May 30, 2015, 12:54:52 AM

Author Topic: Say What? The 5DS, f/11 and Confusing Circles  (Read 7411 times)

Busted Knuckles

  • Rebel T5i
  • ****
  • Posts: 102
Re: Say What? The 5DS, f/11 and Confusing Circles
« Reply #30 on: May 02, 2015, 08:44:45 AM »
Give me pixel density or give me death! or something like that.

I had an aha moment in the circles of confusion discussion.  The concept creates an mental image of a few circles falling neatly onto a photosite.  Depending on the diameter of the photosite either neatly contained within the microlens or horror if the photosite is too small overlapping onto the next.

My moment is that there is typically infinite airy disks and circles in every direction.  So ever smaller photo sites will better resolve next infinitely close circle of confusion or airy disk

I can't recall if I posted it before, but asked the rhetorical question, as we approach infinitely small photosites (excluding read noise) do not we not approach the equivalence of an analog sensor?

Lastly do the circles of confusion when infinitely evaluated look like the specular highlight bokeh circles for each lense? 

Its early and I am still enjoying the excessive consumption of fermented grain and fruit juices.

canon rumors FORUM

Re: Say What? The 5DS, f/11 and Confusing Circles
« Reply #30 on: May 02, 2015, 08:44:45 AM »

AlanF

  • 1D X
  • *******
  • Posts: 1392
Re: Say What? The 5DS, f/11 and Confusing Circles
« Reply #31 on: May 03, 2015, 07:18:13 AM »
The concern over the effects of diffraction for high resolution sensors is completely misplaced.

You don't lose sharpness to diffraction by increasing pixel density, because the size of the diffraction effect is strictly determined by the lens.  To understand why, suppose you have two camera systems that are identical in every respect except that one has twice the linear pixel density than the other (i.e., every pixel in the low-resolution sensor is split into four pixels in a 2x2 arrangement in the high-resolution sensor).  Ignoring the effect this has on noise (and noise on perceived resolution), it is true that, as an increasing function of f-number, the higher resolution sensor will be able to reveal the effect of diffraction sooner than the low-resolution sensor.  But the reason for this is because the low-resolution sensor is unable to resolve that effect, not that the effect is stronger in the high-resolution sensor.  The Airy disks are IDENTICAL in the two systems because the lens is identical.

Therefore, increasing sensor resolution does not confer any disadvantage with respect to diffraction.  You always have something to gain, and you never do any worse than the low-resolution sensor.  You might not gain as much as you theoretically could (i.e., a high-resolution sensor might not realize the full sharpness in the sharpest plane of focus at f/16 compared to when it is shot at with a near-ideal lens at f/2.8 ), but you won't do worse than a low-resolution sensor that couldn't SEE the diffraction at f/16 in the first place.

The hesitation to go with higher resolution because of fears of diffraction reveals a complete misunderstanding of the phenomenon.  If you said "I don't want high resolution because I want better dynamic range," then I can be on board with that statement.  But if you said "I don't want high resolution because I would be more severely diffraction-limited," I would tell you that you don't understand what you're talking about.

I think it is a concern/hesitation due to "diminishing returns" by using higher sensor resolutions - as I
read the comments here.

With a 5D classic I see losses of detail from f/8 on while doing macro shots at 1:3 (I think it
is actually comparable to f/11). From that I extrapolate that the 5Ds can be used at a
max aperture of f/5.6 to use the lenses capabilities in similar situations. The lens used was
the near ideal EF 2.8 100mm Macro (non-L).
To use the full potential of the 5Ds you need stacking at least for macro but I think it will
be necessary for landscape and architecture as well.

Is that a valid extrapolation? Go to the slrgear site where they measure IQ at different apertures for both full frame and crop. For all the good lenses, there is minimal diffraction degradation at f/8, a tiny amount at f/11, and then more significant at f/16. It is exactly the same for FF and the 7D with 5Ds size pixels.
5D III, 7D II, EOS-M, Powershot SX50, 300/2.8 II, 1.4xTC III, 2xTC III, 70-200/4 IS, 24-105, 15-85, 100-400 II, Sigma 10-20, EOS-M 18-55, f/2 22.

