Understanding Dispersion

Discussion in 'Canon EOS' started by keith_plechaty, Jan 1, 2010.

  1. I'm a little confused about dispersion. I know that if I want maximum depth of field, I should use a very small aperture. However, I read that if you use your lens's smallest aperture you lose sharpness from dispersion. So how do you know how small to go with aperture to avoid dispersion? I own a 10-22 and 24-105L.
    Thanks for the help!
     
  2. Keith I think you mean diffraction.
    The best thing about digital is you can test all this stuff for yourself. It is all related to print size and viewing distance as well as aperture and sensor size and also the amount of detail in the scene so just run some tests at your normal print size and see what you are happy with.
    As a very general rule of thumb f11 is generally as small as you want to go on an APS-C camera but that is a very broad generalization.
     
  3. Ah - Scott beat me to it...
     
  4. So how much do you want to know?
    PN's own Bob Atkins did a good job explaining it:
    http://www.bobatkins.com/photography/technical/diffraction.html
    Another general overview:
    http://www.luminous-landscape.com/tutorials/understanding-series/u-diffraction.shtml
    This page has a calculator at the bottom:
    http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm
     
  5. Hi Keith, Happy New Year!
    I believe you are talking about diffraction, which does indeed affect fine detail at small apertures. Rather than reinvent the wheel here, I can point you to a page that explains it better than I can - with an interactive portion so you can see how your own sensor/aperture/final usage will affect the degree of diffraction and if your intended use will be diffraction limited or not.
    http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm#
     
  6. Wow - some fast fingers on this New Years Day! Well, apparently great minds think alike and I'd like to extend my best wishes to the posters before me as well as to the rest of the photo.net community.
     
  7. Hell's fire, there were no replies when I started typing this. Now I'm at the back of the queue! Photo.net is busy tonight!
    Dispersion, commonly called diffraction happens at different apertures depending on the lens. As a general rule, most lenses are at their optimum when set at around 2 stops down from their maximum aperture i.e. both your 10-22mm and 24-105mm f4 will both be sharpest at around f8 or so. You probably won't notice any loss of sharpness until f16 and beyond. Even then, it's hardly devastating unless you're a real pixel peeper.
    I was shooting at f32 yesterday and am quite happy with the results. If you want front to back sharpness take a few test shots first (or use depth of field preview or live view). That way you can use the optimum aperture without using an unnecessarily small one.
     
  8. "So how do you know how small to go with aperture to avoid dispersion (caused by diffraction)?"
    When you begin to see sensor dust you didn't know you have, then you know you have gone too far :)
    FWIW, IMHO. It is a trade off one must learn to deal with. There is no simple answer depending, on sensor size, print size, subject size, aperture sizes, lens aberrations and detail needs.
     
  9. You can't avoid diffraction, but if you don't stop down past f16 for APS-C crop sensors and f22 for full frame, diffraction effects won't generally be a problem. Use f11 and f16 if you're picky!
     
  10. You've had correct responses here regarding diffraction. I'll point out for the record that dispersion is not at all the same as diffraction. Dispersion is the separation of light into its component colors, caused by glass refracting different colors to different degrees. It's the cause of chromatic aberration. It's not made worse by stopping down. Lens designers can reduce dispersion by choosing the optical materials used for the lens.
    Diffraction is the spreading of light passing through a small aperture or the bending of light around a small object. It's determined by the lens aperture.
    http://en.wikipedia.org/wiki/Diffraction
    http://en.wikipedia.org/wiki/Dispersion_%28optics%29
     
  11. Keith I think you mean diffraction.​
    The words diffraction and dispersion are intertwined in physics. Dispersion is the separation of light into colors by refraction or diffraction.
    I shoot at F11 and F16 all the time with APS-C and can't see any diffraction degradation, save extreme pixel peeping. It's not something you'll notice in a typical 8x12 or 11x14 print. You do notice the increased DOF.
     
  12. Well yeah, but use of the the word "dispersion" is wrong in this context - we're obviously talking about diffraction, and they're not synonymous.
     
  13. dispersion and diffraction are both important optical terms; but they are diffenent things. Saying they are the same is like saying a dog is cat; or a nail is a board; or a tripod socket is a camera case; or a football is a helmet.
    Dispersion is where the refraction index of glass varies with wavelength; ie color.
    Thus it is what makes rainbows; or causes a glass prism to emit a beam of randow colors. Dispersion causes chromatic aberration. Thus a simple achromat uses a crown and flint glass types in design to reduce chromatic errors.

