Full Frame DoF vs Crop Frame DoF

ivarivar Registered Users Posts: 8,395 Major grins
edited September 16, 2009 in Cameras
This thread has some posts about the difference in Depth of Field from a Crop Frame camera versus that of a Full Frame Camera.

I've always known that a FF camera has a relatively shallower DoF, but was wondering if my reasoning why that is so, is correct.

Here is what is in my mind:

Take two exact same setups, one Crop frame, one Full frame. Everything else is the same.
Focus is not straightlined, but spherical.
Result, the distance between the focus and the sensor is greater on the edge of FF sensors as opposed to Crop sensors, creating OOF (blur).

* removed the image to not confuse other people who do understand :D

It's written down in a kind of simple way, and I hope it makes sense. But can someone tell me either if this reasoning is correct, or if i'm missing something?

Comments

  • Tim KamppinenTim Kamppinen Registered Users Posts: 816 Major grins
    edited September 14, 2009
    I don't think that's right at all. I've never heard of the focal plane being spherical; after all, it's a plane, right? Maybe it is, but anyway I don't think that it has anything to do with the DOF difference.

    The difference only exists when taking photographs with the same framing. If you take a headshot with a 50mm on full frame, and then take the same photo with a 50mm on a crop sensor, the full frame image will have less depth of field. Why? Because you have to stand closer to the subject to achieve the same framing. Shorter focal distances result in shallower DOF.
  • QarikQarik Registered Users Posts: 4,959 Major grins
    edited September 14, 2009
    I don't really get your diagram. Tim is correct though.
    D700, D600
    14-24 24-70 70-200mm (vr2)
    85 and 50 1.4
    45 PC and sb910 x2
    http://www.danielkimphotography.com
  • ivarivar Registered Users Posts: 8,395 Major grins
    edited September 14, 2009
    Okay, so maybe I'm skipping a few steps that I have in my mind.

    Let me explain the 'spherical' focus thing, at least how I have it in my head, and see where I go wrong.

    A lens focusses on a certain distance. Let's say you focus at 10 feet. If you take 10 people and all position them at exactly 10 feet from the camera to have them in focus, next to each other, they will not form a straight line, but the line they will be on will be curved. Right?
  • NikolaiNikolai Registered Users Posts: 19,035 Major grins
    edited September 14, 2009
    ivar wrote:
    Okay, so maybe I'm skipping a few steps that I have in my mind.

    Let me explain the 'spherical' focus thing, at least how I have it in my head, and see where I go wrong.

    A lens focusses on a certain distance. Let's say you focus at 10 feet. If you take 10 people and all position them at exactly 10 feet from the camera to have them in focus, next to each other, they will not form a straight line, but the line they will be on will be curved. Right?
    Dude, you need to freshen up your optics 101. mwink.gif
    If your film/sensor were not a plane, but a sphere, you'd be correct. But since it's a plane - well, the focal plane will be a plane, too. deal.gif

    So, if you position 10 people on an equal distance from the camera lens and focu on a middle one, most of them (on a side) will be OOF.
    If, however, you position 10 (or 100) on a line parallel to the sensor/film (I'm not talking TS lnes here) plane and focus on any one of them - they ALL will be in focus.

    Sorry, that's how things work. ne_nau.gif
    "May the f/stop be with you!"
  • Manfr3dManfr3d Registered Users Posts: 2,008 Major grins
    edited September 15, 2009
    “To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
    ― Edward Weston
  • ivarivar Registered Users Posts: 8,395 Major grins
    edited September 15, 2009
    Okay, so I obviously have no idea what I'm talking about lol3.gif
  • ivarivar Registered Users Posts: 8,395 Major grins
    edited September 15, 2009
    Manfr3d wrote:
    Thanks for that, that actually made some sense (parts of it).

    So, let me see if get this correct, now.

    The size of the sensor doesn't have a direct influence on the DoF, but it's because for the same photo a different focal length has to be used, and that creates the difference in DoF? Am i getting this correct now?
  • ziggy53ziggy53 Super Moderators Posts: 24,132 moderator
    edited September 15, 2009
    ivar wrote:
    Thanks for that, that actually made some sense (parts of it).

    So, let me see if get this correct, now.

