Is large sensor DOF aesthetics real or a myth?

[Guillermo Luijk]

I have no practical use or experience on large formats, but was wondering why is that large formats are regarded as producing a special feeling in their DOF aesthetics that make their images look more 3D than those obtained in smaller formats, even if DOF is kept equal.

I tried to find out if something could explain that lens blurring evolves in front of and behind the focused subject differently according to focal length/aperture used (which are the changing parameters if we are to achieve the same FOV and DOF with different size formats). But according to a simulation I have just done using Max Lyons' DOF Simulator, APS vs FF, the blurring profiles for a given subject distance are almost indistinguisable:

Camera 1:
Sensor: APS crop factor 1,6
Focal length: 50mm
Aperture: 1.8
CoC: 0.019mm
DOF: 2m 77.37cm

Camera 2:
Sensor: FF crop factor 1
Focal length: 80mm
Aperture: 2.89 (the one needed to match DOF)
CoC: 0.0304mm
DOF: 2m 77.5cm


.



Could anyone give some ideas that could explain the different feeling large formats provide regarding DOF? or that is just a myth, and using appropiate focal length and aperture any format can produce virtually the same output?. Maybe the simulation should be done comparing more differently sized senors, or perhaps the differences become relevant at distances not covered by the simulation.

Regards

[theguywitha645d]

I would say you have one assumption that is wrong--working aperture. It is possible to set up two formats to give a similar DoF, but folks choose specific aperture for exposure or to "maximize/minimize" DoF. So if you compare two formats working at the same aperture, say f/2.8 or f/11, you will find there is a change in DoF. So when I setup my camera for the narrowest DoF at f/2.8, then the results of the format are significant.

But there are other things going on. Larger formats tend to have more accutance. Small formats optimize their lenses for resolution which lowers accutance.

I think the problem with your model is it is too simple.

[Sheldon N]

There are a couple things at work, I believe.

One is MTF of the lens. By going to a larger format, the lens is required to output fewer line pairs per mm for any given final image resolution. This means that the smaller format is generally going to be demanding higher resolution out of a lens than a larger format. Since MTF will increase when you lower the required resolution, what you will essentially see is improved contrast and edge acuity in the larger format at the plane of focus (for a given image size).  Simply stated, if the lenses for the smaller format and the larger format are of equal quality, then the larger format will look better because the lens doesn't have to "work as hard' to produce a given image. If the plane of focus looks better relative to the blurred areas, this can enhance the perception of 3D or overall aesthetics.

Secondly, I think there is a subtle advantage to the larger formats as you approach the hyperfocal distance to having shallower depth of field. I think the net effect of this is that you can maintain shallower depth of field at farther distances with a larger format. This is probably more noticeable if you compare significantly different format sizes (ie. 35mm vs 4x5).

In terms of total quantity of background blur, a quick rule of thumb is to look at the physical size of the aperture opening. Larger aperture will generally produce more background blur for any given DOF.

[ejmartin]

There are a couple things at work, I believe.

One is MTF of the lens. By going to a larger format, the lens is required to output fewer line pairs per mm for any given final image resolution. This means that the smaller format is generally going to be demanding higher resolution out of a lens than a larger format. Since MTF will increase when you lower the required resolution, what you will essentially see is improved contrast and edge acuity in the larger format at the plane of focus (for a given image size).  Simply stated, if the lenses for the smaller format and the larger format are of equal quality, then the larger format will look better because the lens doesn't have to "work as hard' to produce a given image. If the plane of focus looks better relative to the blurred areas, this can enhance the perception of 3D or overall aesthetics.

I'm confused.  Higher resolution means MTF is increased at higher spatial frequencies; won't that typically lead to higher MTF across the board, and therefore improved contrast and edge acuity?  Why would lower resolution lead to higher MTF, and therefore higher contrast and acuity?

[amsp]

Unlike most people on this forum I could care less about the science behind it, numbers and charts bore me to death   My experience is based on one thing only, namely 15+ years of working with both small and medium format cameras. To me there's a simple answer, yes there's a big difference. So much so in fact that I only use my Canons nowadays when I'm absolutely forced to. Of course it's not the only reason I prefer medium format, but the "look" is a big part of it.

If you're actually interested in the aesthetics I suggest you get some real-world experience of your own instead of running "simulations", only this way will you know if the difference is obvious to YOU, and if it warrants the investment.

