Depth of field (DOF) does NOT change with sensor size

Discussion in 'Photography Beginners' Forum' started by donny1963, Nov 18, 2018.

  1. Ysarex

    Ysarex TPF Noob!

    Joined:
    Nov 27, 2011
    Messages:
    6,695
    Likes Received:
    3,171
    Location:
    St. Louis
    Tim, I like the But... this has long been a topic of interest for me. It begs the question, how do we present an image such that the viewer experiences as faithful a rendition as possible to what we experienced when taking the photo. And that begs the question of why a standard or normal lens is normal. It's taken on faith (possibly a mistake in this business) that a normal lens presents a natural perspective. You hear, a normal lens mimics human vision, or the normal lens sees the scene the way you do, etc. Apart from just empirically testing that, (in most cases with the scene photographed no longer available), what's the criteria for assigning "normal" to a lens focal length for a given format? The answer that a lens is normal if it's focal length matches the diagonal measure of the film/sensor never sounded to me like an adequate answer if achieving natural perspective was the goal.

    So this should interest you: Leslie Stroebel was likewise interested in this question and added some additional information for us to consider. It's obviously limited but what he did was measure an average viewing distance that people naturally stood back to view displayed images. Given the freedom in a gallery to view an image from a distance of choice he was able to say that people on average preferred to view an image from a distance equal to twice the long side of the image. So viewing an 8 x 10 print = 20 inches, viewing a 16 x 20 print = 40 inches, etc. That gives us a number to plug into an equation, and if we run that math the answer is that a normal lens as determined by the film/sensor diagonal comes up a little short.

    So by conventional determination a normal lens for a 35mm camera is 43mm (we round up to 50). Using what I'll call the Stroebel method a normal lens for a 35mm is 70mm. I've long had this fantasy that I'd mount a 75mm f/1.4 Summilux on an M4 and live happily ever after.

    Joe


     
  2. markjwyatt

    markjwyatt TPF Noob!

    Joined:
    Apr 24, 2018
    Messages:
    194
    Likes Received:
    126
    Location:
    So. Cal
    I just got this book from Amazon and look forward to digging into it: Photographic Materials and Processes

    As an aside, you stated, "how do we present an image such that the viewer experiences as faithful a rendition as possible to what we experienced when taking the photo", is this the "standard for photography" (I think it is one standard, but maybe not the only one). I proposed a group on Photrio, which was not well received: PHOTRIO.COM

    I also picked up the domain name http://Standards.photography , currently empty, but...
     
  3. Ysarex

    Ysarex TPF Noob!

    Joined:
    Nov 27, 2011
    Messages:
    6,695
    Likes Received:
    3,171
    Location:
    St. Louis
    Leslie Stroebel is the best.

    Not at all, but it is a unique opportunity for photography and I think worth exploring. Just like it's really good that we have lots of different camera types to chose from there is richness in diversity and there are lots of different ways to approach and practice photography.

    Joe

     
  4. john.margetts

    john.margetts TPF Noob!

    Joined:
    Dec 3, 2015
    Messages:
    1,103
    Likes Received:
    275
    Location:
    Lincoln
    I think that I can safely say that none of my photographs are faithful renditions of what I see. I am more interested in conveying smell (usually musty), temperature (usually cold), age (usually ancient) and other non visual aspects of my subject (mediaeval churches). Lenses pay only a small part in this.

    (I do note that Joe says 'experienced' in the quote above rather than saw. I am really commenting on the desire for faithful which I (and many other photographers) do not share)



    Sent from my 8070 using Tapatalk
     
  5. Tim Tucker 2

    Tim Tucker 2 TPF Noob!

    Joined:
    Jun 8, 2017
    Messages:
    333
    Likes Received:
    241
    Thanks, just ordered his book on visual principles to take a look, promises to be interesting.

    Though I don't doubt this....

    I do question this. You can't equate human vision to equations that easily.

    The best illustration I can think of is the 3D pavement art. As you walk around it then you have a fairly consistent view of what you see, until you reach the exact point at which the perspective is designed to create the illusion. At that point your *understanding* or interpretation of the visual cues changes dramatically and you see the 3D effect. It is much the same with images, in that your view of them will be fairly consistent until you stand at a point that coincides with the camera position and a different reality is presented. This different interpretation then takes over as being more consistent with what you understand through experience and therefore the one you brain registers as *the truth*. It is not a gradual process but a more sudden one as your brain is also trying to maintain a *consistent* understanding of what you see. For instance you do not see a world in motion as you drive down the *freeway* but understand a static world and yourself moving through it. It's quite a computation of the absolute reality that your eyes *see*.

    EDIT: For the diagonal I agree, I don't think that it's actually based on anything other than observation and a useful coincidence.

    This is further compounded because we *learn* through experience. As photographers we learn the effects of wide angle lenses and so recognise them more readily than someone who has never seen a photograph. As photographers our interpretation of the image will have a grounding in out experience of learning to see through a camera whereas the non-photo aware eyes will base understanding on reality alone. It is a very real truth in photography that we teach ourselves to view images in terms of our understanding of the photographic process alone.
     
    Last edited: Nov 21, 2018
  6. Ysarex

    Ysarex TPF Noob!

    Joined:
    Nov 27, 2011
    Messages:
    6,695
    Likes Received:
    3,171
    Location:
    St. Louis
    I made no such suggestion -- just that Leslie's observation gives us a hard number. Further below you agree the use of the format diagonal isn't especially meaningful. So otherwise we have nothing but repeated empirical observation. We know as you pointed out that perspective is also dependent on print viewing distance. What is or should that distance be -- a normal distance? If we have an average distance that applies to most viewers we can back calculate the lens focal length required to match the FOV to the print. It is then fair to say that if you view a 16 x 20 inch print from 40 inches you will be in the correct position to experience perspective as it appeared from the camera position using a 70mm lens on a 35mm camera. It's also fair to say that if you stay at that 40 inch distance and are presented a 16 x 20 image taken with a 20mm lens on a 35mm camera you're likely to experience the image as distorted, but not necessarily as it's actually pretty easy to fool your brain about what it's seeing.

