-
Please use real names.
Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed
Firstname Lastname
Register
We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know! -
Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!
About depth of field
- Thread starter Doug Kerr
- Start date
Doug Kerr
Well-known member
Hi, Bart,
But not very carefully. There is a lot of very questionable stuff in his piece.
I am not at first really sympathetic with his notion that a camera with a lens with severe aberrations will exhibit lesser depth of field since the aberrations will already have used up so much of our tolerance for blurring, leaving less to be caused by misfocus blurring.
But examining this closely requires us to use a more accurate model that we usually do for the joint effect of blurring from finite camera resolution and blurring from misfocus.
Perhaps if we use a more accurate model of that, his point will have some validity.
I can't deal with that before breakfast.
Best regards,
Doug
Hi Cem,
He must have been following the discussion here ...
But not very carefully. There is a lot of very questionable stuff in his piece.
I am not at first really sympathetic with his notion that a camera with a lens with severe aberrations will exhibit lesser depth of field since the aberrations will already have used up so much of our tolerance for blurring, leaving less to be caused by misfocus blurring.
But examining this closely requires us to use a more accurate model that we usually do for the joint effect of blurring from finite camera resolution and blurring from misfocus.
Perhaps if we use a more accurate model of that, his point will have some validity.
I can't deal with that before breakfast.
Best regards,
Doug
Doug Kerr
Well-known member
In this thread, I have called attention to the fact that as we prepare to determine the projected depth of field for a shot being planned, we must adopt a value for the parameter COCDL. This is our way of telling the calculator the amount of blurring from imperfect focus that we are willing to "accept" for our subjects at different distances from the camera.
I pointed out that there are two widely-used premises for choosing a COCDL value.
A. Here, we treat as "acceptable" a blurring that is comparable to the resolution of human vision (for a viewer with "20/20" vision) when viewing the image as a print of a certain size at a certain viewing distance, with perhaps a "fudge factor" put in to make the requirement less stringent.
The rationale here is to make the blurring (due to imperfect focus) of subjects we consider important "invisible" to the viewer of the image under the assumed conditions.
B. Here, we treat as "acceptable" a blurring that is comparable to the resolution of the camera itself. Sometimes for simplification, we treat that as the sensel pitch, with perhaps a "fudge factor" put in to make the requirement less stringent.
The rationale her is that the blurring (due to imperfect focus) of subjects we consider important will not degrade image sharpness "below" that attainable by virtue of the camera's in-focus resolution.
******
I thought it would be interesting to demonstrate the difference in calculated depth of field using a hopefully-realistic example under these two premises.
We assume a Canon EOS 6D camera, with a frame size of 36 x 24 mm and a sensel pitch of 6.58 um.
We will assume a lens of focal length 50 mm operated at an aperture of f/3.5, focused at a distance of 10 m.
Under premise A, we will base our COCDL on the viewer viewing a print of dimensions 12" x 8" from a distance of 16.7".
This viewing situation gives "life-size" viewing of the subjects in this case.
We will use no "fudge factor" to "ease" the requirement.
This premise suggests a COCDL of 0.0145 mm.
Under premise B, we will use the simplified basis, involving the sensel pitch, and will use no "fudge factor" to "ease" the requirement.
This premise suggests a COCDL of 0.00658 mm.
The results
For the shot setup parameters mentioned before, the results are:
For the "premise A" COCDL
Near limit of field: 8.32 m Near depth of field: -1.68 m
Far limit of field: 12.53 m Far depth of field: +2.53 m
Total depth of field: 4.21 m
For the "premise B" COCDL
Near limit of field: 6.16 m Near depth of field: -0.84 m
Far limit of field: 11.01 m Far depth of field: +1.01 m
Total depth of field: 1.85 m
The message here:
If we want no perceptible degradation of the images of the important subjects due to imperfect focus, we must confine the subjects to a substantially shorter range of distances than if our objective was to not have the blurring due to imperfect focus visible to a viewer viewing the image at "life size".
