[Cannikin]

So many "theories" and "interpretations" in this thread...

"Depth of Field" is dependent on exactly three things:

1. Aperture diameter, in units of length (e.g. a 50mm f/2 has an aperture of 25mm). Larger aperture diameter decreases DOF, smaller diameter increases DOF.
2. Distance to point of focus. Closer focus decreases DOF, farther focus increases DOF.
3. Final viewing magnification. Larger viewing size decreases perceived DOF, smaller viewing size increases perceived DOF (note the keyword "perceived", as of course changing the viewing magnification does not actually alter an image already taken).

What does this mean? It can be shown through optics and geometry why DOF is dependent on only these three things, but that goes into a looooong discussion. Instead, here's a straightforward example:

If you take a 100mm f/4 and a 50mm f/2 (both with 25mm apertures), focus on the same target at the same distance, and view the final images such that the subjects appear the same visual size (irrespective of total field of view), the DOF will appear to be exactly the same. Assuming you have enough pixels to be able to view the subjects at the same size, the sensors play no part.

It is from these fundamental properties that allow one to mathematically simplify things into a "crop factor". Multiplying the focal length and f-number by the same "crop factor" obtains the same aperture diameter (in millimeters). If the lenses are chosen to obtain the same angle of view on any two formats, they would be viewed at the same final size, giving the same final DOF.

Likewise, "light gathering power" and "resolving power" (the ability to gather light and resolve detail from a particular given subject) is dependent once again on aperture diameter, not sensor size or f-numbers, or whatnot. Ask any astronomer. There is a reason why the iPhone with its f/2.4 aperture (two point four) has nowhere near the light gathering or resolving power of the Hubble Space Telescope with its f/24 (twenty four) aperture for any given subject (no, the answer is not "sensor size"). The iPhone's lens does not have 100 times the light gathering power of the HST, regardless of what people who say "f-number is f-number is f-number" want you to believe. A larger sensor's "low light advantage" and "shallower depth of field" fundamentally come from having a larger aperture diameter for a given angle of view and exposure settings.

F-number by itself tells you only two things: luminous intensity (light intensity per unit area) for uniform incoming light and size of the Airy disk (one aspect of diffraction's effect on resolution) on the sensor plane. By itself it tells you nothing about DOF or real subject light gathering/resolving power.

---

Last edited by Cannikin; 07-15-2015 at 11:26 PM.

[oxidized]

The way I see it,

Put your 50 F1.4 on your APS-C camera. Take a head and shoulders portrait of someone standing right in front of you. While you, keep your position, pick up a full frame camera, slap your 50 1.4 on that camera. Whoa!!! The person now looks further away! I need to step in closer to take the same head and shoulders portrait. By stepping in closer my depth of field shrank. Alternatively, I can take a 75mm F1.4 lens and that will look the same way the 50 1.4 looked on the APS-C. Again, depth of field got more shallow.

End of argument.

[Ian Stuart Forsyth]

It's true you don't need to understand circles of confusion to take great photos, and there are tons of calculators out there to give at least a general idea of how DoF is changed with the many variables. However, for the math minded who have an interest it's still worth looking at. Here's a handy derivation of the equations involved with DoF and, for me at least, working through it was far more insightful than trying to understand explanations from photographers who only have working understandings:

The DOF equations



No need to trade anything for situations where you can keep the iso down with a slower shutter speed + tripod, or by cranking up the lights (turn up the flashes or throw more hydrogen into the sun). I don't consider these to be "serious qualifications" These won't apply to every situation, or even to everyone, but for me this would be most of the time when I'm not being exceedingly lazy.

Be fair or you risk sounding like a propaganda machine in your own right*.


(* and to continue with the policy of fairness, for my own use I'm nowhere near pushing the limits of APS-C, that doesn't mean the rest of the photographic world isn't pushing them, or that they're incapable of making their own rational purchasing decisions)

And let’s be honest
We can shoot FF handheld with a 100mm lens at F8 all the way down to LV 12.6 without cranking up the iso
50mm F8 down to LV 11.6
28mm F7.1 down to LV 11
16mm F6.7 down to LV 10
That covers a lot of the shooting conditions that one would normally see and shoot & without IS
One can easily use a 600mm lens just before sunset at F6.7 with VR & not have the need to bump up the iso
There is also the so important factor that may of the FF cameras have higher resolutions and with that added resolution one can trade that resolution with NR for a cleaner photo.

[philbaum]

Okay, the guy said that the increase in SNR of FF vs APS was due to larger photosites. That actually makes sense, one need to only look at the A7S performance :-)

Then he said that Canon and Nikon have been working on compression algorithms that will provide the same pixels density as crop sensors BUT with the nearly the same SNR performance as the older FF sensors. I guess they are talking about the Canon 50mp sensor and the Sony 42mp sensor. Will be interesting to see the results when DXO tests them.