Have fun.

[Sheldon N]

I'm confused.  Higher resolution means MTF is increased at higher spatial frequencies; won't that typically lead to higher MTF across the board, and therefore improved contrast and edge acuity?  Why would lower resolution lead to higher MTF, and therefore higher contrast and acuity?

I'm referring to the resolution at which MTF is measured, not the resolution of the lens itself. If you look at a typical MTF graph, you'll see two sets of lines: a thick line measured at 10 lp/mm and a thin line measured at 30 lp/mm. The thick line is always higher up the chart which indicates higher measured contrast.  The same lens measured at a lower resolution equates to higher contrast output.

So, assume you have two identical lenses that both make 100 lp/mm resolution all the way across the frame. Also assume you are making equal sized 8x10 inch prints. The lens that is used on medium format is only going to need to resolve 20 lp/mm to make the 8x10 print, but the lens that is used on 35mm is going to need to resolve 35 lp/mm (I'm totally making these numbers up, just a hypothetical example). The medium format image will have higher perceived contrast on the print (higher MTF measurement).

I'm terrible at math, these are just general principles.

[Guillermo Luijk]

you have a different magnification. That changes the symmetry of the DOF. Also LF lenses are corrected in an other way.

Could you explain more about the symmetry of DOF changing with magnification? the two cases I presented have different magnification for using different formats. They have the same relative magnification with respect to sensor size. The way LF lenses are designed and corrected seems to me a more likely reason to explain differences in the DOF behaviour. The Rayleigh equations DOF calculators use could not be 100% accurate with certain type of lenses.

Unlike most people on this forum I could care less about the science behind it, numbers and charts bore me to death  My experience is based on one thing only, namely 15+ years of working with both small and medium format cameras. To me there's a simple answer, yes there's a big difference.

That is fine, but I not only want to know if there is a real difference, but find out what's the reason behind it. I could spend 15+ years getting real-world experience and still not being able to explain why.


Regards

[ErikKaffehr]

Hi,

One factor may be that a lens at f/1.8 may be less sharp than a longer lens at f/2.8. I have seen many comparisons between 50/1.4 and 80/2.8 lenses and the 50/1.4 lenses I have seen are all simply awful at full aperture.

Best regards
Erik

I have no practical use or experience on large formats, but was wondering why is that large formats are regarded as producing a special feeling in their DOF aesthetics that make their images look more 3D than those obtained in smaller formats, even if DOF is kept equal.

I tried to find out if something could explain that lens blurring evolves in front of and behind the focused subject differently according to focal length/aperture used (which are the changing parameters if we are to achieve the same FOV and DOF with different size formats). But according to a simulation I have just done using Max Lyons' DOF Simulator, APS vs FF, the blurring profiles for a given subject distance are almost indistinguisable:

Camera 1:
Sensor: APS crop factor 1,6
Focal length: 50mm
Aperture: 1.8
CoC: 0.019mm
DOF: 2m 77.37cm

Camera 2:
Sensor: FF crop factor 1
Focal length: 80mm
Aperture: 2.89 (the one needed to match DOF)
CoC: 0.0304mm
DOF: 2m 77.5cm


.



Could anyone give some ideas that could explain the different feeling large formats provide regarding DOF? or that is just a myth, and using appropiate focal length and aperture any format can produce virtually the same output?. Maybe the simulation should be done comparing more differently sized senors, or perhaps the differences become relevant at distances not covered by the simulation.

Regards

[Guillermo Luijk]

One factor may be that a lens at f/1.8 may be less sharp than a longer lens at f/2.8. I have seen many comparisons between 50/1.4 and 80/2.8 lenses and the 50/1.4 lenses I have seen are all simply awful at full aperture.

I agree with that happening (after all the 0-blurriness of the simulations at the focused plane is ideal, that would be the point of minimum blurriness), but I wonder if that is going to dramatically change a subjective perception such as the '3D aesthetics' of the DOF.

[ErikKaffehr]

Hi,

I have seen a couple of demos by Hasselblad and Phase One based on the idea. It is much easier to correct a lens for f/2.8 than f/1.4. Would you compare 80/5.6 and 50/2.8, it would be a different thing.

Best regards
Erik

I agree with that happening (after all the 0-blurriness of the simulations at the focused plane is ideal, that would be the point of minimum blurriness), but I wonder if that is going to dramatically change a subjective perception such as the '3D aesthetics' of the DOF.

