This topic is in the 2012 archive. You can read it and like its posts here. It is read only.
Archive 2012Canon 24mm T/SII vs ZE/ZF 25mm f2.0
p.3 #1Canon 24mm T/SII vs ZE/ZF 25mm f2.0
My brain hurts... with the C the grass up close is in focus, and so is the BG. But the pole tip is OOF.
PMNZ - could the TS have been tilted slightly by accident?
PMNZ - could the TS have been tilted slightly by accident?
Jan 16, 2012 at 08:29 PM
p.3 #2Canon 24mm T/SII vs ZE/ZF 25mm f2.0
What was the distance from the camera to the pole tip? If less than the distance to the ground (appx 7-8 feet), then it could be oof while the ground is in focus.
Edited on Jan 16, 2012 at 08:50 PM · View previous versions
Edited on Jan 16, 2012 at 08:50 PM · View previous versions
Jan 16, 2012 at 08:36 PM
p.3 #3Canon 24mm T/SII vs ZE/ZF 25mm f2.0
No it wasnt tilted. It was locked at zero. Ill put a couple of comparative boring shots up when I get home. They have more solid objects in the scene. It was actually the shot I printed.
Jan 16, 2012 at 08:38 PM
p.3 #4Canon 24mm T/SII vs ZE/ZF 25mm f2.0
Cool, thanks, that would be awesome!!
Jan 16, 2012 at 08:44 PM
p.3 #5Canon 24mm T/SII vs ZE/ZF 25mm f2.0
RustyBug wrote:
The C has a significantly larger image circle ... should account for a shallower DOF, a smoother bokeh and less vignetting.
Don't think it impacts the DOF / Bokeh. Definitely impacts sharpness into corners and vignetting.
The C has a significantly larger image circle ... should account for a shallower DOF, a smoother bokeh and less vignetting.
Don't think it impacts the DOF / Bokeh. Definitely impacts sharpness into corners and vignetting.
Jan 16, 2012 at 08:54 PM
p.3 #6Canon 24mm T/SII vs ZE/ZF 25mm f2.0
Think opposite of m4/3 where it gains DOF by having the smaller image circle
Jan 16, 2012 at 08:57 PM
p.3 #7Canon 24mm T/SII vs ZE/ZF 25mm f2.0
RustyBug wrote:
Think opposite of m4/3 where it gains DOF by having the smaller image circle
That is due to a smaller sensor size. The sensor size dictates the circle of confusion which is a factor for DOF.
Could be wrong but I don't think the image circle plays a role.
Think opposite of m4/3 where it gains DOF by having the smaller image circle
That is due to a smaller sensor size. The sensor size dictates the circle of confusion which is a factor for DOF.
Could be wrong but I don't think the image circle plays a role.
Jan 16, 2012 at 09:07 PM
p.3 #8Canon 24mm T/SII vs ZE/ZF 25mm f2.0
It only works that way (less DOF with larger image circle) if you are using the larger image circle in the image. Think using a Zeiss 35/1.4 lens made for FF on a crop sensor camera. You increase DOF at f/1.4 due to the smaller sensor.
Jan 16, 2012 at 09:10 PM
p.3 #9Canon 24mm T/SII vs ZE/ZF 25mm f2.0
So how does a smaller sensor change the projection coming from the lens ... the smaller sensor only captures the central portion of the exact same projection from the lens, i.e. it can't change the DOF being projected.
Jan 16, 2012 at 09:56 PM
p.3 #10Canon 24mm T/SII vs ZE/ZF 25mm f2.0
You need to take a look at a DOF calculator and do the math 
As Skumar25 stated the smaller sensor size changes the CoC, which changes the DOF.
As Skumar25 stated the smaller sensor size changes the CoC, which changes the DOF.
Jan 16, 2012 at 10:15 PM
p.3 #11Canon 24mm T/SII vs ZE/ZF 25mm f2.0
The image circle's diameter doesn't affect DOF.