    Diffraction is where the optical design is good enough to where one is diffraction limited; a GOOD thing.
    You Want a lens to be diffraction limited; just like a normal person wants low taxes or good health.

    Diffraction limits the performance of a lens; it cannot be better than perfect ie diffraction limited. In the slower fstops of say F16, F22, F32 a len might be diffraction limited but have a lessor performance when say at F11 or F8; since the diffraction limits boxes in the max performace. A rough swag is line pairs per mm time fstop might be say 1500; thus an F50 lens cannot be better than 30 line pairs. The same lens say at F11 might be better; say 50 line pairs but not be fully diffraction limited; if it was it might be higher in performance.

    One really wants to have a giant million dollar income (fast lens F1.0 ); and also pay no taxes ( diffraction limited). It is not that paying less taxes or diffraction is bad; you really just want it to occur when one is at a faster fstop; or makes when one makes more money too. In a lay sense; folks who say they want to avoid diffraction sound odd; when it is the ultimate goal in optics; a perfect lens.

    With most lenses ever made; the best performance is not wide open or totally stopped down either. A common 50mm F2 lens is almost never best at F22; or F2 either; typically it is say F5.6 to F11. Same goes with cars; the best MPG is not 10 mph; and not 110 mph; it is usually about 35 to 45 mph. Where a lens peaks is not at F22; one has diffraction BUT F16 is about diffraction limted too; and thus opening up gives a better performace. A lens is not best at F2; one has alot of abberations. If the lens was diffraction limited at F2; one would have a super lens; many many hundreds of lines per mm; more than film can record.

    Imagine if the same job paid 10000 with no taxes; or 50000 with no taxes; or 100k with 20 percent taxes; or 500k with 50 percent taxes; or 1 million with 90 percent taxes; which is best? one gets 10k; or 50k; or 80k; or 250k; or 100k. You take the F8 job; ie you take home 1/4 million. If one was diffaction limited one would pay no tax and one would take home all; ie 500k. The peak in take home is NOT where one pays no taxes; it is where the takehome peaks; a balance of income and what is taken out to fund the dole.

    In cars the peak in mpg is about where rolling drag and air drag are equal. Air drag goes up with the square of speed; horsepower thus goes up as the cube. At slow speeds rolling resistance dominates; at high speed air drag dominates. Due to the way gas engines are geared and because extracting a tiny fraction of horsepower is not effecient; MPG rises at first; then peaks; then drops. This happens with model T's; trains; Ford Falcons; any gas engine car; the queen mary too.
     
  14. Kelly, I have a PhD in the physical sciences, and I didn't understand a word you just said! Maybe I should study less physics and more accounting. ; )
    Dispersion: the change of "direction" of light as a function of it's wavelength (color) when passing through the lens; these different wavelengths then focus at slightly different points along the optical axis of the lens; depends on quality of the lens; typically more of an issue near the lens edges, so becomes a factor on image sharpness at the widest apertures
    Keith, you want to increase DOF while avoiding decrease in sharpness due to diffraction : the spread of ALL wavelengths of light when passing through very small apertures; the diffraction limit kicks in at slightly different apertures depending on the details of the lens/image plane geometry, but for 35mm cameras is typically somewhere between f8 and f16.
     
  15. SLR gear tests different lenses for various criteria. Here are links to the two lenses. You click on the blur index chart and can check the lens for the different f-stops and focal lengths.
    10-22
    http://www.slrgear.com/reviews/showproduct.php/product/135/cat/11
    24-105
    http://www.slrgear.com/reviews/showproduct.php/product/145/cat/11
     
  16. What Asa said, but it is important to recognise that diffraction effects are frequency-dependent and so can be one cause of dispersion, which is the separation, by whatever means, of different frequencies. The dependence of refractive index on frequency is another, more familiar, cause of dispersion. Canon's DO lenses are (?were) an attempt to use these effects in combination, but working in opposite directions, to correct for chromatic aberration. It is clear that what the OP (Keith) meant was diffraction not dispersion.
    A couple of months back I posted what to me was an interesting example of ring patterns ("onion rings") in an image with out-of-focus bright highlights, and I conjectured that these were some form of diffraction pattern (NB this was NOT about DO lenses). I didn't get any definitive answers from those who posted to the thread, but I did my own homework (no too hard for a (former) professional mathematician, unfortunately completely inaccessible for anyone without a maths/physics background) and reached the conclusion that what I was seeing was explained by the Fresnel theory of the diffraction of spherical wavefronts at a circular aperture, applied to what happens away from the plane of focus. The basic ideas aren't too hard, but there is a good deal of tecnical complexity over turning them into "formulae" for what is going on and then in turn producing actual calculated results, and this last stage defeated the nineteenth-century mathematicians who did not have computers available. I also reached the view that the accounts of diffraction offered to photographers were over-simplified to the point that they failed to draw attention to the sort of very visible effects that I was seeing, since they typically took the much simpler Fraunhöfer theory of plane wave diffraction and used it as an approximation to what is going on in the plane of focus. The effects there are of progressive image blurring of a not very interesting kind at small apertures, whereas out of the plane of focus relatively large onion-ring pattens can arise. Since these, rather than nice simple uniform discs, are the true building blocks of the out-of-focus part of any image, it is not surprising that the nature of "bokeh" is rather mysterious.
     