[theguywitha645d]

There is actually nothing wrong with the theories behind optics and image formation. Usually they are broken down so you can understand different elements. In the case of the OP, he has simply taken a very simple concept and applied it to the problem. However, it will not give him an answer as it is not taking into account all of the different factors that are working together. His assumptions of how people use equipment is going to help either.

The funny thing is if the OP thinks it is a myth, the easiest thing for him to do would be to shoot two images with two different formats and simply compare them. Which is actually where he should start. It always helps to identify the problem before trying to solve for it. Analyzing real images will help him identify the criteria he should be looking for.

[BernardLanguillier]

It could be interesting to also consider where a stitched image fits in this picture.

You will typically end up with a virtual sensor area sized between 4x5 and 8x10, meaning that you will use a focal lenght similar also for a given framing.

This could end up explaining the magical quality of stitched images. Larger imaging device.

Cheers,
Bernard

[theguywitha645d]

I have been doing more stitching lately and it is a really interesting question. It really is not the same as going to a larger chip. It is a hard question as well as the projection that is used I imagine would influence the results just as the fact you are using a small sensor system in a big pixel image. It would be interesting to map systemic MTF over these different systems.

[ErikKaffehr]

Hi,

I guess that the OP is not referring to very large formats. The example he gives is APS-C vs full format. Also, he compares actual blur profiles. It is clear that with using a very large format camera at close distance DoF will behave differently, but that difference would be small between say APS-C and full format or full format and MFDB.

Another factor is what the blur looks like. Lenses for small formats have often worse "bokeh" than lenses that may be less well corrected. I got the impression that over/under correcting spherical aberration plays a significant role.

At large apertures most lenses suffer from axial chromatic aberration, also called "color bokeh" this effect is probably not very relevant if the lens is stopped down to f/5.6 or f/8.

Regarding the original question, "Myth or reality?" I guess that the answer is both.

Best regards
Erik

1. I amsp is 100 % correct – if we talk about photography in terms of making pictures and looking at pictures.

2. If we talk about physics/technology I can give you an explanation. I've done this several times on this forum but why not once more?

You make a photograph of a head with 8x10 inch LF camera. Your magnification is almost 1:1 (long side of the format is 25.4 cm and a head also). If you use for the same head a 35mm camera the head on the sensor is only 3.6 cm high. That gives a magnification of something like 7:1 (25.4:3.6).

Why does this change the symmetry of the dof? Think of the extreems: Is the magnification 1:1, then the dof is symmetrical to the plane of focus. If the magnification is infinity:1 and you focus for the hyperfocal distance then the dof is form the close point to infinity. That will say: dof is finite in front of the hyperfocal distance and infinite behind the hyperfocal distance. Everything else is in between. Also this stupid 1/3 to 2/3 rule.

From your posts I see that you come from the technical side. Do you know this books?
Manual of Photography
http://www.amazon.com/Manual-Photography-Ninth-Media/dp/0240515749/ref=sr_1_1?ie=UTF8&qid=1318796371&sr=8-1
or
Applied Photographic Optics
http://www.amazon.com/Applied-Photographic-Optics-HonFBIPP-HonFRPS/dp/0240515404/ref=sr_1_1?s=books&ie=UTF8&qid=1318796410&sr=1-1

BUT: Things are not that easy. In most of the books you find only linear approximations. And your subjects are non linear. Calculations and measurements are made for point sources and high contrast. But you don't photograph perfectly shaped circular dots – perhaps some do. I found an approximation to calculate at which f-stop Lens 1 on format A behaves like Lens 2 on format B. But real world showed that it is a lot more complicated. One more example: For the Hasselblad were 2 different 250 mm /f 4 lenses available. They differed in dof behaviour because one was better color corrected.

REMARK: If you look in photo journals of the 1920ies and 1930ies theses questions were already discussed over and over. And at that period even the lens developers like Rudolf wrote about that.

Back to amsp. He could spent more time to understand what he is doing technically. But perhaps that would make his images worse. That at least happend to me. I hope I recoverd from this calculate-an-image-before-you-take-it-virus.

Best,
Johannes

[BernardLanguillier]

I have been doing more stitching lately and it is a really interesting question. It really is not the same as going to a larger chip. It is a hard question as well as the projection that is used I imagine would influence the results just as the fact you are using a small sensor system in a big pixel image. It would be interesting to map systemic MTF over these different systems.