Best regards,
Doug
I pointed out that there are two widely-used premises for choosing a COCDL value.
A. Here, we treat as "acceptable" a blurring that is comparable to the resolution of human vision (for a viewer with "20/20" vision) when viewing the image as a print of a certain size at a certain viewing distance, with perhaps a "fudge factor" put in to make the requirement less stringent.
The rationale here is to make the blurring (due to imperfect focus) of subjects we consider important "invisible" to the viewer of the image under the assumed conditions.
B. Here, we treat as "acceptable" a blurring that is comparable to the resolution of the camera itself. Sometimes for simplification, we treat that as the sensel pitch, with perhaps a "fudge factor" put in to make the requirement less stringent.
The rationale her is that the blurring (due to imperfect focus) of subjects we consider important will not degrade image sharpness "below" that attainable by virtue of the camera's in-focus resolution.
******
I thought it would be interesting to demonstrate the difference in calculated depth of field using a hopefully-realistic example under these two premises.
We assume a Canon EOS 6D camera, with a frame size of 36 x 24 mm and a sensel pitch of 6.58 um.
We will assume a lens of focal length 50 mm operated at an aperture of f/3.5, focused at a distance of 10 m.
Under premise A, we will base our COCDL on the viewer viewing a print of dimensions 12" x 8" from a distance of 16.7".
This viewing situation gives "life-size" viewing of the subjects in this case.
We will use no "fudge factor" to "ease" the requirement.
This premise suggests a COCDL of 0.0145 mm.
Under premise B, we will use the simplified basis, involving the sensel pitch, and will use no "fudge factor" to "ease" the requirement.
This premise suggests a COCDL of 0.00658 mm.
The results
For the shot setup parameters mentioned before, the results are:
For the "premise A" COCDL
Near limit of field: 8.32 m Near depth of field: -1.68 m
Far limit of field: 12.53 m Far depth of field: +2.53 m
Total depth of field: 4.21 m
For the "premise B" COCDL
Near limit of field: 6.16 m Near depth of field: -0.84 m
Far limit of field: 11.01 m Far depth of field: +1.01 m
Total depth of field: 1.85 m
The message here:
If we want no perceptible degradation of the images of the important subjects due to imperfect focus, we must confine the subjects to a substantially shorter range of distances than if our objective was to not have the blurring due to imperfect focus visible to a viewer viewing the image at "life size".
Best regards,
Doug
Doug Kerr
Well-known member
This is further to Ctein's assertion that a camera with a lens with severe aberrations will exhibit lesser depth of field since the aberrations will already have used up so much of our tolerance for blurring, leaving less to be caused by misfocus blurring.
I will of course here have to make many assumptions, one of which is that the camera aberrations produce an essentially Gaussian point spread function, and that the blurring caused by misfocus also has an essentially Gaussian point spread function.
I will also assume that we will consider a decline in resolution (in terms of cycles per picture height) of 20% (from the camera resolution for a perfectly-focused object) to be "unacceptable". (This is essentially what we do when we set the COCDL at the sensel pitch, even though we may not realize that.)
These principles are described here:
http://www.openphotographyforums.com/forums/showthread.php?t=17501
We begin with a camera with 2000 sensels per picture height. We assume its resolution to be 750 cycles per picture height (that is a test on the sensor only, as if a "prefect" lens were in place).
To calculate the expected depth of field, we adopt a COCDL of 1/2000 picture height (the sensel pitch).
Using some parameters I arbitrarily used in a recent illustration, that would lead to a "total" depth of field of 1.85 m.
The result will be that, for an object at a limit of the calculated depth of field, the resolution will be 600 cycles per picture height, a decline of 20% from the in-focus value of 750 cycles per picture height.
Next we replace the "essentially-perfect" lens with another, really crummy one. Now, the perfect-focus resolution is only 500 cycles per picture height.