The image sensor's size pertains to DOF only to the extent that its resolving power and desired enlargement ratio affect the choice of the maximum circle of confusion diameter.
Small sensors are usually (but not always) synonymous with high pixel density. However, if the pixel spacing is identical between two sensors of different dimensions, the maximum acceptable circle of confusion is also the same, because your ability to resolve spatial frequencies higher than the distance between adjacent pixels has not changed. It is because the smaller sensor tends to have more closely spaced pixels that contributes to the reason why the DOF criterion changes. However, it is not the ONLY reason. The other component is enlargement ratio, defined as the ratio of the size of the displayed image to the size of the sensor. That is to say, if you have an 18 MP full frame sensor and an 18 MP APS-C sensor, and you take the same image from the same subject distance at the same aperture, but you print the full frame image at exactly 1.6x the size of the APS-C image, the DOF will also be the same (assuming you don't print the image at such an unreasonably large size that you see individual pixels).
So, to recap, the image circle of a lens doesn't relate to its DOF. The sensor size only relates to DOF indirectly; it is actually the pixel spacing and maximum resolving power of the sensor that affects how we choose the maximum acceptable circle of confusion, with higher resolving power corresponding to smaller COC. Enlargement ratio (display size / image size) also plays a role. All of the above assumes we do not change focal length, subject distance, or f-number when comparing images.
Edited on Jan 16, 2012 at 10:18 PM · View previous versions
The image sensor's size pertains to DOF only to the extent that its resolving power and desired enlargement ratio affect the choice of the maximum circle of confusion diameter.
Small sensors are usually (but not always) synonymous with high pixel density. However, if the pixel spacing is identical between two sensors of different dimensions, the maximum acceptable circle of confusion is also the same, because your ability to resolve spatial frequencies higher than the distance between adjacent pixels has not changed. It is because the smaller sensor tends to have more closely spaced pixels that contributes to the reason why the DOF criterion changes. However, it is not the ONLY reason. The other component is enlargement ratio, defined as the ratio of the size of the displayed image to the size of the sensor. That is to say, if you have an 18 MP full frame sensor and an 18 MP APS-C sensor, and you take the same image from the same subject distance at the same aperture, but you print the full frame image at exactly 1.6x the size of the APS-C image, the DOF will also be the same (assuming you don't print the image at such an unreasonably large size that you see individual pixels).
So, to recap, the image circle of a lens doesn't relate to its DOF. The sensor size only relates to DOF indirectly; it is actually the pixel spacing and maximum resolving power of the sensor that affects how we choose the maximum acceptable circle of confusion, with higher resolving power corresponding to smaller COC. Enlargement ratio (display size / image size) also plays a role. All of the above assumes we do not change focal length, subject distance, or f-number when comparing images.
Edited on Jan 16, 2012 at 10:18 PM · View previous versions
Jan 16, 2012 at 10:17 PM
p.3 #12Canon 24mm T/SII vs ZE/ZF 25mm f2.0
Okay ... so a FF sensor with pixel density equal to pixel density of crop sensor (both using same lens) will yield same DOF as projection and CoC remain the same.
Now, use a lens with a larger image circle (i.e. different projection) on same pixel density, i.e. same CoC and what changes will occur ??
Edited on Jan 16, 2012 at 10:28 PM · View previous versions
Now, use a lens with a larger image circle (i.e. different projection) on same pixel density, i.e. same CoC and what changes will occur ??
Edited on Jan 16, 2012 at 10:28 PM · View previous versions
Jan 16, 2012 at 10:17 PM
p.3 #14Canon 24mm T/SII vs ZE/ZF 25mm f2.0
wickerprints wrote:
The image circle's diameter doesn't affect DOF.
The image sensor's size pertains to DOF only to the extent that its resolving power and desired enlargement ratio affect the choice of the maximum circle of confusion diameter.
