Doug Kerr
Well-known member
The concept of depth of field is wholly man-made (like the notion of the age at which one can vote, or how long one should wait after the death of a spouse before remarrying). It is not a fundamental relationship dictated by the laws of optics (although certainly the laws of optics are involved as we manipulate the concept).
Briefly, the concept of depth of field is the answer to this question:
We adopt that bogey in terms of a certain limiting diameter of the circle of confusion, which is the blur figure produced on the focal plane from a point on the object when the object is not in perfect focus. I call that limiting diameter the circle of confusion diameter limit (COCDL).
Traditionally, going back to film practice, it was common to choose a COCDL as a fraction of the diagonal size of the image. This was predicated on the concept that:
• blurring would only be noticeable if the angular diameter of the circle of confusion, as seen on the print, exceeded the eye's angular resolution, based on a print of size (for a 3:2 format) 12" x 8", viewed from a distance of about 16".
Based on that rationale, we would choose a COCDL of about 1/1400 the frame diagonal dimension.
Note that the COCDL is not a property of the camera (we choose it), but when we choose one using that "rule of thumb", the format size of the camera is involved.
In modern times, many workers advocate choosing a COCDL on this basis:
• Any blurring in which the diameter of the circle of confusion exceeds the resolution of the sensor system is "not negligible".
This leads to the adoption of a COCDL of perhaps twice the sensel pitch of the camera.
We must be aware that the choice of a COCDL for our depth of field calculations has no effect on camera behavior nor on the image produced. The camera has no idea what value we choose.
Rather, our reckoning of depth of field may govern our decision as to what "setup" to use in a particular photographic situation, and through that affects the camera behavior (and thus the nature of the image).
For example, suppose we have a situation in which, with a focal length of 50 mm, we wish to focus at a distance of 3 m, and want the total depth of field (the distance between the distances, nearer and farther than 3 m, at which a object will be in "acceptable" focus) to be 1 m. Suppose we are using a full-frame 35-mm format, and choose a COCDL of 1/1400 of the frame diagonal (about 0.031 mm).
The depth of field equations tell us that, if we use an aperture of f/4.5, the total depth of field (under that "bogey") will be about 1 m. So we might decide to use that aperture.
Now suppose we think it more prudent to use the "modern", camera-resolution-based rule of thumb for choosing a COCDL. Suppose the camera is a Canon EOS 1Ds Mark 3. Its sensel pitch is about 7 µm. Based on one version of the "new" rule of thumb, that would suggest a COCDL of 14 µm (0.014 mm).
Now, based on our original scenario, to attain the 1 m total depth of field (as defined under that COCDL value), we would need to use an aperture of about f/10.
Now, would the photographic result be "better"? Well, the image would be less blurred for objects at distances of 2.57 m and 3.57 m (our near and far limits of the "desired" depth of field in either case). The background (at any given distance) would be less blurred. We might have to use a longer exposure time, and that might have repercussions. Etc, etc.
Hyperfocal distance
The hyperfocal distance for any given focal length and aperture, based on some chosen COCDL as the definition of our bogey for "acceptable" blurring, is that distance such that, if the camera is focused at that distance, the far limit of the depth of field just falls at infinity. A corollary is that the near limit of the depth of field falls at half the focus distance. In other words, objects at any distance greater than half the focus distance will be imaged with blurring not worse than our "bogey".
Suppose we plan to use a lens of a certain focal length and a certain aperture for a landscape shot, and decide to use the hyperfocal distance as our focus distance. We first reckon it using, as the COCDL, 1/1400 of the frame diagonal (the "traditional" COCDL guideline). We get a certain hyperfocal distance. In "shot A", we set the focus to that distance.
But then we reconsider, and reckon the hyperfocal distance using 0.014 mm as the COCDL. We get a substantially greater hyperfocal distance. In shot "B", we set the focus to that distance.
How do the results differ? Well in shot B, we find that for objects closer the original hyperfocal distance, the blurring (in absolute terms, as the actual diameter of the actual circle of confusion) is greater.
So, in the reckoning of hyperfocal distance, as a guide to focus setting, is the use of the, "modern more-stringent" COCDL value more conservative (foolproof)? Not really.
The bottom line is this: please keep in mind what the determination of depth of field, or the related matter of hyperfocal distance, is and isn't, what it does and does not do.
Best regards,
Doug
Briefly, the concept of depth of field is the answer to this question:
For a given focal length and aperture, and for focus at a certain distance, over what range of object distances will the object be imaged on the focal plane with blurring, resulting from imperfect focus, not worse than a certain "bogey" we adopt.