Now, how do we think about depth of field?
Here is one view:
We have a right to expect a resolution of 750 cycles per picture height (since that's what we get with a "perfect" lens on the camera).
By our previous decisions, we have (without really knowing it) said that (with that really nice lens on board) we will accept a resolution degraded to 600 cycles per picture height to be the limit of what is
"acceptable" as a result of misfocus, and calculate the limits of object distance that will keep us within that.
So what is the depth of field with the new crummy lens in place? For what range of object distances will the net resolution (degraded by misfocus) be not less than 600 cy/PH?
No range at all. This will never happen, even at perfect focus - there, the resolution will be 500 cy/PH.
So her Ctein is right: this lens exhibits a smaller depth of field - zero, in fact.
But I don't find that outlook "useful".
Now here's another.
With the new, crummy lens aboard, we will calculate the range of object distances for which the overall resolution, including the effect of misfocus, is not degraded by more than 20% from the value for perfect focus (with this lens).
That will happen if we use a COCDL that is 1.5 times the pixel pitch (750/600 times what we used before).
Then, the calculated depth of field will be 2.80 m (substantially more than before)!
So what's the bottom line? The comment about the field of view being less because the performance of the crummy lens has burned part of (in my example, more than all of) our "quota" of blurring is of no actual significance to us in shot planning.
Best regards,
Doug
I will of course here have to make many assumptions, one of which is that the camera aberrations produce an essentially Gaussian point spread function, and that the blurring caused by misfocus also has an essentially Gaussian point spread function.
I will also assume that we will consider a decline in resolution (in terms of cycles per picture height) of 20% (from the camera resolution for a perfectly-focused object) to be "unacceptable". (This is essentially what we do when we set the COCDL at the sensel pitch, even though we may not realize that.)
These principles are described here:
http://www.openphotographyforums.com/forums/showthread.php?t=17501
We begin with a camera with 2000 sensels per picture height. We assume its resolution to be 750 cycles per picture height (that is a test on the sensor only, as if a "prefect" lens were in place).
To calculate the expected depth of field, we adopt a COCDL of 1/2000 picture height (the sensel pitch).
Using some parameters I arbitrarily used in a recent illustration, that would lead to a "total" depth of field of 1.85 m.
The result will be that, for an object at a limit of the calculated depth of field, the resolution will be 600 cycles per picture height, a decline of 20% from the in-focus value of 750 cycles per picture height.
Next we replace the "essentially-perfect" lens with another, really crummy one. Now, the perfect-focus resolution is only 500 cycles per picture height.
Now, how do we think about depth of field?
Here is one view:
We have a right to expect a resolution of 750 cycles per picture height (since that's what we get with a "perfect" lens on the camera).
By our previous decisions, we have (without really knowing it) said that (with that really nice lens on board) we will accept a resolution degraded to 600 cycles per picture height to be the limit of what is
"acceptable" as a result of misfocus, and calculate the limits of object distance that will keep us within that.
So what is the depth of field with the new crummy lens in place? For what range of object distances will the net resolution (degraded by misfocus) be not less than 600 cy/PH?
No range at all. This will never happen, even at perfect focus - there, the resolution will be 500 cy/PH.
So her Ctein is right: this lens exhibits a smaller depth of field - zero, in fact.
But I don't find that outlook "useful".
Now here's another.
With the new, crummy lens aboard, we will calculate the range of object distances for which the overall resolution, including the effect of misfocus, is not degraded by more than 20% from the value for perfect focus (with this lens).
That will happen if we use a COCDL that is 1.5 times the pixel pitch (750/600 times what we used before).
Then, the calculated depth of field will be 2.80 m (substantially more than before)!
So what's the bottom line? The comment about the field of view being less because the performance of the crummy lens has burned part of (in my example, more than all of) our "quota" of blurring is of no actual significance to us in shot planning.
Best regards,
Doug