Small sensors are usually (but not always) synonymous with high pixel density. However, if the pixel spacing is identical between two sensors of different dimensions, the maximum acceptable circle of confusion is also the same, because your ability to resolve spatial frequencies higher than the distance between adjacent pixels has not changed. It is because the smaller sensor tends to have more closely spaced pixels that...Show more →
Interesting, thanks!
The image circle's diameter doesn't affect DOF.
The image sensor's size pertains to DOF only to the extent that its resolving power and desired enlargement ratio affect the choice of the maximum circle of confusion diameter.
Small sensors are usually (but not always) synonymous with high pixel density. However, if the pixel spacing is identical between two sensors of different dimensions, the maximum acceptable circle of confusion is also the same, because your ability to resolve spatial frequencies higher than the distance between adjacent pixels has not changed. It is because the smaller sensor tends to have more closely spaced pixels that...Show more →
Interesting, thanks!
Jan 16, 2012 at 10:28 PM
p.3 #15Canon 24mm T/SII vs ZE/ZF 25mm f2.0
RustyBug wrote:
Okay ... so how does the smaller sensor change the CoC?
I explained it already. A sensor with more densely spaced pixels produces an image that, when displayed at a larger reproduction ratio, will have a DOF that corresponds to a smaller COC. This is because the system is able to resolve higher spatial frequencies, assuming that the sharpness of the image is not limited by diffraction, lens aberration, or some other optical phenomenon.
For instance, all else being equal, sensor A with size 10 x 10mm at an inter-pixel spacing of 10 micrometers will be able to resolve spatial frequencies twice as high as sensor B with size 20 x 20mm at an inter-pixel spacing of 20 micrometers, even though both sensors have the same number of pixels. Using the same optically ideal lens for each, if you reproduce the images obtained by each sensor at the same display size--say, 30 x 30 cm, then your reproduction ratio for sensor A is twice that for sensor B, meaning your choice of COC to model the DOF in each image should be twice as small for A as for B.
If, however, you displayed the image from sensor A at the SAME reproduction ratio as for sensor B--so for example, image A is shown at 15 x 15 cm whereas B is shown at 30 x 30 cm, then the apparent DOF will again look identical despite the smaller pixel spacing of A. That is, display size matters when considering DOF. How much you enlarge the image will affect your ability to see whether an image point is out of focus. And the inter-pixel spacing also affects DOF because it influences whether or not the sensor can record data that shows whether an image point is out of focus.
Okay ... so how does the smaller sensor change the CoC?
I explained it already. A sensor with more densely spaced pixels produces an image that, when displayed at a larger reproduction ratio, will have a DOF that corresponds to a smaller COC. This is because the system is able to resolve higher spatial frequencies, assuming that the sharpness of the image is not limited by diffraction, lens aberration, or some other optical phenomenon.
For instance, all else being equal, sensor A with size 10 x 10mm at an inter-pixel spacing of 10 micrometers will be able to resolve spatial frequencies twice as high as sensor B with size 20 x 20mm at an inter-pixel spacing of 20 micrometers, even though both sensors have the same number of pixels. Using the same optically ideal lens for each, if you reproduce the images obtained by each sensor at the same display size--say, 30 x 30 cm, then your reproduction ratio for sensor A is twice that for sensor B, meaning your choice of COC to model the DOF in each image should be twice as small for A as for B.
If, however, you displayed the image from sensor A at the SAME reproduction ratio as for sensor B--so for example, image A is shown at 15 x 15 cm whereas B is shown at 30 x 30 cm, then the apparent DOF will again look identical despite the smaller pixel spacing of A. That is, display size matters when considering DOF. How much you enlarge the image will affect your ability to see whether an image point is out of focus. And the inter-pixel spacing also affects DOF because it influences whether or not the sensor can record data that shows whether an image point is out of focus.