We adopt that bogey in terms of a certain limiting diameter of the circle of confusion, which is the blur figure produced on the focal plane from a point on the object when the object is not in perfect focus. I call that limiting diameter the circle of confusion diameter limit (COCDL).
Most often in modern photographic work that value is just called the "circle of confusion". The problem with that is if I want to actually speak of the circle of confusion itself (a figure on the focal plane, not a number), I have no name for it. There is a similar problem if I wish to speak of the actual diameter of some real or hypothetical circle of confusion.
Accordingly, I call:
• The circle of confusion "the circle of confusion"
• The diameter of come circle of confusion "the circle of confusion diameter"
• The limit we adopt for the diameter of the circle of confusion to establish a bogey for misfocus "the circle of confusion diameter limit".
Accordingly, I call:
• The circle of confusion "the circle of confusion"
• The diameter of come circle of confusion "the circle of confusion diameter"
• The limit we adopt for the diameter of the circle of confusion to establish a bogey for misfocus "the circle of confusion diameter limit".
Traditionally, going back to film practice, it was common to choose a COCDL as a fraction of the diagonal size of the image. This was predicated on the concept that:
• blurring would only be noticeable if the angular diameter of the circle of confusion, as seen on the print, exceeded the eye's angular resolution, based on a print of size (for a 3:2 format) 12" x 8", viewed from a distance of about 16".
Based on that rationale, we would choose a COCDL of about 1/1400 the frame diagonal dimension.
Note that the COCDL is not a property of the camera (we choose it), but when we choose one using that "rule of thumb", the format size of the camera is involved.
In modern times, many workers advocate choosing a COCDL on this basis:
• Any blurring in which the diameter of the circle of confusion exceeds the resolution of the sensor system is "not negligible".
This leads to the adoption of a COCDL of perhaps twice the sensel pitch of the camera.
We must be aware that the choice of a COCDL for our depth of field calculations has no effect on camera behavior nor on the image produced. The camera has no idea what value we choose.
Rather, our reckoning of depth of field may govern our decision as to what "setup" to use in a particular photographic situation, and through that affects the camera behavior (and thus the nature of the image).
For example, suppose we have a situation in which, with a focal length of 50 mm, we wish to focus at a distance of 3 m, and want the total depth of field (the distance between the distances, nearer and farther than 3 m, at which a object will be in "acceptable" focus) to be 1 m. Suppose we are using a full-frame 35-mm format, and choose a COCDL of 1/1400 of the frame diagonal (about 0.031 mm).
The depth of field equations tell us that, if we use an aperture of f/4.5, the total depth of field (under that "bogey") will be about 1 m. So we might decide to use that aperture.
Now suppose we think it more prudent to use the "modern", camera-resolution-based rule of thumb for choosing a COCDL. Suppose the camera is a Canon EOS 1Ds Mark 3. Its sensel pitch is about 7 µm. Based on one version of the "new" rule of thumb, that would suggest a COCDL of 14 µm (0.014 mm).
Now, based on our original scenario, to attain the 1 m total depth of field (as defined under that COCDL value), we would need to use an aperture of about f/10.
Now, would the photographic result be "better"? Well, the image would be less blurred for objects at distances of 2.57 m and 3.57 m (our near and far limits of the "desired" depth of field in either case). The background (at any given distance) would be less blurred. We might have to use a longer exposure time, and that might have repercussions. Etc, etc.
Hyperfocal distance
The hyperfocal distance for any given focal length and aperture, based on some chosen COCDL as the definition of our bogey for "acceptable" blurring, is that distance such that, if the camera is focused at that distance, the far limit of the depth of field just falls at infinity. A corollary is that the near limit of the depth of field falls at half the focus distance. In other words, objects at any distance greater than half the focus distance will be imaged with blurring not worse than our "bogey".
Suppose we plan to use a lens of a certain focal length and a certain aperture for a landscape shot, and decide to use the hyperfocal distance as our focus distance. We first reckon it using, as the COCDL, 1/1400 of the frame diagonal (the "traditional" COCDL guideline). We get a certain hyperfocal distance. In "shot A", we set the focus to that distance.
But then we reconsider, and reckon the hyperfocal distance using 0.014 mm as the COCDL. We get a substantially greater hyperfocal distance. In shot "B", we set the focus to that distance.
How do the results differ? Well in shot B, we find that for objects closer the original hyperfocal distance, the blurring (in absolute terms, as the actual diameter of the actual circle of confusion) is greater.
So, in the reckoning of hyperfocal distance, as a guide to focus setting, is the use of the, "modern more-stringent" COCDL value more conservative (foolproof)? Not really.
The bottom line is this: please keep in mind what the determination of depth of field, or the related matter of hyperfocal distance, is and isn't, what it does and does not do.
Best regards,
Doug
