Erik's web

  • Increase font size
  • Default font size
  • Decrease font size

Effects of diffraction

E-mail Print PDF

There is a tendency that each new generation of DSLR cameras has sensors with higher resolutions. My cameras may be an illustration


2005 Konica Minolta 7D 6 MP
2007 Sony Alpha 100 10MP
2007 Sony Alpha 700 12 MP
2008 Sony Alpha 900 24 MP (full frame)
2011 Sony Alpha 77 SLT 24 MP


Obviously, the newer cameras with higher resolution can make better use of the best lenses than the older models, but they also raise the ribbon regarding lens quality and shooting technique.


One of the factors limiting resolution is a property of light called diffraction. As said diffraction is a property of light, nothing lens designers can do a lot about. Diffraction will affect the best Zeiss lens in the exactly same way as a cheap plastic lens. But, the more you have the more you have to loose.

This article will try to illustrate the effects of diffraction and dispel some of the misconceptions about it.


What is diffraction?

I will not dwelve to much into diffraction, as there are excellent articles about it like this one.
Any way,  when light passes an aperture it will bend. This bendig of the makes that parallel rays of light will not be focused in a single point but in a distribution of light, illustrated by this figures from the aforementioned article:


The left image illustrates that the typical effect of diffraction is a central disk of light with a few rings around. The central disk is not uniform. The second figure illustrates the distribution of light n what is now as the Airy pattern.


Normally, the spot image produced by a lens is dominate by aberrations at large apertures, stopping down will reduce the diameter of the spot image, until the reduction runs in the increasing radius aóf the Airy pattern. If the first Airy circle is visible on a spot image on a optical bench at a given aperture it is said that the lens is diffraction limited at that aperture. The better the lens the more is it limited by diffraction.


So, what does this mean in practice?

It is well known that almost all lenses needed to be stopped down to achieve maximum performance. Once we stop down to optimal aperture resolution will start to drop when stopping down further, this drop off is due to diffraction. This effect can be quite noticable.

These images were take with a decent macro lens at f/5.6 and f/8 respectively. It doesn't take much of eyesight to note that the f/5.6 image is noticable sharper. The image was taken with a 16MP APS-C cameram, with 4.77 micron pixel pitch. This corresponds pretty close to the new Nikon D800 full frame at 36 MP. The figures below the images show the diameter of the Airy disk for green light in realtion to the pixel pitch.



f/5.6 f/8

Diffraction obviously kills image qulity when stopping down. Real life subjects are on the other hand three dimensional, so that we need to stop down. If we assume that a part of the image is slightly defocused the focusing error will cause a disc of unsharpness normally called "circle of confusion" (CoC). Stopping down reduces the size of the "CoC". But stopping down will also increase the diamter of the Airy disc. When the diameter of the Airy disc exceeds the diameter of the CoC, sharpness will decline when stopping down. The images below were defocused 6 cm at 300 cm object distance.


f/5.6 f/8 f/16

The reason I choose to use f/16 to illustrate the drop of the image quality is that the Airy disk is not uniformly lit. Most of the light is falling in the center, so the sharpness fallof is not clearly visible at f/11 but very much so at f/16.


Does incresing sensor resolution make the sensor more sensitive to diffraction?

No, not really. But, the more you have the more you have to loose. I measured MTF 50 using Imatest on unsharpened images from thre generations of SLRs using the same lens in the same setup. The results are shown below:

Here we can see that the 24 MP Alpha 77SLT outresolves the 12 MP Alpha 700 and the 6 MP Dimage 7D at any aperture. We also see that the Dimage 7D looses about 23% of it's resolution at f/16, while the Alpha 77SLT looses around 30%. We also see that the Alpha 700 gained much on Dimage 7D but the Alpha 77SLT gained little over the Alpha 700. This may indicate that the Alpha 77SLT may be limited by lens.


The above image shows the full MTF curve for the 6MP Dynax 7D at f/5.6 and f/16 respectively. We can see that altough the MTF curve is well below the MTF limit even at f/16 the effect of diffractio is visible. If we look at 2000 LW/PH the MTF is 12.3% at f/5.6 and 6.1% at f/16.

Now, turning to the Alpha 77, we can see that MTF at 2000 LW/PH is now around 35%, so the higher resolution sensor transfers three times more contrast at what would be the Nyquist limit of the Dynax 7D sensor. Stopping down to f/16 we only transfer 20% but that is still three times as much as what we had with the 6MP sensor. In addition we still have decent detail up to 3000-3500 LW/PH, that can be emphasized by adequate sharpening.


What can we regain with sharpening?

If we use modern sharpening methods, that is deconvolution with an adequate approximation of the Pointt Spread Function we can regain a lot of the MTF lost due to stopping down. But once MTF drops below some level the resolution will be lost.
This thread discusses deconvulution in good detail.
The figure below ilustrates the issue. It is a crop of my test target with what I consider optimal sharpening (to taste) for the f/16 and the f/5.6 image. Both images may look similar in sharpness, depending on viewing conditions, but the f/16 image does not resolve some of the fine line pattern on the bank note. The pattern is gone.
Sharpening was in Lightroom 4.1 (detail set to 100 activates deconvolution in LR).
Aperture Amount Radius Detail Masking
f/16 75 1.3 100 17
f/5.6 55 0.8 100 17



Last Updated on Thursday, 26 July 2012 20:52  


OS : Linux u
PHP : 5.2.9
MySQL : 5.0.67
Time : 13:52
Caching : Enabled
GZIP : Disabled
Members : 1510
Content : 72
Web Links : 1
Content View Hits : 359342