Jan 16, 2012 at 10:29 PM
p.3 #16Canon 24mm T/SII vs ZE/ZF 25mm f2.0
RustyBug wrote:
Okay ... so a FF sensor with pixel density equal to pixel density of crop sensor (both using same lens) will yield same DOF as projection and CoC remain the same.
Only if the reproduction ratio is also the same, so that the image recorded by the FF sensor is displayed proportionately larger than the image recorded by the crop sensor. So your print sizes have to be proportional to the sensors they were printed from.
Now, use a lens with a larger image circle (i.e. different projection) on same pixel density, i.e. same CoC and what changes will occur ??
If you compare two lenses that are identical in every way except that the image circle of one lens is larger than the image circle of the other, and place behind them the same sensor at the same back focus distance, and you take the same shot at the same f-number and same subject distance, the COC is also the same. (We assume that the image circle of both lenses are large enough to cover the sensor.) A larger image circle doesn't affect DOF. All that it does is admit more of the scene, but in order to see the periphery of that larger circle, you need a sensor that is correspondingly larger.
Here is a thought experiment that may help to illustrate what is happening. Say you are working with the EF 24/1.4L II on an APS-C body. I picked this lens because we can reasonably assume that its image circle is only just large enough to cover a full frame sensor, but we're not using full frame. So this lens projects an image circle that is comfortably larger than what we can capture with the crop sensor. Now say we are able to put a baffle between the lens and the sensor--a field stop, if you will--that blocks light in such a way that it reduces the resulting image circle diameter to be exactly equal to the APS-C diagonal. Would you see any difference in the resulting image? Of course not. The image looks identical because you have only eliminated the light that never would have struck the APS-C sensor to begin with. So the image circle's size doesn't affect the apparent DOF.
Edited on Jan 16, 2012 at 10:45 PM · View previous versions
Okay ... so a FF sensor with pixel density equal to pixel density of crop sensor (both using same lens) will yield same DOF as projection and CoC remain the same.
Only if the reproduction ratio is also the same, so that the image recorded by the FF sensor is displayed proportionately larger than the image recorded by the crop sensor. So your print sizes have to be proportional to the sensors they were printed from.
Now, use a lens with a larger image circle (i.e. different projection) on same pixel density, i.e. same CoC and what changes will occur ??
If you compare two lenses that are identical in every way except that the image circle of one lens is larger than the image circle of the other, and place behind them the same sensor at the same back focus distance, and you take the same shot at the same f-number and same subject distance, the COC is also the same. (We assume that the image circle of both lenses are large enough to cover the sensor.) A larger image circle doesn't affect DOF. All that it does is admit more of the scene, but in order to see the periphery of that larger circle, you need a sensor that is correspondingly larger.
Here is a thought experiment that may help to illustrate what is happening. Say you are working with the EF 24/1.4L II on an APS-C body. I picked this lens because we can reasonably assume that its image circle is only just large enough to cover a full frame sensor, but we're not using full frame. So this lens projects an image circle that is comfortably larger than what we can capture with the crop sensor. Now say we are able to put a baffle between the lens and the sensor--a field stop, if you will--that blocks light in such a way that it reduces the resulting image circle diameter to be exactly equal to the APS-C diagonal. Would you see any difference in the resulting image? Of course not. The image looks identical because you have only eliminated the light that never would have struck the APS-C sensor to begin with. So the image circle's size doesn't affect the apparent DOF.
Edited on Jan 16, 2012 at 10:45 PM · View previous versions
Jan 16, 2012 at 10:37 PM
p.3 #17Canon 24mm T/SII vs ZE/ZF 25mm f2.0
wickerprints wrote:
Using the same optically ideal lens for each, .
But the TSE is not optically ideal for the FF format (it is significatly oversized), whereas the Zeiss is.
So ... you're saying that the DOF of my Oly 24 and the 24 TSE II should be identical on FF? Is this correct?
Using the same optically ideal lens for each, .
But the TSE is not optically ideal for the FF format (it is significatly oversized), whereas the Zeiss is.