Well, I am of course speaking of planar projection. If you haven't done the test, you might want to compare a planar stitch shot with a 100mm lense to a wide angle shot.

They will be identical from a geometry standpoint, less the wide angle lense imperfection that will be totally absent from the stitch.

The difference will be in DoF and resolution, which does for sure change the look of an image.

Cheers,
Bernard

[Bart_van_der_Wolf]

I have no practical use or experience on large formats, but was wondering why is that large formats are regarded as producing a special feeling in their DOF aesthetics that make their images look more 3D than those obtained in smaller formats, even if DOF is kept equal.

Hi Guillermo,

It's a complicated subject and at least part of the impressions (the unfounded mythical ones) are not based on reality. But there are differences, the main one being that larger sensor array sizes allow shallower depth of field due to the longer focal length needed to cover the image circle. That longer focal length will produce a higher on sensor magnification of the projected image, and thus the resolution of a given lens is coming from a lower spatial frequency point on its MTF curve.

There is something else going on though. Even in a comparison with matched DOF, by closing down the aperture with increasing sensor dimensions (which changes the shape of the MTF by diffraction limitation, see my earlier post here), defocus also changes the MTF characteristics depending on the F-number. Different f-numbers produce different defocus effects on the MTF (see my earlier post here).

Could anyone give some ideas that could explain the different feeling large formats provide regarding DOF? or that is just a myth, and using appropiate focal length and aperture any format can produce virtually the same output?. Maybe the simulation should be done comparing more differently sized senors, or perhaps the differences become relevant at distances not covered by the simulation.

Different lens designs will make a direct comparison difficult, because the bokeh (quality of defocus) is also important for image quality. So besides MTF differences at different apertures to get similar DOF range, the quality of the defocus is another factor in the subjective appreciation of image quality, what remains is probably unfounded myth.

Cheers,
Bart

[madmanchan]

Guillermo, it depends on what you're comparing. If you're comparing two shots with comparable field of view, then the subject-to-camera distance will vary between the two sensor/film sizes. This affects the balance between the DOF in front of and in back of the plane of focus.

[Doug Peterson]

There are too many variables in any real world application of different formats to answer this question definitively - except by spending significant time using both formats.

Any mathematical or physics based answer will focus on one or two elements of the image-quality chain.

A partial list of the image quality change. Most of these are different between any two cameras, and are very different between cameras of different formats from different manufacturers:
[Lens Hood / Flare > Lens coating > lens > aperture/shutter > body's internal blackness > IR filter > microlenses > AA filter (or lack thereof) > sensor size > sensor pixel type > readout speed > sensor-to-AD-convertor path, A/D convertor (both bit depth and quality) > heat sinking / cooling > raw file compression > black calibration > in camera raw data manipulation > characteristic curve > ICC profile > demosaic algorithm > deconvolution algorithm > noise reduction type > up-res or down-res algorithm > sharpening]

This list doesn't even include the influence any particular kind of gear has on your manner/style/method of shooting. You don't dig a hole the same way if you have a shovel, a backhoe, or a stick of dynamite. You don't shoot images the same way if you have a 4/3rds with a EVF, a medium format body with a digital back,  and an 8x10 camera (those are not meant to be directly analogous to the shovel/backhoe/dynamite - simply illustrating a large range of things that do the same basic task).

The answer in my case has been "yes" (real). The plane of focus to me seems to transition more smoothly from sharp focus to soft out of focus when using a camera which happens to have a larger sensor. Am I using slightly different focal lengths, apertures, distance to the subject, lens types, lighting than I would have otherwise used? I'm sure I do. What specific element of the physics and/or usage of these systems is responsible for the difference? I can name several candidates (most of which are discussed here) but frankly: I do not care. I have the great perk to my job (see signature) that I can shoot a huge variety of <35mm, 35mm, 1.3 MF, 1.1 MF, 1.0 MF, tech cameras, view cameras etc etc with a huge variety of lenses. When I want beautiful shallow DOF I select 1.0 MF every time. The Phase One 150mm D f/2.8 and Hassy 100mm / 2.2 are my current favorites.

