How (and why) does sensor size change DOF?

[bdunbar79]

I............CAN'T................BELIEVE...............THIS............IS............SO............HARD............TO...........UNDERSTAND

[BozillaNZ]

DoF does not rely solely on optics, so badgerpiper's  statement is false. DoF relies on apparent aperture (optics) and subject magnification (optics, reproduction size and viewing distance).

If you look at same sized prints, as common sense dictates you must, the crop camera capture is enlarged more, so the CoC is smaller, so the DoF is less.

Why doesn't everybody who is inclined to post read the links I have provided? It is all in there. Depending on how you make your comparison, and you have to clearly state the way you want to compare the captures, a smaller sensor can be shown to have more DoF than a ff camera, the same DoF, or as in this instance, less DoF than that ff camera.

You cannot separate DoF from subject magnification and viewing distance at the output size.

Finally, someone who knows what I was talking about! I am not crazy,hahaha!

When you enlarge a picture, CoC gets bigger, DoF gets shallower. Don't believe me? Take a mild shallow DoF photo and down scale it to thumbnail image, see, suddenly everything is in focus! Magical isn't it? 


Note when you crop a photo, you essentially zooming in and stopping down. On the other post we had a very detailed discussion about what this does to the background blur, in short, when you zoom in and stop down, close background blurs less, distant background blurs more, and life is complicated!

http://www.canonrumors.com/forum/index.php?topic=15904.0

[ecka]

DoF does not rely solely on optics, so badgerpiper's  statement is false. DoF relies on apparent aperture (optics) and subject magnification (optics, reproduction size and viewing distance).

If you look at same sized prints, as common sense dictates you must, the crop camera capture is enlarged more, so the CoC is smaller, so the DoF is less.

Why doesn't everybody who is inclined to post read the links I have provided? It is all in there. Depending on how you make your comparison, and you have to clearly state the way you want to compare the captures, a smaller sensor can be shown to have more DoF than a ff camera, the same DoF, or as in this instance, less DoF than that ff camera.

You cannot separate DoF from subject magnification and viewing distance at the output size.

Finally, someone who knows what I was talking about! I am not crazy,hahaha!

When you enlarge a picture, CoC gets bigger, DoF gets shallower. Don't believe me? Take a mild shallow DoF photo and down scale it to thumbnail image, see, suddenly everything is in focus! Magical isn't it? 


Note when you crop a photo, you essentially zooming in and stopping down. On the other post we had a very detailed discussion about what this does to the background blur, in short, when you zoom in and stop down, close background blurs less, distant background blurs more, and life is complicated!

http://www.canonrumors.com/forum/index.php?topic=15904.0

That's completely insane .
Imagine that you have a magical shrinking machine. You print the photo say 30x20, you shrink it 100 times and it looks super sharp to you. Now you shrink yourself 100 times too and see that the picture didn't change. The problem is that your printer and monitor do not have an infinite number of pixels to show you that, and even if they did, then your eyes wouldn't see it, because they would be too small. Different print sizes that include the destructive reduction of resolution dots isn't doing any magic, you are shrinking the image, but the dots stay the same size. CoC is about the perception, not about magical information transformation. There's no such thing as magic. I say - you're both mad.

[J.R.]

DoF does not rely solely on optics, so badgerpiper's  statement is false. DoF relies on apparent aperture (optics) and subject magnification (optics, reproduction size and viewing distance).

If you look at same sized prints, as common sense dictates you must, the crop camera capture is enlarged more, so the CoC is smaller, so the DoF is less.

Why doesn't everybody who is inclined to post read the links I have provided? It is all in there. Depending on how you make your comparison, and you have to clearly state the way you want to compare the captures, a smaller sensor can be shown to have more DoF than a ff camera, the same DoF, or as in this instance, less DoF than that ff camera.

You cannot separate DoF from subject magnification and viewing distance at the output size.

Finally, someone who knows what I was talking about! I am not crazy,hahaha!