So ... you're saying that the DOF of my Oly 24 and the 24 TSE II should be identical on FF? Is this correct?
Jan 16, 2012 at 10:42 PM
p.3 #18Canon 24mm T/SII vs ZE/ZF 25mm f2.0
wickerprints wrote:
Using the same optically ideal lens for each, .
RustyBug wrote:
But the TSE is not optically ideal for the FF format (it is significatly oversized), whereas the Zeiss is.
Per the previous discussion, what I mean by "optically ideal" is not that one lens is sharper than the other or optimized for a specific format size, but that we can reasonably assume that effects such as diffraction or differences in MTF are not a significant contributor to the ability to discern whether an image point is within acceptably sharp focus.
Please take note that I make NO claim whatsoever as to the relative imaging performance of these two lenses and I will not do so. My sole interest in this thread is in explaining how various imaging system properties influence how we model the depth of field. As such, it suffices that the lens(es) in question is(are) reasonably sharp enough that they are capable of exceeding the resolving power of the sensors that are placed behind them, in the area of the image for which the DOF is being measured.
Using the same optically ideal lens for each, .
RustyBug wrote:
But the TSE is not optically ideal for the FF format (it is significatly oversized), whereas the Zeiss is.
Per the previous discussion, what I mean by "optically ideal" is not that one lens is sharper than the other or optimized for a specific format size, but that we can reasonably assume that effects such as diffraction or differences in MTF are not a significant contributor to the ability to discern whether an image point is within acceptably sharp focus.
Please take note that I make NO claim whatsoever as to the relative imaging performance of these two lenses and I will not do so. My sole interest in this thread is in explaining how various imaging system properties influence how we model the depth of field. As such, it suffices that the lens(es) in question is(are) reasonably sharp enough that they are capable of exceeding the resolving power of the sensors that are placed behind them, in the area of the image for which the DOF is being measured.
Jan 16, 2012 at 10:49 PM
p.3 #19Canon 24mm T/SII vs ZE/ZF 25mm f2.0
Understood "optically ideal" at designed for specific format size ... i.e. Oly 24/2.8 designed for FF, TSE designed for FF ... both "optically ideal" for FF, yet two different image circles. Would that represent a fair comparison test with both on the same FF sensor?
Edited on Jan 16, 2012 at 10:56 PM · View previous versions
Edited on Jan 16, 2012 at 10:56 PM · View previous versions
Jan 16, 2012 at 10:55 PM
p.3 #20Canon 24mm T/SII vs ZE/ZF 25mm f2.0
RustyBug wrote:
So ... you're saying that the DOF of my Oly 24 and the 24 TSE II should be identical on FF? Is this correct?
To a first-order approximation, yes.
I have thus far intentionally avoided discussing the intricacies of modeling depth of field that would require explicit ray-tracing of the specific optical formula, because this would confuse the topic further. As we customarily approximate depth of field for a symmetric lens design (unitary pupil ratio), we care only about three variables (focal length, subject distance, and f-number) and one parameter (choice of maximum acceptable circle of confusion). But this model is not perfect because real-world lenses are generally complex and not thin-lens Gaussian idealizations. To take the optical formula into account would require simulation that is beyond the scope of what I would consider to be reasonable.
So ... you're saying that the DOF of my Oly 24 and the 24 TSE II should be identical on FF? Is this correct?
To a first-order approximation, yes.
I have thus far intentionally avoided discussing the intricacies of modeling depth of field that would require explicit ray-tracing of the specific optical formula, because this would confuse the topic further. As we customarily approximate depth of field for a symmetric lens design (unitary pupil ratio), we care only about three variables (focal length, subject distance, and f-number) and one parameter (choice of maximum acceptable circle of confusion). But this model is not perfect because real-world lenses are generally complex and not thin-lens Gaussian idealizations. To take the optical formula into account would require simulation that is beyond the scope of what I would consider to be reasonable.
Jan 16, 2012 at 10:56 PM