[Fogel70]

Okay, the guy said that the increase in SNR of FF vs APS was due to larger photosites. That actually makes sense, one need to only look at the A7S performance :-)

Then he said that Canon and Nikon have been working on compression algorithms that will provide the same pixels density as crop sensors BUT with the nearly the same SNR performance as the older FF sensors. I guess they are talking about the Canon 50mp sensor and the Sony 42mp sensor. Will be interesting to see the results when DXO tests them.

It'is a bit of both, pixel size and sensor size can affect the SNR performance, but it also depends on a number of other factors.
A larger pixel capture more light so it gives higher SNR, a larger sensor capture more light so it gives higher SNR. The only difference between them is that you either compare pixel level noise or sensor level noise.

A7S seems to use a high ISO specialized sensor that give advantage on high ISO so I don't think it's only pixel size that give A7s an advantage, but on lower ISO it fall behind in DR compared to sensors with smaller pixels (FI Nikon D810 with 3x pixel count). On base ISO A7s has lower DR than Pentax K-3 with twice as many pixels on less than half the sensor size.

Comparing Pentax K-5IIs with D810 that both have same pixel size it's clear that the larger sensor get SNR improvements that is linear to the difference in sensor size, and when comparing pixel level SNR they are both the same. And when comparing SNR between A7S and D810 they are about the same at sensor level, and both of them are ahead of K-5IIs.
Sony A7S versus Nikon D810 versus Pentax K-5 IIs - Side by side camera comparison - DxOMark

When comparing pixel by pixel a larger pixel has of course higher SNR (as larger pixel capture more light), but when it comes to comparing whole sensors it's about the total light capturing area that matters (the sensor that capture most light get best SNR). Today when sensors start to reach maximum QE and fill factor possible, it does/will not matter much if pixels are small or big as the total amount of light captured by both sensor will be the same as long as the sensors are of same size. Smaller pixel still have a little disadvantage, but with each new generation this disadvantage get smaller, and with BSI sensors pixel size might not matter at all on future sensors.

Sony A7s compared to Nikon D810 at sensor level performance.

[amoringello]

The way I see it,

Put your 50 F1.4 on your APS-C camera. Take a head and shoulders portrait of someone standing right in front of you. While you, keep your position, pick up a full frame camera, slap your 50 1.4 on that camera. Whoa!!! The person now looks further away! I need to step in closer to take the same head and shoulders portrait. By stepping in closer my depth of field shrank. Alternatively, I can take a 75mm F1.4 lens and that will look the same way the 50 1.4 looked on the APS-C. Again, depth of field got more shallow.

End of argument.

DOF == Aperture, Focal length and Distance to Subject.
WHY WHY WHY do people continue to change a defining attribute of DOF, such as distance, and explain it away as the sensor being the culprit??

This is a prime example of the pure FAIL in these arguments when people do not fully comprehend what it is that they are talking about.

[ElJamoquio]

I hate to keep harping on this, but the work of the FF propagandsits has been insidious and relentless. It would seem 90% of the photographic world actually believe you get something for nothing when you buy a larger format camera. What you get is the ability to trade DOF for noise shooting wide open, if you want that.

And cheaper, smaller, lighter lenses if you don't. And the ability to purchase less lenses, because you can always crop.

And hey, FF cameras are now $1k.

There's really not a lot of reason for a non-pocket-able camera to have a sensor smaller than 36x24 in my book. That doesn't make older APS-C cameras bad... but technology marches on. My K200D and my K-5 have sentimental places in my heart. They made some great pictures. I wish I had a FF back then, but I didn't.

I remember when 'FF propagandists' were arguing that you can get the same SNR and DOF as APS-C cameras. Apparently acceptance of change is a long process.

---------- Post added 07-16-15 at 06:20 AM ----------

One can easily use a 600mm lens just before sunset at F6.7 with VR & not have the need to bump up the iso

Two flaws in practice in my book -

1) The pictures I'm taking with a 600mm lens are generally after sunset. the better pics are right before sunset, true, but that owl gets sleepy.

2) At 600mm, a lot of the time what I'm taking a picture of is moving. So 1/200 or whatever - although sharp for non-moving objects - isn't acceptable anyway.

[normhead]

were arguing that you can get the same SNR and DOF as APS-C cameras.

Well no, they were arguing that you got a stop better. IN fact the propagandists had to be arguing they got better of this that or the other or they wouldn't have been propagandists would they?

