area roughly 2.44*f*lambda/D, where f is the focal length of the lens, lambda is the light wavelength, and D is the lens's aperture diameter. Since the f-number is defined as f/D, the blur reduces to 2.44*lambda*f-number. My camera's pixel size is 4.1 micrometers. If my math is correct, and assuming an average wavelength of 0.55 micrometers, even at f/4 the blur is 1.4 pixels. At f/16 it is 5.2 pixels.

What I don't know is how much this matters to image quality. In the future I plan to examine my photos closely to see whether the theoretical blurring vs f-number actually matters. Has anyone looked into this? I would appreciate some experienced feedback!

Many thanks,

Ed

(Trying to watch Giants vs Dodgers while writing!)

[Edward1064]

EB,

Thank you. Those panos of yours are really interesting. Yes, and image quality is not as important as it is with, say, bird photography. My interest here is to get the best quality photographs that my equipment can muster, not so limited my skills! 

About different lens focal lengths, the diffraction blurring does not depend on the lens focal length, but only on the f-number. This is because the f-number (image distance/lens iris diameter) contains both the lens opening and the image distance. So it is easy to think about.

Please see my comments to Waddizzle. The tree photo was shot at a zoom of 23 mm and f/8 and 1/125 sec.

Ed

[ebiggs1]

Ed,

This is not correct.

"the diffraction blurring does not depend on the lens focal length, but only on the f-number."

Diffraction is worse with WA and UWA lenses because of the short FL. Why?  Because the hole in the diaphram is smaller at any given f-stop compared to the same f-stop in a normal or tele typle lens. In a WA/UAW the light path is bent at a greater  angle.

Another factor is sensor size and pixel size. A camera like the 7D Mk II has fairly small pixels. Its going to be affected sooner than a camera with larger sensor and pixels.

It becomes a photographers decission as shooting at wide open apertures all lenses suffer form optical aberrations.  This is one reason most all lenses show their best when stopped down one or two stops. Diffraction is usually even where aberation is not. This makes diffraction less destructive to lens IQ.

How about DOF. DOF affects different parts of a scene with differing amounts. Remember we said diffraction is even, however, DOF is not.  The farther away from the focal plane you are, the more blurry your picture will get.

Diffraction is easy to correct in Photosop with smart sharpening and/or unsharp mask. Resolving power of most lenses is at its highest or best in the f4 to f8 range. Just anoher factor to consider. 

[Edward1064]

---

---

@ebiggs1 wrote:

Ed,

This is not correct.

"the diffraction blurring does not depend on the lens focal length, but only on the f-number."

Diffraction is worse with WA and UWA lenses because of the short FL. Why?  Because the hole in the diaphram is smaller at any given f-stop compared to the same f-stop in a normal or tele typle lens. In a WA/UAW the light path is bent at a greater  angle.

__________________________________________________________________

EB,

After your interesting comments, I looked into this topic some more.  My conclusion about the importance of light diffraction in lenses was based on thinking about a singlet lens with an aperture right next to it. Of course, modern lenses are complicated affairs and designed my specialized optics engineers. Nevertheless, my conclusion still holds, namely, that the diffraction blurring on the camera sensor is proportional to:

                   2.44 * lambda * f-number

The entrance pupil diameter, which makes up part of the f-number, is defined as the image of the len's aperture stop as seen from the object (front) side of the lens. The link below corroborates the conclusion above. (Since I haven't tried to include a link on this site before, if it does not work just go to Wikipedia and search "Diffraction-limited System".

https://en.wikipedia.org/wiki/Diffraction-limited_system 

For me, it is always helpful to be able to understand things in the simplest, most reduced way, and this is an example. It makes learning easier.

Respectively,

Ed

[ebiggs1]

Ed,

I have no idea about math so that stuff is wasted on me. I only know from experience. WA and UWA suffer more diffraction from a given f-stop than teles do.

So I must repeat..........

This is not correct.

"the diffraction blurring does not depend on the lens focal length, but only on the f-number."

Sometimes stuff is made a lot more difficult by reading too much. Sometimes it is the simple explanation.

[johnrmoyer]

Diffraction blur depends upon F number and the size of pixels on the sensor. It does not depend upon focal length or physical aperture diameter except that F number is calculated from both focal length and aperture diameter. For the same lens, a different sensor will have a different diffraction blur. There might be sources of blur other than diffraction blur, for example depth of field or motion. The combination of focal length and F number will change depth of field blur.

There is a simplified explanation with diagrams illustrating the math and example calulators at https://www.cambridgeincolour.com/tutorials/diffraction-photography.htm

When using a Richardson/Lucy deconvolution to remove diffraction blur, the radius must be chosen based upon the F number and the spacing of pixels on the sensor.

[Edward1064]

Agreed, johnrmoyer.

[ebiggs1]

"Diffraction blur depends upon F number and the size of pixels on the sensor."

I already stated that.

" t does not depend upon focal length or physical aperture diameter except that F number is calculated from both focal length and aperture diameter."

Yes it does. Why? Because the light rays are bent more in a WA/UWA than it is in a tele lens. As FL increase the diameter of the aperture is also increase. This bends the light less. This is simple to understand if you don't understand or reject it I have done all I can.

[Edward1064]

EB,

How do you do a pano, composed of several shots, with movimg subjects?

Ed

[Waddizzle]

---

@Edward1064 wrote:

EB,

How do you do a pano, composed of several shots, with movimg subjects?

Ed

---

I have never purposely tried it.  But, you can probably pull it off using a high frame rate, provided your subjects are neither moving quickly or close to the camera.  

For example,I have captured pano shots of a bridge with moving vehicles,  The individual frames are fine, and are free of any motion blur from moving vehicles.  The only potential for real issues would occur at where the images overlap.  The stitching algorithms tend to throw away differences, and keep what is similar.  

Provded your subjects are not moving any significant difference across the frame from one shot to the next, you should be fine under those circumstances.

[ebiggs1]

Ed,

"How do you do a pano, composed of several shots, with movimg subjects?"

You really can't.  Or at least, you need to be aware of some compromise. The bridge photo, which I have seen many timers on here is beautiful, for sure, but you notice there is no definition in the water in the foreground. That is probably due to the fact it was moving waves or ripples.

A person or animal in a single frame will still remain sharp unless it is next to a seam area. If there is too much motion in the various shots Lightroom will not merge them.

I don't necessarily agree with a multi-shot pano being better IQ. It is certainly a bigger file but the individual shots are no more  higher IQ than any one single shot. This would greatly depend on camera, lenses and conditions. Panos are different than multi-focus stacking which can increase IQ. Panos take multiple shots and puts them together using LR, or some other software, to allow a wider final photo.  Usually because it can exceed what can be contained within just a single frame.

What you may not know is panos also have issues with parallax. A way to correct for parallax error is to use a nodal slide. I don't bother because I am not interested in the highest IQ of these shots because of their final usage. Which is another factor in your decission making, how is the photo going to be used. In fact the two shots shown were hand held using my 1DX, 24-70mm f2.8L lens @ f9.5 (using P mode), SS 1/350 and ISO 200.