[swanlefitte]

I have done supplemental reading to stay afloat. Now I am on why 4 standard deviations with sunlight and 16 bits is overkill or 8 bits at ISO3200. Yeah I am drowning. My last statistics course was in 1984.

[clackers]

To summarise: Contrary to what f-stoppers purport, DOF is not affected by focal length but by changes to the aperture diameter.

Of course it does, @ClassA, the video makes it clear to you that the distance, the focal length and F-stop are what determines it, not sensor size.

The aperture diameter is just a derivation of Lee's focal length and F-stop.

F-Stoppers are absolutely correct, as the equation below shows. You can see that focal length is much more important than the F number because it's a squared quantity. Even with a slow kit lens, you should be able to get as close as possible to the subject at the long end and produce blur.

Also note the sentences: "Sensor size affects DOF only in that changing the sensor size on a camera requires changing the focal length to get the same picture. It is the change in focal length that then affects the DOF."

It is quite useless of this 'Equivalence' tosh to suggest otherwise!

[clackers]

If you keep everything constant (except sensor size, presumably) then you get two wildly different images which cannot be reasonably compared to each other.

Wrong.

You get *identical* images except for the field of view. One is just the crop of the other.

Here are two screengrabs I did from their video. They used different cameras, rather than the K-5/K-1 example I like to use because it really is the same sensor wafer with the same pixel pitch, same lens and imaging engine.

One is with a Nikon D850 and the other a Panasonic m43. Same focal length, distance, F-stop, just the Nikon's frame is cropped.

You can claim these are 'wildly different', I put 'em up here not for you, but to any non-equivalence cultists reading this thread!

[clackers]

The "Print" tab isn't any less "real".

It's *phony*, @ClassA, stop defending that crap company's methodology!

As I pointed out to you to get the Print screen values they never used their DxO software to resample the RAW files to 8Mp then measured ISO and dynamic range, they did this infantile 'normalization' to the very real values of the Screen tab.

We know by the data I showed you in Post 123 that in both the K-5 and K-1 that at ISO 12800 the noise in their RAW files is identical to within variation - both measure at about 18dB. That's about four pixels out of every thousand being noise.

In real world use, Normhead has shown blind testers can't tell which of his pics posted were full frame or APS-C.

In Image Resource's sample images, the same struggle goes on:

Attached Images

View Picture EXIF

File Date

 

View Picture EXIF

File Date

 

---

Last edited by clackers; 11-21-2019 at 10:52 PM.

[swanlefitte]

Wrong.

You get *identical* images except for the field of view. One is just the crop of the other.

Identical except different. Part is identical and part is as different as existence to non existence. That seems a huge distinction to me.
A 44 kg weight and a 22 kg weight are identical except 1 is a crop of the other. A 400m sprint is identical to a 100m sprint.

[Rondec]

Wrong.

You get *identical* images except for the field of view. One is just the crop of the other.

Here are two screengrabs I did from their video. They used different cameras, rather than the K-5/K-1 example I like to use because it really is the same sensor wafer with the same pixel pitch, same lens and imaging engine.

One is with a Nikon D850 and the other a Panasonic m43. Same focal length, distance, F-stop, just the Nikon's frame is cropped.

You can claim these are 'wildly different', I put 'em up here not for you, but to any non-equivalence cultists reading this thread!

(I think) everyone knows that if you stand way back with a 100mm lens and snap a photo with two camera, one with a large sensor and the other a small sensor and then crop the large sensor camera's image way down you will get the same image. The same is also true if you would frame your image normally for full frame and then also snapped an image with the same lens from the same distance on the smaller sensor. The thing is that while the depth of field may be the same, the images are vastly different.

The point most people make is that you want a particular framing regardless of your format. If you want a portrait on full frame, you don't stand back as though you are using a 200mm lens, you walk forward and frame for your format. And if you want to emulate that with micro four thirds, you either back up or you use a 50mm lens from the same spot. And if you use a 50mm lens and stand in the same spot, you will definitely have more depth of field unless you open your aperture more.