And cheaper, smaller, lighter lenses if you don't. And the ability to purchase less lenses, because you can always crop.

You can get higher pixel densities in APS-c. As long as that is true you can't really match APS-c with FF in some situations. Cropping FF to 15 MP when everyone is shooting 24 MP APS-c is a really half assed solution.

[kh1234567890]

WHY WHY WHY do people continue to change a defining attribute of DOF, such as distance, and explain it away as the sensor being the culprit??

Because otherwise their whole belief system falls apart. The equivalence people have obviously never stitched a panorama (the obvious equivalent of having a bigger sensor) else they would realise that nothing changes except that you get more in the picture.

[ElJamoquio]

Because otherwise their whole belief system falls apart. The equivalence people have obviously never stitched a panorama (the obvious equivalent of having a bigger sensor) else they would realise that nothing changes except that you get more in the picture.

Who are these people that you're insulting? I want to meet one of them.

---------- Post added 07-16-15 at 07:20 AM ----------

Well no, they were arguing that you got a stop better. IN fact the propagandists had to be arguing they got better of this that or the other or they wouldn't have been propagandists would they?

I remember the propagandists saying that APS-C had a stop more DOF, and DOF was good, therefore APS-C is superior.

But maybe we were on different forums.

[normhead]

Because otherwise their whole belief system falls apart. The equivalence people have obviously never stitched a panorama (the obvious equivalent of having a bigger sensor) else they would realise that nothing changes except that you get more in the picture.

Well, yes and no. I'm not sure what the effect of stitching is. Is Stiching two 35mm images together the same as one 17mm image? Because a 17mm lens would give you way more DoF than a 35 mm lens. That's a whole different line of inquiry.

SO say you sticth two 35 mm images together. Then you reduce that image to the same size one 20mm image. You're shrinking you circles of confusion by half, and that should double your DoF. Shrinking the circles of confusion should make more of the image...in acceptable focus.

Clearly the actual DoF which has as part of it's definition acceptable focus, well reducing an image in size definitely increases the percentage of the image in acceptable focus. Over time I've leaned how out of focus an area can be at 6000 x 4000, to know what will look sharp at 1080x 800.

I'm just speculating here. Does anyone actually know how to interpret stitched images in terms of DoF?

I don't know, I just think this could be a lot more complicated than you're making it out to be.

[Fogel70]

Because otherwise their whole belief system falls apart. The equivalence people have obviously never stitched a panorama (the obvious equivalent of having a bigger sensor) else they would realise that nothing changes except that you get more in the picture.

With stitching you get the same advantages as using a larger sensor, just like you get the same advantage by cropping as you get with a smaller sensor.
Fi by stitching the noise level in the image get lowered just like using a larger sensor, and the FOV increase

As always, the more light that is captured for the final image the greater the SNR get. It does not really matter how it is done; longer exposure, larger sensor, stacking, stitching, using flash to add light, or waiting for the sun to shine brighter...

[amoringello]

Well, yes and no. I'm not sure what the effect of stitching is. Is Stiching two 35mm images together the same as one 17mm image? Because a 17mm lens would give you way more DoF than a 35 mm lens. That's a whole different line of inquiry.

SO say you sticth two 35 mm images together. Then you shrink that image to the same size a one image. You're shrining you circles of confusion by half, and that should double your DoF.

I'm just speculating here. Does anyone actually know how to interpret stitched images in terms of DoF?

I don't know, I just think this could be a lot more complicated than you're making it out to be.

DOF again is created by Focal length, Aperture and Distance to subject. Each image will have the same DOF, regardless of how many you stitch together. Just as if you have a 1mm x 1mm sensor or a 1meter by 1meter sensor, the DOF does not change.

Someone will come along and say that printing the larger image, blah blah blah,etc... well yes -- once you get past the sensor itself there are a whole host of variables that come into interpretation of image quality; crop or magnification of image, acceptable sharpness, substrate, DPI, viewing distance, visual acuity of viewer, etc...

Getting away from all the "equivalence" theories and additional myriad of variables required to determine their effects, light coming from the lens does not get altered just because the size of the plane on which it is cast changes.
Of course, if you don't have your nodal point exactly correct you won't only get potentially obscene image deformation but inconsistencies in DOF can occur. Same DOF, they simply may not line up correctly. (although I suspect that would be quite minimal if noticeable at all)

[boriscleto]

Yes, right after we resolve that "chicken or the egg" conundrum.

As chickens are descended from older egg laying species the egg came first...

[normhead]

As chickens are descended from older egg laying species the egg came first...

Are your referring to neolithic pre-chickens?
Cause I'd rather talk about neolithic pre-chcikens than DOF right now.