Busted Knuckles

  • Rebel T5i
  • ****
  • Posts: 102
Re: Say What? The 5DS, f/11 and Confusing Circles
« Reply #32 on: May 03, 2015, 08:11:54 AM »

Is that a valid extrapolation? Go to the slrgear site where they measure IQ at different apertures for both full frame and crop. For all the good lenses, there is minimal diffraction degradation at f/8, a tiny amount at f/11, and then more significant at f/16. It is exactly the same for FF and the 7D with 5Ds size pixels.
[/quote

Roger at lens rental posted as much in his blogs a couple months ago.  An upside down "U" in max resolution w/ f8/f11 sometimes being better/worse than wide open depending on the starting point.  f16 starting to tail off pretty quickly

mb66energy

  • Canon 7D MK II
  • *****
  • Posts: 508
Re: Say What? The 5DS, f/11 and Confusing Circles
« Reply #33 on: May 03, 2015, 08:34:48 AM »

[...]

With a 5D classic I see losses of detail from f/8 on while doing macro shots at 1:3 (I think it
is actually comparable to f/11). From that I extrapolate that the 5Ds can be used at a
max aperture of f/5.6 to use the lenses capabilities in similar situations. The lens used was
the near ideal EF 2.8 100mm Macro (non-L).
To use the full potential of the 5Ds you need stacking at least for macro but I think it will
be necessary for landscape and architecture as well.

Is that a valid extrapolation? Go to the slrgear site where they measure IQ at different apertures for both full frame and crop. For all the good lenses, there is minimal diffraction degradation at f/8, a tiny amount at f/11, and then more significant at f/16. It is exactly the same for FF and the 7D with 5Ds size pixels.

I think it is a valid extrapolation. Look at the plot of the resolution @ photozone
http://www.photozone.de/canon_eos_ff/508-zeiss100f2eosff?start=1
http://www.photozone.de/canon_eos_ff/847-canon35f2isff?start=1
http://www.photozone.de/canon_eos_ff/430-canon_135_2_5d?start=1
and
http://www.photozone.de/canon_eos_ff/489-canon_100_28_5d?start=1
The sharpness loss with higher f-numbers is less dramatic if you take the suppressed zero into account.

or compare the 2.8 100 L macro @ f/4 and f/16 with the 7D ii (you are right to use that camera as a good extrapolation basis for 5Ds):
http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=674&Camera=963&Sample=0&FLI=0&API=2&LensComp=674&CameraComp=963&SampleComp=0&FLIComp=0&APIComp=6

Less pronounced with the 1Ds iii:
http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=674&Camera=453&Sample=0&FLI=0&API=2&LensComp=674&CameraComp=453&SampleComp=0&FLIComp=0&APIComp=6


With very good lenses you can spot the decrease of center sharpness from f/4 on on a 20 MPix sensor consistently (in the graphs of photozone). The corner quality increases usually while closing one or two f-stops due to the minimization of lens errors.
With better sensors you will get more detailed images but if you stop down to f/16 you will loose a lot of this gain and I think it will start with f/5.6 or f/8 if "detailedness" is a strong concern.

And never underestimate the sensitivity of our eye-brain system which detects a 5% or 10% loss in detail easily - I do it very often too :)

But I am open to update my understanding while the new data is coming in!
« Last Edit: May 03, 2015, 08:38:50 AM by mb66energy »
Most used tools: 2 x EOS 5D i & 2.8 24 + 2.8/40 | 2.8 100 M + 2.0/100 ||| EOS M & 22 + Ef-S 60

canon rumors FORUM

Re: Say What? The 5DS, f/11 and Confusing Circles
« Reply #33 on: May 03, 2015, 08:34:48 AM »