Question regarding sharpness, circle of confusion and stopping down.

Messages
23,494
Name
Toby
Edit My Images
No
I've always thought that perceived sharpness is due to the circle of confusion, and that the smaller the circle of confusion the sharper the perceived image. I've also known that stopping down a lens generally improves sharpness. However, I've just posted an example of how aperture affects depth of field and the CoC is the same at f2.8 than it is at f11 (0.03mm in the example below). Now I understand that the zone of focus is larger at f11 therefore overall the image will appear sharper as will have less sharpness fall off, but if the CoC is the same at f2.8 as f11 why can I see my images look sharper even at the point of critical sharpness when the lens is stopped down? I thought that it would be due to greater spherical aberrations with wider apertures, but then I thought aberrations (spherical aberrations at least) were related to the CoC.

As you can see, I'm a bit confused. Can anyone clear this up for me please? ;)

Screen Shot 2017-07-25 at 18.06.16.png
 
You've changed the subject distance?
Yes, to get to the Hyperfocal distance. But I would still have thought that critical sharpness would be sharper with a lens that is stopped down?
 
It depends on magnification and that depends on focal distance (ask any macro tog), as far as my limited knowledge goes, I'm sure someone more knowledgable than me will know.

Also, hyperfocal distances put a very movable value on 'acceptable' critical focus, I suspect that your examples fall outside CoC but inside 'acceptable'.
 
I've just tried the calculator again and the CoC doesn't change if you have them both set to the same subject distance, e.g. Both set to 6m the CoC is still 0.03mm. Perhaps the calculator is wonky?
 
The distance between the focus point and the subject is greater in the second example, so I would expect it to be softer than the first example. Working with infinity is odd. The first example, the subject is infinity minus 5.22m away from the focal point. The second example, the subject is infinity minus 1.32m away from the focal point. Compared to infinity, the difference seems a pointless comparison, but it will likely be the reason as both subjects are at the limit of the focus field.
 
The distance between the focus point and the subject is greater in the second example, so I would expect it to be softer than the first example. Working with infinity is odd. The first example, the subject is infinity minus 5.22m away from the focal point. The second example, the subject is infinity minus 1.32m away from the focal point. Compared to infinity, the difference seems a pointless comparison, but it will likely be the reason as both subjects are at the limit of the focus field.
As I said above, if I set both f2.8 and f11 to 6m subject distance the CoC is still 0.03mm.

However, I think I've figured it out. The CoC they've given is not the actual CoC but the reference CoC for FF cameras in order to calculate DOF so that after enlargement the CoC is still within perceived sharpness boundaries.

http://www.dofmaster.com/digital_coc.html
 
As I said above, if I set both f2.8 and f11 to 6m subject distance the CoC is still 0.03mm.

However, I think I've figured it out. The CoC they've given is not the actual CoC but the reference CoC for FF cameras in order to calculate DOF so that after enlargement the CoC is still within perceived sharpness boundaries.

http://www.dofmaster.com/digital_coc.html


Glad to be of service :thinking: :ROFLMAO:

Presumably the CoC changes with FL?
 
Glad to be of service :thinking: :ROFLMAO:

Presumably the CoC changes with FL?
Not on that calculator no. As I said, the 0.03mm is just the CoC they're using as a reference as what is deemed acceptably sharp for a given format. The only time it changes is if you change the format, eg to APS-C which has to have a smaller CoC reference as the image is enlarged more.
 
Circle of confusin is not part of the measurement of quality produced by a lens.

It is a variable standard we chose to calculate the properties of a lens.
It remains the same what ever the aperture is that we set.

If we need to only make 10x8 prints with a lens, our calculations for depth of field will be very different to that, if we are producing 20x16 prints.
We reflect this by chosing a different coc. We do not use coc as a measure of resolving power of the lens.

In the table 0.03 is the coc chosen as the minimum acceptable.
If we want to resolve even finer detail over the depth of field because we want to make giant print, or a very fussy pixel peeper we may choose an even smaller coc.
The calculator would then produce a correspondingly smaller depth of field.

Though that particular calculator is confusing because it does not allow you to set the coc.

The table is also confusing as it talks about subject distance, when it means the distance that the lens is focussed on. Which they also correcty call the hyper focall distance, further down the table.

The subject may of course be any where between the near and far limit.
 
Last edited:
Circle of confusin is not part of the measurement of quality produced by a lens.

It is a variable standard we chose to calculate the properties of a lens.
It remains the same what ever the aperture is that we set.

If we need to only make 10x8 prints with a lens, our calculations for depth of field will be very different to that, if we are producing 20x16 prints.
We reflect this by chosing a different coc. We do not use coc as a measure of resolving power.

In the table give 0.03 is the coc chosen as the minimum acceptable.
If we want to resolve even finer detail over the depth of field because we want to make giant print, or a very fussy pixel peeper we may choose an even smaller coc.
The calculator would then produce a correspondingly smaller depth of field.

Though that particular calculator is confusing because it does not allow you to set the coc.
Thanks. So CoC is a predetermined quantity derived from what we perceive as sharp when printing at certain sizes? When you read about CoC on t'interweb it says that it's due to the lens not creating a pin point focus area of the incoming light and rather a circle, but this is also pretty much the explanation of spherical aberration. It's making more sense now.