    The numbers along with understanding are useful tools.

    Joe

     
  7. Tim Tucker 2

    Tim Tucker 2 TPF Noob!

    Joined:
    Jun 8, 2017
    Messages:
    333
    Likes Received:
    241
    We've gone well off topic here, but it's interesting so I'll throw one more thought into the mix:

    What if perspective only exists in a 2D representation of an object?

    It's not as weird as it at first sounds. It's just as photographers we automatically assume that what we capture is a representation of a 3D object and therefore how we interpret a picture on a 2D plane is the same as how we see and interpret the actual object in 3D.

    What if that wasn't true?

    It is very difficult to demonstrate on a computer screen because everything I show you will be a 2D representation and will therefore be seen as a 2D representation. Here's a little experiment you can do with your hands, with palm facing you hold them in front of your eyes with one at arms length and the other about half way but do it so they don't overlap. They look of a very similar size.

    Now move them together so they overlap. It's called consistency scaling, you understand that both your hands are the same size and so your interpretation of them will try to be of the same size until they overlap. At which point the illusion is destroyed and you see them in relation to each other and the difference in size becomes apparent.

    The camera though will always present a literal recording free of human perceptive effects on a 2D surface through monocular vision. You may at first glance try to see them the same size, but in a static flat 2D rendition of monocular vision you allow yourself to compare actual scaling because you allow the introduction of imaginary *vanishing points* that are just converging lines on a 2D surface.

    Here is another:

    ex-1.jpg

    On the left is either an oval or a circle, it's ambiguous in a 2D representation. On the right you clearly see it as a circle within a rectangle. Think about this for a second, if you understand it as a rectangle then you understand and see it to have parallel sides so your human perception in the real 3D world has no vanishing point. It's only when I represent it in 2D that the vanishing point becomes clear, that you see clearly how a 3D object is represented in 2D space. I've added something in a 2D space that clearly influences your interpretation of what you see and makes you equate it to a 3D object. The reality is that it is an oval in a rhomboid on a 2D surface yet you don't interpret the reality, you see the illusion.

    Artists used to do it by either tracing lines on a sheet of glass or through use of a camera obscura. Vanishing points and perspective only become visible in 2D representations, in the 3D representation you eye/brain is doing as much as it can to subtract or cancel perspective in order to allow you a consistent view as you walk through the landscape. One where objects maintain a consistent size and distance relative to each other, one where they don't change relative to your position.

    I can distort our rectangle and circle in many ways by moving the imaginary vanishing point around, from the Degas style of *off the page* to one with a virtual horizon on the page:

    ex-2.jpg

    It still looks like a circle in a rectangle. But ask yourself if it were a real 3D object do vanishing points move as you walk around the object? How dizzy would you get if they did?

    Here's another. Imagine yourself walking past a 2D image, is the perspective or vanishing point actually fixed or is it entirely dependant on the position of the observer?

    img082_sRGB_ss2.jpg

    In this 2D representation there is a clear indication of a vanishing point towards the middle of the frame. But if we move to the right then the implied vanishing point moves off the left of the frame:

    ex-3.jpg

    And yet your impression is that you are still standing directly in front of the chair. If you walked past the subject then your impression would be entirely different, that the objects are static and their edges parallel.

    2D representations clearly do not behave in the same way as real 3D objects. This begs the question of whether trying to understand the 3D world in terms of how perspective works in a 2D image is preventing you from understanding how we actually see the real 3D world? Perhaps there is no real correlation between how we create the illusion of depth in a 2D image and we interpret the 3D world. Perhaps 2D images will always be a distortion and never a representation, that we only see them as representative because the *hypothesis generator* in our brains equates them to the view we are most familiar with - reality. In images we create an illusion of reality, not reality itself. Illusions work because we do not see them properly not because they are an accurate representation of reality. Even tilting the camera and displaying the image vertically distorts your perception...

    Renaissance perspective actually has five elements, linear perspective is but one. Trying to understand 2D images by only linear perspective is also trying to limit understanding into a rationalisation that is already understood and fails to recognise anything outside what is understood.
     
    Last edited: Nov 22, 2018
  8. Designer

    Designer TPF Noob!

    Joined:
    Apr 13, 2012
    Messages:
    18,491
    Likes Received:
    4,823
    Location:
    Iowa
    Speaking of 3D perception of a 2D object:

    (Scroll to 7:23.)

     
  9. Tim Tucker 2

    Tim Tucker 2 TPF Noob!

    Joined:
    Jun 8, 2017
    Messages:
    333
    Likes Received:
    241
    The illusion actually works in 3D as well. It's because you make the assumption that best fits your experience, that the head is a positive fold rather than negative. It highlights that a lot of what we see isn't so much real but the interpretation that makes to most realistic sense to us, the one that's most consistent to our experience and learning.
     
    • Informative Informative x 1
  10. ORourkeK

    ORourkeK TPF Noob!

    Joined:
    Dec 30, 2013
    Messages:
    311
    Likes Received:
    101
    Location:
    Philadelphia, PA


    Well, this has been a blast to read. Sure have missed this site. Thanks for the entertainment, all. :trink39:
     

Share This Page

Search tags for this page

depht of field an sensor size

,

dof and coc

,

how does dof change with sensor size