  17. If a 50mm F2 lens was Diffraction limited at ALL fstops then it would be best wide open at F2. Thus folks who do not understand difraction would avoid this lens because it is perfect.
    example (1) Crummy 50mm F2 lens is diffraction limited at F32; peak performance at F22; performance is worse at F16; and more worse at each faster fstop; is crap at F4; total crap at F2.
    example (2) Normal 50mm F2 lens; peak in performance in the F5.6 to F11 area; diffraction limited at F8 thru F32
    example (3) perfect 50mm F2 lens peaks in performance wide open at F2; is diffraction limited at ALL fstops The best performance is wide open; since the lens is diffraction limited.
    example (4) a super normal lens; it peaks in performance at F4 is diffraction limited at F8 to F32
    **** When a lens is diffraction limited at several fstops; the fastest diffraction limited fstop is where the spot size is smallest; ie best resolution ****
    By saying one wants to avoid diffraction; most lay non optical folks really mean is to avoid poor performance. This means to seek the best fstop; to where a lens is diffraction limited; ie the FASTEST FSTOP where diffraction really STARTS to occur.
    With an optical bench one can place any lens and observe a point light source; like an artifical star from a special projector. A lens like a 50mm F2 will show a blob of light wide open at the focus; even on axis. As one stops down the iris the spot size drops in size; then is hits a minimum in size when it is about diffraction limited. At this point the Airy disc ie rings are seen ; then further stopping down the rings/spot grow in size. In practice one normally see the first bright ring; the 2nd bright ring is more difficult to see.

    *****With the graphically example below; a 50mm F2 lens is best at F8; where diffraction just starts. The lens is diffraction limited beyond F8 too; ie it is diffraction limited at F11; F16; F22. Further stopping down makes the spot size grow larger. As a summary; this example is diffraction limited from about F8 to F22. It is best at F8; where diffraction just starts . At F4 the lens is NOT diffraction limited; thus the spot size is LARGER than a lens designers ultimate dream of a perfect lens ie diffraction limited.

    This stuff is in many optical books. It is in Sam Brown Edmund books on optics that were a 15 cents in the 1950's. It is in the 1930's Kodak books . It is in a book on photography we had in Indiana in a dinky town; the book was from about 1890.
    The lens is better at F22 than F4.
    [​IMG]
     
  18. If one takes a rather poor 50mm F2.8 triplet like an Exakta Domiplan and a great 50mm F1.8 Canon and tests them at F22; both might have sort of similar results; decent but not the best each can perform. Both are diffraction limited; one will see the airy disc on each. It will be about the same size too.

    At F22 where each is diffraction limited both perform better than when then are wide open.
    With the Better lens like the Canon F1.8; the fstop at which the * best performance occurs* will be a faster one; maybe say F5.6 on the Canon; maybe F16 on the ill Domiplan.

    With most all lenses ever made for photo usage; enlargers etc; the *best* performing fstop is usually not the fastest one; it is one stopped down a few stops. In a few super super rare cases; the fastest fstop is the best. The old famed 300mm F2.8 Canon FL-F SSC lens that Modern Photo tested back in March 1974 as like this. It was sharpest wide open; but contrast peaked when stopped down abit.

    The goal is to get the best image; ie best ideal fstop. If your camera or subject moves none of this matters. A faster fstop and thus a faster shutter speed is often better for shooting sports; kids stuff other than landscapes and granite blocks.
    If one uses a homemade Walmart magnifier 50mm focal length lens and one makes homemade fstops from F2 to F22; the best fstop will probably be F22.
     
  19. Sheesh, you guys scared the hell out of the OP...
     

Share This Page