So back to why a lens is sharper stopped down, is this mainly due to less bending of light creating less/smaller spherical aberrations?
 
Thanks. So CoC is a predetermined quantity derived from what we perceive as sharp when printing at certain sizes? When you read about CoC on t'interweb it says that it's due to the lens not creating a pin point focus area of the incoming light and rather a circle, but this is also pretty much the explanation of spherical aberration. It's making more sense now.

So back to why a lens is sharper stopped down, is this mainly due to less bending of light creating less/smaller spherical aberrations?

A lens is only sharper as we stop down over a limited number of stops, as aberations are reduced this is normally only two or three stops. At that point diffraction starts to show its ugly head and negates any further improvement, and in fact progressively reduces resolving power. Which is why lenses are said to have a sweet spot.

However stopping down further, continues to increase deph of field, but not resolving power. This duality is always confusing to newcomers.
 
A lens is only sharper as we stop down over a limited number of stops, as aberations are reduced this is normally only two or three stops. At that point diffraction starts to show its ugly head and negates any further improvement, and in fact progressively reduces resolving power. Which is why lenses are said to have a sweet spot.

However stopping down further, continues to increase deph of field, but not resolving power. This duality is always confusing to newcomers.
Thanks (y)
 
Glad to be of service :thinking: :ROFLMAO:

Presumably the CoC changes with FL?

No it does not. Coc is a choice. It is only a standard of percieved sharpness that we set ourselves.
 
A perfect lens using light that doesn't obey the laws of physics will produce an image of a point that is itself a point but only when that image is in focus. In all other cases, you get a circle (roughly). This diagram shows light rays passing through a lens and coming to a point of focus where the lines intersect. As you move to each side of that single point, the lines diverge, and the point becomes a disk. The circle of confusion is the maximum size that that disk can attain and still appear as a point to our eyes. Magnify an image 100 times, and clearly the disk in the image must be smaller than the disk required for a 10 times magnification. View the image from further away (where things look smaller) and the disk can be larger. The circle of confusion is based on a set of assumptions about visual acuity, magnification to produce the image and image viewing distance.

DOF3.jpg

Take the thin lines as indicating the arbitrarily chosen size of the circle of confusion. Different object distances affect the plane of focus (nearer objects are brought to a focus further from the lens) so you can read this as indicating the size of the depth of field, in that the further you can travel from the plane of focus and still see a point the greater the depth of field. The red lines pass only through the middle area of the lens, and the lines are at a shallower angle, which implies that you can travel further from the point of focus before the circle reaches the maximum - hence stopping down increases depth of field.

When we do take the laws of physics into account, diffraction limits resolution. To give a rough idea of how much, the maximum possible resolution of a perfect lens is the f number divided by 1800 when diffraction is allowed for. All sorts of caveats apply to this value - like the circle of confusion there are a lot of assumptions built in to give a nice, easy to use figure - but it gives the flavour.

Edit to add: see post #18 below. The formula I've just made bold is the wrong way round :oops: :$. It should read 1800/f not f/1800.
 

Attachments

  • DOF3.jpg
    DOF3.jpg
    39.3 KB · Views: 8
Last edited:
I've always thought that perceived sharpness is due to the circle of confusion,


And in response to that, I'll add - contrast affects perceived sharpness at least as much as resolution. A lens of lower resolving power but higher contrast will give an image that we perceive as sharper than one of higher resolution and lower contrast. This disconnect between resolution and sharpness was one reason that resolving power in terms of line pairs per millimeter passed out of fashion in favour of MTF which factors in contrast.
 
And in response to that, I'll add - contrast affects perceived sharpness at least as much as resolution. A lens of lower resolving power but higher contrast will give an image that we perceive as sharper than one of higher resolution and lower contrast. This disconnect between resolution and sharpness was one reason that resolving power in terms of line pairs per millimeter passed out of fashion in favour of MTF which factors in contrast.

It is also the reason that in the 50's/ 60's, the best German lenses which were low contrast high resolution, lost out in the battle with the Japanese cheaper but much higher contrast lenses.
To day, lenses are both higher resolution and high contrast, thanks to new designs but more importantly have almost perfect transmission coatings.
For a very long time some people regretted the advent of high contrast lenses, because they were seen as less atmospheric and had less tonality especially for artistic landscapes and portraiture.

Today pixel peeping is king, and even the circle of confusion chosen for depth of field calculations is is thought inadequate, and is usually reduced in size.
As a result people spend good money on software to smooth out complexions, that are so well rendered with all their faults by modern lenses.
The most recent trend, is for camera makers to reduce the effects of diffraction with special compensatory algorithms in the cameras firmware, so as to increase apparent sharpness even further.

Photography is going though a phase where it is in love with ultimate sharpness and at the same time adores perfect bokeh.
 
Last edited:
Oops! :oops: :$

If you read the second sentence of the final paragraph in post #15 above, you'll realise that something is wrong somewhere, because as it stands the formula shows resolution increasing as you stop down - and I just said the exact opposite. For "divide the f number by 1800" read either "divide the f number into 1800" or "divide 1800 by the f number".

So, f/18 gives 100; f/2 gives 900. For a perfect lens.
 
Back
Top