When you enlarge a picture, CoC gets bigger, DoF gets shallower. Don't believe me? Take a mild shallow DoF photo and down scale it to thumbnail image, see, suddenly everything is in focus! Magical isn't it? 


Note when you crop a photo, you essentially zooming in and stopping down. On the other post we had a very detailed discussion about what this does to the background blur, in short, when you zoom in and stop down, close background blurs less, distant background blurs more, and life is complicated!

http://www.canonrumors.com/forum/index.php?topic=15904.0

That's completely insane .
Imagine that you have a magical shrinking machine. You print the photo say 30x20, you shrink it 100 times and it looks super sharp to you. Now you shrink yourself 100 times too and see that the picture didn't change. The problem is that your printer and monitor do not have an infinite number of pixels to show you that, and even if they did, then your eyes wouldn't see it, because they would be too small. Different print sizes that include the destructive reduction of resolution dots isn't doing any magic, you are shrinking the image, but the dots stay the same size. CoC is about the perception, not about magical information transformation. There's no such thing as magic. I say - you're both mad.

[BozillaNZ]

Prepared to shrink yourself 100 times and tell me that I am mad, LOL! I'm laughing and crying at the same time!

Even you have confirm that shrinking picture increase DoF do have a real world implication, BUT

Shrink a picture 100 times? Yes! Shrink a human 100 times? OMGWTFBBQChickenWings!

In the end, isn't photography all about perception? 

[ecka]

Prepared to shrink yourself 100 times and tell me that I am mad, LOL! I'm laughing and crying at the same time!

Even you have confirm that shrinking picture increase DoF do have a real world implication, BUT

Shrink a picture 100 times? Yes! Shrink a human 100 times? OMGWTFBBQChickenWings!

In the end, isn't photography all about perception? 

I'm glad that you are enjoying the discussion. It wasn't my intent to offend anyone.
I didn't confirm the nonsense. Shrinking the picture simply makes it's details imperceptible to you. If you can't see a bacteria, it doesn't mean that there are none. In case with a magic shrinking machine, the details are made smaller so you can still see them by shrinking yourself or maybe using a microscope. However, when you shrink the image on your screen or print a thumbnail, you are just losing the information. Just like for a half-blind person all your images can look same "sharp" or same "blurry". In fact, for him, sharp and blur looks the same. CoC is about perception. DoF is not, it is about information, same as photography. Once the light of an optical image hits the sensor, it is gone, all that's left is the information gathered by the electronics. If you shoot a picture that has nothing in focus, it doesn't matter to what resolution you downsize it, no new information will occur (except the false one). You can manipulate the image in any way you want, but in relativity to reality DoF won't change a bit. If photography is just a form of art for you and perception is the only thing that matters, then perhaps you are not even trying to understand what I'm talking about.

[BozillaNZ]

I'm glad that you are enjoying the discussion. It wasn't my intent to offend anyone.
I didn't confirm the nonsense. Shrinking the picture simply makes it's details imperceptible to you. If you can't see a bacteria, it doesn't mean that there are none. In case with a magic shrinking machine, the details are made smaller so you can still see them by shrinking yourself or maybe using a microscope. However, when you shrink the image on your screen or print a thumbnail, you are just losing the information. Just like for a half-blind person all your images can look same "sharp" or same "blurry". In fact, for him, sharp and blur looks the same. CoC is about perception. DoF is not, it is about information, same as photography. Once the light of an optical image hits the sensor, it is gone, all that's left is the information gathered by the electronics. If you shoot a picture that has nothing in focus, it doesn't matter to what resolution you downsize it, no new information will occur (except the false one). You can manipulate the image in any way you want, but in relativity to reality DoF won't change a bit. If photography is just a form of art for you and perception is the only thing that matters, then perhaps you are not even trying to understand what I'm talking about.

Well by your standard, in reality there is nothing really 'in focus'. The focus 'plane' is a hypothetical thing that has zero thickness. Also on the 'true' focus plane every light point has diameter of 0. Anything in front, or behind this zero thickness hypothetical plane is deemed out of focus because they have a CoC > absolute 0.