[Ian Stuart Forsyth]

You get *identical* images except for the field of view. One is just the crop of the other.

If we use this line of thought then we could say that the DOF for the ist ds gives us a different DOF than what we see with the K3 when using the same lens at the same ƒ8, if we go about viewing the image by cropping the k3 image to the same pixel count as the ist DS.

In photography DOF is a perceptual property and to determine the DOF you should hold how we are going to view the final image, If we are not going to hold the same viewing parameters and ignore them completely we could say all kinds silly things.

Take for instance I could say that there will be the same DOF when using a 50mm lens at ƒ8 at 8 feet as using the same camera and the same lens ƒ2.8 at the same 8 feet the only difference is that one image is printed much larger they will have the same DOF

While this is true, for how the majority use their cameras this reality is not a way of defining how DOF really works and for how they are going about setting up DOF

The argument of using different final output images as a basis to show that there is no differences DOF is as silly as my argument with my ist DS and the K3 having different DOF because they are being viewed differently at 100%, and that we should have different DOF's for the same settings on both cameras. Would this fly for most photographers?

Or Just like saying that we have the same DOF using the same camera and lens at the same distance while using 2 different ƒ stops. How would this fly with most photographers ?

F-Stoppers are absolutely correct, as the equation below shows.

One overlooked part of the DOF calculations is the enlargement factor but it is there in every DOF calculator. Without it we cannot calc DOF

In their equation they are using the same enlargement factor so of course they will have the same DOF, but that would mean that one of the images will be printed 2.25 larger than the other,.
How is this a useful form of calculating DOF ?
And how practical is this in how most people use their cameras?

[swanlefitte]

Ian, the fstoppers video agrees. It covers perception and image size around 1:35.
Covers pixel density at around 2:05
And covers cropping around 3:45
It glosses over them though and doesn't explain how they are comparing small vs large images.

[Rondec]

Ian, the fstoppers video agrees. It covers perception and image size around 1:35.
Covers pixel density at around 2:05
And covers cropping around 3:45
It glosses over them though and doesn't explain how they are comparing small vs large images.

Sure. The basic point is that if I shoot with a lens and crop it way down I have not changed depth of field. What you can't say is that it is really the same image. There is a reason why Pentax marketed the DA *55 as a portrait lens for APS-C. It is because an 85mm on APS-C would be too long. And if you compare a similarly framed 55mm shot on APS-C and an 85mm shot on full frame, you will have a really similar field of view, which is actually the important starting place for these sorts of discussions.

Edit: I should say that when you crop, you do change depth of field. Typically the depth of field looks more narrow when you crop heavily. But of course the whole image is completely different too.

[swanlefitte]

Sure. The basic point is that if I shoot with a lens and crop it way down I have not changed depth of field. What you can't say is that it is really the same image.

You have changed the image size. If you now change the image size of the crop by enlarging it to equalize, or you change you viewing distance to equalize by moving closer to see it like the bigger image the DOF has changed. DOF is enlargement and viewing distance. Basically a COC of 0.2mm from 25cm depending on who defines it. Enlarge the image and the COC is enlarged or look at it closer than 25cm and you see the COC clearer.

[Class A]

You can see that focal length is much more important than the F number because it's a squared quantity.

If you take the formula
DOF = 2*u^2 * N * c / f^2
and insert the definition of N
N = f/D

you get

DOF = 2*u^2 * c / (f * D)

This suggests that the DOF decreases linearly with focal length (not with the square of the focal length), if we keep D constant (which we must in order to create comparable images).

For two different formats that require two different focal lengths (f1 & f2) to achieve the same framing, we have

DOF1 = 2*u^2 *c / (f1 * D)
DOF2 = 2*u^2 *c / (f2 * D)

with DOF1 being the DOF for the small format and DOF2 being the DOF for the larger format.

However, we must also consider the enlargement factor required for producing an output at a certain size.
For a smaller format, the enlargement factor is larger, precisely by a factor k, the crop factor k = sqrt(larger format area / smaller format area).