One side note about DOF and lenses: I consider it personally very important that lenses have little or no chromatic aberration when shot near wide open (at least for any lens which you expect you will use near wide open in real usage). While clearly delineated chromatic aberration can be dealt with fairly easily in Capture One (and to a lesser extent ACR or LR) areas that are out of focus and contain fuzzy chromatic aberration are much harder to deal with. For images that are technical in nature (product / architecture etc) the green/magenta is just plain "wrong" but even for more creative images I find it a pet peeve and it really annoys me when making a black and white conversion since anytime you raise or lower the amount of green being used in the back and white you'll add halos (black or white) to any object with chromatic aberration.

Anyway, I mostly wanted to say that you're not going to get anywhere at all with this question without spending meaningful time with a larger sensor system doing the kinds of photography you want/need to do. And it' probably not even the question you want to be asking; the specific question of whether large sensors will give you a different DOF aesthetic when all variables are held even is minor compared to the overall question of whether, with a large sensor system you get images you find more compelling, emotionally connected, persuasive, intriguing, challenging, serene, or whatever the heck it is that you want in your images (mixed with whatever economic considerations you need to make and any technical considerations such as usability of the file for the print sizes or level of post-processing you need/want).

Though my expertise is with Medium Format I suspect most of these comments apply to someone with a <35mm moving up to FF sensors. Though perhaps to a slightly lesser extent given that more variables hold the same between a sub-frame and FF canon than between a FF Canon and a MF system.

Doug Peterson (e-mail Me)
__________________

Head of Technical Services, Capture Integration
Phase One Partner of the Year
Leaf, Leica, Cambo, Arca Swiss, Canon, Apple, Profoto, Broncolor, Eizo & More

National: 877.217.9870  |  Cell: 740.707.2183
Newsletter | RSS Feed
Buy Capture One 6 at 10% off

Masters Series Workshop:
New England Landscape - Fall Color (Oct 5-8)

[ondebanks]

I have no practical use or experience on large formats, but was wondering why is that large formats are regarded as producing a special feeling in their DOF aesthetics that make their images look more 3D than those obtained in smaller formats, even if DOF is kept equal.

I tried to find out if something could explain that lens blurring evolves in front of and behind the focused subject differently according to focal length/aperture used (which are the changing parameters if we are to achieve the same FOV and DOF with different size formats). But according to a simulation I have just done using Max Lyons' DOF Simulator, APS vs FF, the blurring profiles for a given subject distance are almost indistinguisable:

Camera 1:
Sensor: APS crop factor 1,6
Focal length: 50mm
Aperture: 1.8
CoC: 0.019mm
DOF: 2m 77.37cm

Camera 2:
Sensor: FF crop factor 1
Focal length: 80mm
Aperture: 2.89 (the one needed to match DOF)
CoC: 0.0304mm
DOF: 2m 77.5cm

The thing I notice is that the simulations are using different circles of confusion. Should they not be the same, say both 0.019mm?

Ray

[Guillermo Luijk]

The funny thing is if the OP thinks it is a myth, the easiest thing for him to do would be to shoot two images with two different formats and simply compare them. Which is actually where he should start. It always helps to identify the problem before trying to solve for it. Analyzing real images will help him identify the criteria he should be looking for.

I don't think it's a myth, I wonder if it is and that is why I opened this thread. I also shot images with my 5D and 350D, but to be honest couldn't experience any enhanced 3D feeling from the former at the same DOF and that is why I opened this thread in this section. I will probably never have a large format camera or back since I am not interested in having one. Right the opposite, moving from FF and APS to 4/3 cameras with smaller lenses I can carry everywhere. Just wanted to find out about the real or myth of existing differences.

The thing I notice is that the simulations are using different circles of confusion. Should they not be the same, say both 0.019mm?

Nope, if we want DOF to be defined with the same level of demand (criteria), the CoC must be proportional to the format size: 0.0304mm / 0.019mm = 1,6

Reading all your answers, I have a feeling the differences in aesthetics of larger formats come from the particular characteristics of the optic designs each format allows. The ability to obtain the same DOF at closer apertures (the curves I plotted, in the real world are probably closer to the ideal 0-blurriness point the larger the format used), lighter lens corrections, lower level of demand in lpmm,... all play in favour of large formats. That will explain an apparently subjective characteristic of large format sizes based just on image quality (sharper images in the focused areas might mean a stronger 3D effect, less corrected lenses might mean a more pleasant transition to the defocused areas and nicer bokeh).

Regards