The sensor sees something in focus not because they are in focus, but simply because the CoC is smaller than sensor's pixel could distinguish. So what you say? That the image the sensor captures is the real world? It is not.

If above assumption is correct, then take an example, if I shoot a photo with a 320x240 pixel FF sensor, what is my DoF? Even my lens gives a blurry mess I would still get a 320x240 photo that is sharp at pixel level. Does this represent the 'reality'?

The thing is, reality is far weirder than you can ever imaging. We are in a photographic forum, so yes, photography is just a form of art for me and perception is the only thing that matters. I learn from my output photos and prints so I can control my equipment to get the result I want.

Then we leave the underlying physical, electricial or philosophical discussions for some one else or somewhere else.

[ecka]

I'm glad that you are enjoying the discussion. It wasn't my intent to offend anyone.
I didn't confirm the nonsense. Shrinking the picture simply makes it's details imperceptible to you. If you can't see a bacteria, it doesn't mean that there are none. In case with a magic shrinking machine, the details are made smaller so you can still see them by shrinking yourself or maybe using a microscope. However, when you shrink the image on your screen or print a thumbnail, you are just losing the information. Just like for a half-blind person all your images can look same "sharp" or same "blurry". In fact, for him, sharp and blur looks the same. CoC is about perception. DoF is not, it is about information, same as photography. Once the light of an optical image hits the sensor, it is gone, all that's left is the information gathered by the electronics. If you shoot a picture that has nothing in focus, it doesn't matter to what resolution you downsize it, no new information will occur (except the false one). You can manipulate the image in any way you want, but in relativity to reality DoF won't change a bit. If photography is just a form of art for you and perception is the only thing that matters, then perhaps you are not even trying to understand what I'm talking about.

Well by your standard, in reality there is nothing really 'in focus'. The focus 'plane' is a hypothetical thing that has zero thickness. Also on the 'true' focus plane every light point has diameter of 0. Anything in front, or behind this zero thickness hypothetical plane is deemed out of focus because they have a CoC > absolute 0.

The sensor sees something in focus not because they are in focus, but simply because the CoC is smaller than sensor's pixel could distinguish. So what you say? That the image the sensor captures is the real world? It is not.

If above assumption is correct, then take an example, if I shoot a photo with a 320x240 pixel FF sensor, what is my DoF? Even my lens gives a blurry mess I would still get a 320x240 photo that is sharp at pixel level. Does this represent the 'reality'?

The thing is, reality is far weirder than you can ever imaging. We are in a photographic forum, so yes, photography is just a form of art for me and perception is the only thing that matters. I learn from my output photos and prints so I can control my equipment to get the result I want.

Then we leave the underlying physical, electricial or philosophical discussions for some one else or somewhere else.

No. In reality there is nothing that is a 0. Zero is not a thing, zero is just a tool in mathematics. Every point has dimensions and it can be represented as an image of at least 1 pixel. When you are viewing ~18mp image on a ~2mp screen, then 1 dot (color) on the screen represents a group of 9 pixels of the image. Sensor does not capture the real world. The projection of an optical image on the sensor is limited by all kinds of information manipulation by the lens (diffraction, aberrations, vignetting, coma, color tint, distortion, flares and CoC). If the 9 combined pixels carry enough information to represent 1 real world dot, then it will be sharp. If not, then it will be blur (or noise). At 1:1 (100%) it is similar, but with much more false color and noise. If you shrank the blur into oblivion and got some kind of real world information, then it only means that you've destroyed all the rest and the whole blurriness carried only this little.

The sensor and electronics "sees" nothing in focus, just color and contrast of the neighboring pixels.320x240 pixel FF sensor cannot mimic human vision. There are artificial eye implants that allow blind people to see the world in just a few hundred pixels and trust me, it's nothing like the real thing. It's a blurry mess and they can only see a letter or a digit in close-up.

[Pi]

No. In reality there is nothing that is a 0. Zero is not a thing, zero is just a tool in mathematics.

You are wrong about DOF but I will let others argue about it. But as a mathematician, I strongly object the statement that zero is not a thing, and that there is noting at zero. How many 200-400 lenses do you own?

Every point has dimensions and it can be represented as an image of at least 1 pixel,

What if you are shooting film?

[ecka]

No. In reality there is nothing that is a 0. Zero is not a thing, zero is just a tool in mathematics.

You are wrong about DOF but I will let others argue about it. But as a mathematician, I strongly object the statement that zero is not a thing, and that there is noting at zero. How many 200-400 lenses do you own?

Every point has dimensions and it can be represented as an image of at least 1 pixel,

What if you are shooting film?

I don't use superteles.
Film has it's minimum dot size that can be captured.

[dougkerr]

First recall that DoF is not a distance that comes solely from optical theory (like the distance to the image for a certain object location); it is a description of the range of object distances over which a certain arbitrarily-adopted criterion of "blurred-ness" is not exceeded, with the camera focused at a certain distance.

Often, for consistency, that criterion is a blur circle whose diameter is a certain fraction of the overall image size (maybe its diagonal size). This relates to a consistent diameter blur circle on an image produced at a consistent size (such as a consistent size print).

That having been said, when we compare the DoF attained for two values of some attribute of the camera (such as sensor size), "all other factors being equal", we must adopt and announce what we mean by  "all other factors being equal".

Here is one set of such that we might adopt:

a.  Focal length such that the field of view is the same in both setups.  (Aha!)

b. Camera focused at the same distance.  (Of course.)

c. Same f-number.  (Of course.)

d. Criterion for "acceptable" blurring the same in terms of blur circle diameter as a fraction of the image size (diagonal will do).

Now, if we are comparing two cameras, with "B" having an image size 2x that of "A":

1.  Under rule a, we must use a lens of twice the focal length in B as in A.

2. Under rule c, our acceptable diameter of blur circle is twice for B as it is for A.

Now, with regard to point 2, that means that we are more tolerant of blur in B than in A. Thus this consideration alone would lead to a greater depth of field for B than A.

This works essentially proportionally to the blur circle diameter criterion, and thus to image size (sensor size).

Remember, depth of field is not an creature of optical theory alone. It is a creature of what amount of blurring we consider "acceptable". If we increase the amount we consider "acceptable", then our focus distance can be more "off" and we still consider the result acceptable.

Now, to point 1. Because of the optical situation involved (and I will not attempt to describe this in detail here), for a greater focal length the "incorrectness" in focus distance to cause a certain diameter blur circle is less. Thus this consideration alone would lead to a lesser depth of field for B than A.

This works essentially proportionally to the square of the focal length, and thus (because of rule a) to the square of the image size (sensor size).

So as we move from A to B, we find that issue 2 gives an increase in the depth of field, and issue 1 gives a larger decrease in the depth of field (because it varies inversely as the square of the image size).

Thus, the overall effect is that (under the rules stated above), for an increase in sensor size we have a net decrease in depth of field.

Quod erat demonstrandum.

Best regards,

Doug

[bdunbar79]

Speaking about sensor size changing the DOF is a bit misleading. Do we keep everything else the same? This is actually impossible. You cannot keep the AOV and the FL the same. If we just swap the bodies, without changing (physically) anything else, smaller sensors have less DOF.

For the same AOV and same f/stop, larger sensors with the corresponding lenses have less DOF because the "physical aperture" is larger. For example, 80/2>50/2. There is no rule saying you must shoot at the same f-stops with both sensors, and if you shoot with equivalents ones (like 80/3.2 vs. 50/2) DOF is the same. That means "faster" lenses for the smaller sensor. They do not always exists, but sometimes, they do. For example, the m43 system has f/2 zooms, eq. to f/4 ones. They also have a f/0.95 prime, eq. to an f/2 one. But they do not have a match for f/1.4 on FF.

Bottom line: it is not the sensor size, it is the lens and the sensor but lenses with less DOF for larger sensors are easier to make, more common, cost less, and resolve more.

Yes you are correct obviously.  But what I was getting at, is if you shoot with a 1Dx with a 50mm lens at f/4, and then you change to a 7D, and don't move:  Obviously the DOF is not the same.  Move back to the same AoV and the DOF is STILL not the same.  Now you have to change aperture.  So those that are saying sensor size has nothing to do with it, well...

Look at the things I had to change, simply because of a different sensor size.  I guess I don't get why this has gone so far.  That is all I am trying to say.

[Meh]

Shrinking the picture simply makes it's details imperceptible to you.

If I may, can I suggest that this one sentence sums up some of the disagreement in this thread.   DoF is, in fact, a concept that is rooted in human visual perception.  DoF is defined as the distance in front of and behind the plane of focus that appears in focus to a human being.  The calculation requires assumptions regarding human visual acuity, print size, and viewing distance.

I believe, others can correct me if I'm wrong, it is also implicitly assumed that the print size and resolution is such that the individual pixels in the print are too small for the viewer to see them at the assumed print size and viewing distance.  If the pixels are visible then the entire image would not appear sharp.  That is why sensor resolution does not appear in the calculation.

So yes,  print size matters and yes, if you print small enough the entire image would "magically" appear sharp.  "Appear" is the operative word in that statement but it is relevant because "appears sharp" is fundamental to the concept of DoF.  If you also shrunk yourself down, your visual acuity would likely also change so in fact DoF would be the same.

And it is a concept.  It is a defined value based on some reasonable assumptions.  DoF is not something that exists independent of human vision and is not a strictly defined measurement like mass, distance, size, etc.

If you're looking for a physically defined parameter, it exists.  That is focus distance.  The distance from the image plane that is precisely in focus (in practical terms it would be maximally in focus because there is no perfect focus).  And there is only one distance that is maximally in focus... every plane in front of and behind the focus plane is less focused.  If human visual acuity was infinite and the resolution of a print was infinite you would be able to see the tiniest difference in sharpness.  But that's not the case, more than just the exact plane of focus appears sharp and we can define the depth in the image that appears sharp... i.e. Depth of Field.

[Marsu42]

DoF does not rely solely on optics, so badgerpiper's  statement is false. DoF relies on apparent aperture (optics) and subject magnification (optics, reproduction size and viewing distance).

Thank you very much that you are so persistent explaining it, simple /me really learned a lot in this thread from you and the links you posted, esp. http://www.josephjamesphotography.com/equivalence/light.htm#dof

I was mislead by the dof calculators that output a precise dof range, I never read the fine print :-> ... but on the other hand I knew that the apparent dof changes drastically with perceived image size (like from camera display -> large lcd and your distance from it) which actually requires you to take multiple shots of the same scene with different apertures to have the dof *you* want for every output situation you have in mind. The CoC explains this and having read about it I don't know how I could have ever thought dof would only depend on optics :-p

[privatebydesign]

DoF does not rely solely on optics, so badgerpiper's  statement is false. DoF relies on apparent aperture (optics) and subject magnification (optics, reproduction size and viewing distance).

Thank you very much that you are so persistent explaining it, simple /me really learned a lot in this thread from you and the links you posted, esp. http://www.josephjamesphotography.com/equivalence/light.htm#dof

I was mislead by the dof calculators that output a precise dof range, I never read the fine print :-> ... but on the other hand I knew that the apparent dof changes drastically with perceived image size (like from camera display -> large lcd and your distance from it) which actually requires you to take multiple shots of the same scene with different apertures to have the dof *you* want for every output situation you have in mind. The CoC explains this and having read about it I don't know how I could have ever thought dof would only depend on optics :-p

You are so so welcome.

For another mind numbing factoid related to dof, we all think DSLR's are WYSIWYG when shooting wide open, however look through a fast lens (>f2) mounted on your camera, then take a picture and compare the dof of the image to the scene through your camera. The image will have much narrower dof, focusing screens have an aperture! It is around f2.8 so when you shoot wider than that you get much shallower dof than you expect. As a follow on to this it is impossible to accurately manually focus an ultra fast lens with our standard focusing screens.