    The size of the sensor doesn't have a direct influence on the DoF, but it's because for the same photo a different focal length has to be used, and that creates the difference in DoF? Am i getting this correct now?

    Almost right. It is not exactly the focal length per se. It is more the angle of view that is required and then the relationship of focal length and projected image size (along with aperture setting) that determine the DOF.

    Of course, if you could change both the camera to subject "and" subject to background distances, you could preserve a very similar out-of-focus (OOF) in the background despite the image projection size/focal length relationship.

    Remember too that a background that has a lot of contrast, either color contrast or luminance contrast, will seem "busier" and require more bokeh than a low contrast background. Similarly, curved diaphragm blades will effect a smoother bokeh, realizing a smoother OOF background than a straight bladed diaphragm, yielding better bokeh.
    ziggy53
    Moderator of the Cameras and Accessories forums
  • Scott_QuierScott_Quier Registered Users Posts: 6,524 Major grins
    edited September 15, 2009
    ivar wrote:
    Okay, so maybe I'm skipping a few steps that I have in my mind.

    Let me explain the 'spherical' focus thing, at least how I have it in my head, and see where I go wrong.

    A lens focusses on a certain distance. Let's say you focus at 10 feet. If you take 10 people and all position them at exactly 10 feet from the camera to have them in focus, next to each other, they will not form a straight line, but the line they will be on will be curved. Right?
    An uncorrected, single element, lens will behave in the manner you have described. However, the "modern" lenses mounted to our cameras are corrected in any number of different ways to achieve something very, very close to a plane of critical focus at any distance from the recording media (sensor, film, etc).

    Possibly one of the reasons why a lens will/can be soft in the corners is related, at least in part, to the extent to which a lens fails to achieve a plane of critical focus? ne_nau.gif
  • NikolaiNikolai Registered Users Posts: 19,035 Major grins
    edited September 15, 2009
    Fellas, don't get me wrong, but even on a single basic spherical lens a plane is a plane is a plane.... ne_nau.gif
    http://en.wikipedia.org/wiki/Lens_(optics)

    Multiple lenses in todays lens assemblies provide for other effects (magnification for once), but, again, since the sensor is a plane, so is the focal surface :-)
    "May the f/stop be with you!"
  • James HJames H Registered Users Posts: 92 Big grins
    edited September 16, 2009
    Nikolai wrote:
    Fellas, don't get me wrong, but even on a single basic spherical lens a plane is a plane is a plane.... ne_nau.gif
    http://en.wikipedia.org/wiki/Lens_(optics)

    Multiple lenses in todays lens assemblies provide for other effects (magnification for once), but, again, since the sensor is a plane, so is the focal surface :-)
    Yes, a plane is a plane, except when it's a Petzval surface.

    Now I won't claim to know a lot about 'optics 101' or camera lens design, but the OP's original thought and accompanying graphic appear to have at least some basis in reality. His logic seemed just plausible enough to me that I decided to do some reading and it appears his idea is not so far off the mark. Apparently all lenses have some inherent aberrations which cause a curved field to be the point of critical focus, and not a perfectly flat plane. Manufacturers correct for these aberrations as best as they can but the aberrations are still a fundamental part of lens optics.
    Some interesting reading:
    Field Curvature
    Focus Accuracy
    Astigmatism and Field Curvature
    Petzval Surface
    Joseph Petzval

    Like I said earlier, I'm not an optics guy but, from the reading I've done, I'm not convinced that the OP was entirely wrong with his assumptions. There are certainly other factors involved, and most of those other factors have much more to do with the difference in apparent DOF between a FF sensor and a crop sensor, but I don't think the answer above helps in any way and actually is not totally true either. I'd love for someone with a firm grasp of optics and lens design to weigh in on this, especially if everything I just read and linked to above somehow doesn't matter. More links to better/more in depth info would be welcomed as well.
    --James

    Homepage - James Hill Photography

  • craig_dcraig_d Registered Users Posts: 911 Major grins
    edited September 16, 2009
    James H wrote:
    Like I said earlier, I'm not an optics guy but, from the reading I've done, I'm not convinced that the OP was entirely wrong with his assumptions. There are certainly other factors involved, and most of those other factors have much more to do with the difference in apparent DOF between a FF sensor and a crop sensor, but I don't think the answer above helps in any way and actually is not totally true either. I'd love for someone with a firm grasp of optics and lens design to weigh in on this, especially if everything I just read and linked to above somehow doesn't matter. More links to better/more in depth info would be welcomed as well.

    Well, I'm not a master of optics, but I do know enough to address this issue.

    Look at it this way: If we are talking only about a difference in DOF related to crop-frame vs. full-frame sensors, then the sensor itself must somehow affect DOF. And this is just not true. DOF is a product of the lens: its focal length, aperture, and the distance it is focused to. This, along with a more or less arbitrary value for the CoC size, is all you need to know to calculate DOF. The equations for hyperfocal distance and DOF do not involve the sensor size at all. And if you think about it, it seems really weird that somehow a change of sensor size could change the way light behaves inside the lens.

    This is one of those areas where an incorrect, non-technical description of what seems to be happening leads people into confusion. People say that a larger sensor has shallower DOF because that seems to be the result when full-frame and cropped cameras take the same picture (i.e. same framing). But it's not the sensor that is changing DOF; it's the fact that you have to move the camera or change the focal length to achieve the same framing.

    Another thing that can confuse people in this area is the face that some people (and some online calculators) like to use different CoC values for differently-sized sensors, and the CoC value influences the calculated DOF. Thie problem is that the CoC can be chosen more or less at will based on the intended usage of the resulting images. Some writers try to simplify their presentation by making arbitrary assumptions, such as that images from APS-C sensors will have to be blown up more than full-frame images (which is not necessarily true). Depending on what assumptions you make, it may or may not make sense to use a smaller CoC value for APS-C sensors.
    http://craigd.smugmug.com

    Got bored with digital and went back to film.
  • ziggy53ziggy53 Super Moderators Posts: 24,132 moderator
    edited September 16, 2009
    craig_d wrote:
    Well, I'm not a master of optics, but I do know enough to address this issue.

    Look at it this way: If we are talking only about a difference in DOF related to crop-frame vs. full-frame sensors, then the sensor itself must somehow affect DOF. And this is just not true. DOF is a product of the lens: its focal length, aperture, and the distance it is focused to. This, along with a more or less arbitrary value for the CoC size, is all you need to know to calculate DOF. The equations for hyperfocal distance and DOF do not involve the sensor size at all. And if you think about it, it seems really weird that somehow a change of sensor size could change the way light behaves inside the lens.

    This is one of those areas where an incorrect, non-technical description of what seems to be happening leads people into confusion. People say that a larger sensor has shallower DOF because that seems to be the result when full-frame and cropped cameras take the same picture (i.e. same framing). But it's not the sensor that is changing DOF; it's the fact that you have to move the camera or change the focal length to achieve the same framing.

    Another thing that can confuse people in this area is the face that some people (and some online calculators) like to use different CoC values for differently-sized sensors, and the CoC value influences the calculated DOF. Thie problem is that the CoC can be chosen more or less at will based on the intended usage of the resulting images. Some writers try to simplify their presentation by making arbitrary assumptions, such as that images from APS-C sensors will have to be blown up more than full-frame images (which is not necessarily true). Depending on what assumptions you make, it may or may not make sense to use a smaller CoC value for APS-C sensors.

    Craig_d, welcome the Digital Grin and welcome to the discussion. clap.gif

    Just to continue with your premise, I suggest that even the presentation format, panoramic format for instance, can affect the required/desired CoC to achieve a particular effect, taking into account the different presentation sizes possible.

    This is why it is most important for people to test and understand the capabilities of their own equipment, moreso than online calculators and other peoples' recommendations.
    ziggy53
    Moderator of the Cameras and Accessories forums
  • QarikQarik Registered Users Posts: 4,959 Major grins
    edited September 16, 2009
    craig_d wrote:

    This is one of those areas where an incorrect, non-technical description of what seems to be happening leads people into confusion. People say that a larger sensor has shallower DOF because that seems to be the result when full-frame and cropped cameras take the same picture (i.e. same framing). But it's not the sensor that is changing DOF; it's the fact that you have to move the camera or change the focal length to achieve the same framing.

    You are correct..it is simply a colloquial reference to say that full frame sensor has shallower DOF to avoid having to go through optics 101 lecture to explain the same thing.
    D700, D600
    14-24 24-70 70-200mm (vr2)
    85 and 50 1.4
    45 PC and sb910 x2
    http://www.danielkimphotography.com
  • BGtomBGtom Registered Users Posts: 42 Big grins
    edited September 16, 2009
    Could be simpler...
    Oi! So obtuse...I click shutter...pretty pikture happen. :D

    Let me take a stab, although I think the second poster said it best; for the differently sized sensors to take a similarly framed picture with an equivalently sized lens, the crop camera must move back (farther away from the subject). This, of course, now changes the DoF for the CF camera (DoF gets larger). So it is "true" that FF has shallower DoF than CF...sort of. But not really: they are, in fact, exactly the same.

    Because FF and CF can use the same lenses, the distance from the back of the lens to the sensor is the same, so the focus is the same. It's just that the CF doesn't use all of the image focused upon it; the area beyond the edges of the sensor (crop) is wasted. It is exactly like taking a picture with a FF camera, printing it as an 8x10 and then cutting a 4x6 out of the middle. Hence, Crop Frame.

    As to the curved focal plane, I had to re-read that one...If you array 10 peeps in a line parallel to the focal plane, the guys on the end may be out of focus because the distance to them is greater than the guy in the middle (triangle, yah?). So inorder to maintain the focus, you need to arrange your peeps in a curve (or stop down).

    I know I have added nothing that was not already said more precisly by those more knowlegeable (kudos...), so I will return now to my pikture fotomajigaling.eek7.gif
    Canon 7D! | Baux Pinhole MkIV | 30mm TP Tube Extender | SteadiHand IS system | BiPed 2000 Camera Support | Steely Gaze
  • NikolaiNikolai Registered Users Posts: 19,035 Major grins
    edited September 16, 2009
    OK, guys... Apparently, ages of classic optics and numerous references to special literature are not going to help (reading is hard, special niche reading - even more so:-).

    Here is a visual proof.

    A metal yard stick was placed in front of FF Canon EOS 5DMarkII equipped with TS 24mm (I didn't use any of the T/S stuff, it's simply the widest angle EF glass I currently own) about 18 inches from the front glass element (21" from the focal plane).

    Camera was on a shooting stand at f/3.5 and 1/200sec (lit with Profoto and PW Mini T1), focused manually on the 20" mark in Live View mode (at max zoom, so it's very accurate).

    Here is the result (click on the image for the original)

    651607603_54fvV-XL.jpg

    According to the DofMaster, the DOF in this case is about 4 inches (4.01"). The wooden ruler kinda proves this point - at 2" mark sharpness is getting lost pretty quickly.

    According to Pythagoras's theorem, the distance to the 32" metal mark (which, while being 12" away from the central point, is perfectly in focus) is sqrt(12*12 + 21*21=585) = 24.2" . DOF master gives us far DOF limit as 23.2", meaning 32" mark is one inch beyond the DOF area. While the difference is not that huge, you can see by the wooden ruler what one extra inch means - wooden 3" mark (one inch closer than near DOF limit) is fairly blurred.

    So, according to "put the guys in a circle" crowd, the 32" metal mark should be at least one inch further than the farthest DOF limit, and, as such, rather heavily blurred.

    As you can see with your own eyes - it's not.

    QED.
    "May the f/stop be with you!"
  • QarikQarik Registered Users Posts: 4,959 Major grins
    edited September 16, 2009
    Nikolai, nice data point. I'd like to see that experiment on FF with a 14mm lens at f/2.8 and if you get the same results I will accept that argument for practical purpose of a lens.

    As a thought expereiment..imagine an optically perfect 1mm lens (yeah not going to happen I know). Now put a 100 foot ruler parallel in front of this lens and sensor like you did 6 inches away. The triangle that forms will have 1 leg at 0.5 foot and 1 leg at 50 feet. No way both ends are in focus. (assuming 1mm lens can se the other end of the ruler hehe) As the dimension we are measuring get larger the more the sensor plane looks like a point source instead of a plane.

    Again though for practial purposes within normal camera zoom ranges you may be "close enough".
    D700, D600
    14-24 24-70 70-200mm (vr2)
    85 and 50 1.4
    45 PC and sb910 x2
    http://www.danielkimphotography.com
  • NikolaiNikolai Registered Users Posts: 19,035 Major grins
    edited September 16, 2009
    Qarik wrote:
    Nikolai, nice data point. I'd like to see that experiment on FF with a 14mm lens at f/2.8 and if you get the same results I will accept that argument for practical purpose of a lens.

    As a thought expereiment..imagine an optically perfect 1mm lens (yeah not going to happen I know). Now put a 100 foot ruler parallel in front of this lens and sensor like you did 6 inches away. The triangle that forms will have 1 leg at 0.5 foot and 1 leg at 50 feet. No way both ends are in focus. (assuming 1mm lens can se the other end of the ruler hehe) As the dimension we are measuring get larger the more the sensor plane looks like a point source instead of a plane.

    Again though for practial purposes within normal camera zoom ranges you may be "close enough".

    "No way" is not really a mathematical or physical argument. mwink.gif

    If you're talking abount "mind experiment" you need yo use available optics theory - which nobody seems to like too much here. My background tells me the end of your 100 ft stick will still be in focus in this perfect world. Afterall that is why such short-foucs cameras are typically called focus free :-)

    If you wanna put that to the test - by all means.

    Mind you, super WA lenses are prone to all sort of aberration, and most of them are not even rectilinear ones, so we're getting to the quite a void area of glass impurities and lens assembly design imperfections.

    However, for all the practical purposes of the OP shooting 10 guys I think the results are quite obvious: put your ducks in a row, not in a circle! deal.gifmwink.gif
    "May the f/stop be with you!"
  • craig_dcraig_d Registered Users Posts: 911 Major grins
    edited September 16, 2009
    Qarik wrote:
    As a thought expereiment..imagine an optically perfect 1mm lens (yeah not going to happen I know). Now put a 30 foot ruler parellel in front of this lens and sensor like you did 1 foot away. The triangle that forms will have 1 leg at 1 foot and 1 leg at 15 feet. No way both ends are in focus. As the dimension we are measuring get larger the more the sensor plane looks like a point source instead of a plane.

    Again though for practial purposes withing normal camera zoom ranges you may be "close enough".

    Let's keep in mind that there is a difference between what nature wants to do, what we want it to do, and how close we can get to what we want. A spherical lens does not naturally have a perfectly flat focal plane that extends out to infinity on the sides. Nor does it need to. It need only be reasonably flat out to the limit of its field of view. Some lenses do this better than others. Some lenses do have noticeably curved fields of focus, but this is usually considered a flaw. Then there is also the separate issue that most lenses are not as sharp in the corners as they are in the center. Add in CA, geometric distortion, and flare, and you have several competing requirements that have to be balanced against each other in designing a lens.

    I don't think you've thought through your counter-example very well. Remember that depth of field increases as focal length decreases. A 1mm lens at f/1 would have a hyperfocal length of less than two inches, so yes, if you had even a reasonably good 1mm lens, both ends of the ruler, the center, and everything else in Nikolai's image would be in focus.
    http://craigd.smugmug.com

    Got bored with digital and went back to film.
  • NikolaiNikolai Registered Users Posts: 19,035 Major grins
    edited September 16, 2009
    craig_d wrote:
    Let's keep in mind that there is a difference between what nature wants to do, what we want it to do, and how close we can get to what we want. A spherical lens does not naturally have a perfectly flat focal plane that extends out to infinity on the sides. Nor does it need to. It need only be reasonably flat out to the limit of its field of view. Some lenses do this better than others. Some lenses do have noticeably curved fields of focus, but this is usually considered a flaw. Then there is also the separate issue that most lenses are not as sharp in the corners as they are in the center. Add in CA, geometric distortion, and flare, and you have several competing requirements that have to be balanced against each other in designing a lens.

    I don't think you've thought through your counter-example very well. Remember that depth of field increases as focal length decreases. A 1mm lens at f/1 would have a hyperfocal length of less than two inches, so yes, if you had even a reasonably good 1mm lens, both ends of the ruler, the center, and everything else in Nikolai's image would be in focus.

    +1 15524779-Ti.gifdeal.gifthumb.gif
    "May the f/stop be with you!"
Sign In or Register to comment.