This neutralises the "f" differences in the denominator in the above DOF formulas, as the focal length f2 required for the larger format is exactly k * f1, with f1 being the focal length for the smaller format.

In formulas:

Output-referred-DOF1 = DOF1 / ef1
Output-referred-DOF2 = DOF2 / ef2

with ef1 and ef2 being the required enlargement factors for achieving a certain print size respectively.

Now as we have

ef1 = k * ef2

and

f2 = k * f1

we get

Output-referred-DOF1 = 2*u^2 *c / (f1 * D) / (k * ef2) = 2*u^2 *c / ((f1 * D) * (k * ef2))
Output-referred-DOF2 = 2*u^2 *c / (k * f1 * D) / ef2 = 2*u^2 *c / ((k * f1 * D * ef2))

i.e., Output-referred-DOF1 = Output-referred-DOF2.

Do you now see how the focal length has no impact, when changing the format size at the same time (and using equivalent focal lengths)?
Do you also see that we had to keep D (the entrance pupil diameter) constant to achieve the same DOF?
Due to the fact that D needs to be kept constant, we need to change the f-stop (f/D) from f-stop1 (= f1/D) to f-stop2 = (f2/D) when changing the format size. In other words, we need to stop down (using the crop factor k as a multiplier) when using a larger format.

So, yes, taken out of context, technically my statement that focal length does not influence DOF is not correct.
However, I hope it was always clear that the context I was assuming was a change in format size.

Also note the sentences: "Sensor size affects DOF only in that changing the sensor size on a camera requires changing the focal length to get the same picture. It is the change in focal length that then affects the DOF."

It is quite useless of this 'Equivalence' tosh to suggest otherwise!

The "equivalence tosh" agrees that sensor size per se does not change the DOF.
Arguably, that's one of the main messages of the "equivalence tosh".
It agrees that the lens defines the DOF.
It states that by using equivalent lens parameters that the DOF does not change (which it demonstrably does not).

There are various ways in which one can express the influence of lens parameters on DOF. I find considering the subject magnification and the entrance pupil diameter more helpful compared to focal length (the latter is not helpful anymore in macro situations), but these are details.

[Class A]

Wrong.

You get *identical* images except for the field of view. One is just the crop of the other.

How can you say that "one is just the crop of the other"?

When I stated "wildly different images", I was obviously referring to the wildly differing FOV.
If you compensate for the different FOVs by cropping one of them then you are not doing anything else but emulating the cropping effect of the smaller sensor. No wonder that you then get identical images (in ideal conditions).

If you don't crop but actually use a different focal length (in the f-stopper case, two times the focal length of the smaller format) during capture then you need to also adjust the f-stop in order to achieve the same DOF.

[Class A]

In real world use, Normhead has shown blind testers can't tell which of his pics posted were full frame or APS-C.

If Normhead's testers couldn't see the difference between two images where one was taken with an APS-C camera and the other one was taken with an FF camera while using the same (not equivalent) shooting parameters using reasonable viewing conditions and a scene that supports observing DOF then they indeed were "blind" testers.

[kaseki]

Clackers submission included:



I would make one correction to the implications of this formula and the substituted formula given by Class A in a message above: "c," the optics circle of confusion, applies when the focal plane is to be examined with a microscope or other imaging device with much more resolution than the image can supply. Otherwise, we have to account for the readout with a modified value of "c."

In the case of a film camera, the circle of confusion's energy distribution at any longitudinal point within the depth of field has to be convolved with the film's point spread function, which can be derived from its published MTF.

In the case of an imaging array (digital camera focal plane), the energy distribution has to be convolved with the shape of a pixel, averaged over all offsets of the image point and the center of the pixel (out to at least where the next pixel gets more of the energy).

This implies that DOF actually improves as camera resolution (exclusive of the optics) degrades. Recorded IQ, of course, degrades.

[Alex645]

...and from the late great Charles M. Schultz: