If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#31
|
|||
|
|||
hyperfocal distance
Dave Martindale wrote:
M Barnes wrote: OK, but in a lens lab, don't engineers use a standard line chart to determine the CoC. In other words, no print is generated. Rather, the techs use the chart to determine CoC by focusing the image on the capture area, whether a 35x24 for film or a 66% sensor area for digital. The question being, will this scaling, if a print is not used, cause the HD to scale as well? You're thinking of something else. "Circle of Confusion" is a theoretical concept, not a measured value. Perhaps you are thinking of resolution tests for a lens? Yes, I was under the impression that the two were related, in that the resolution test for a lens determined the CoC value for that lens when computing other values, for example, the HD for a given set of values. In this, it would be like the "tortuosity factor" in filtration physics, a number that doesn't exist and can't be measured, but is essential to compute filtration efficiency in real-world applications. The minimum blur spot size for a lens is normally quite a lot smaller than the CoC used in depth of field calculations. I'm not sure I follow this. Do you mean that the CoC is not a base level of resolution or acuity, but only a semi-useful construct, and that the blur spot is a more meaningful measure of lens acuity? The way you calculate depth of field is (more or less) this: Pick a print size and viewing distance. Pick an angular resolution that will look "sharp enough" for your eye. This gives you the Circle of Confusion size on the print. A standard value is 1/1740 of the diagonal of the print, but there are other possible choices depending on how critical you are about sharpness. Now scale the CoC diameter down by the reciprocal of the printing magnification. This tells you the CoC size on the sensor that gives you the desired CoC size on the print. If you picked diagonal/1740 as the print CoC size, then this will be sensor_diagonal/1740. For this discussion, I'm holding print size at 8x10, lens focal length and all other parameters constant. The only variable will be sensor size -- 35mm and the D100 2/3 equivalent. From here, we use pure geometric optics. snip a very good explanation of the theory of DOF Finally, to determine the hyperfocal distance, just calculate the lens focus setting that places the far limit of the DOF exactly at infinity. (There's one lie in the above. Real DOF tables usually give distance from the film plane, not distance from the lens front principal plane. This makes the exact calculations messier, but there's little practical difference between the two methods for non-macro photography). Well, two things: I'm used to measuring distance to subjects with a tape from the focal-plane mark on the camera body, so that does apply to me, especially when I'm using extension tubes and/or macro lenses; and ... the 64-dollar question -- Does all that add up to an adjustment in the HD when sensor size changes and all else is held constant? There seems to be a circle of confusion regarding the answer, and I'm not smart or experienced enough to figure it out, but you seem to be. |
#32
|
|||
|
|||
hyperfocal distance
"Tony Spadaro" wrote in
. com: Sorry - the intuitive answer is right. Yes the more you enlarge the softer the image gets but it's just plain silly to assume any single size for the final image since as size of the image increases, viewing distance also increases and they cancel each other out. The formula I gave is the formula for hyper focal distance. It behaves exactly as I said. The hyper focal distance is inverse proportional to the size of of the sensor, everything else kept equal. You may not like the definition. You may want something else; but then it is not hyperfocal distance. You may call it Tony Spadero distance if you want. /Roland |
#33
|
|||
|
|||
hyperfocal distance
"M Barnes" writes:
You're thinking of something else. "Circle of Confusion" is a theoretical concept, not a measured value. Perhaps you are thinking of resolution tests for a lens? Yes, I was under the impression that the two were related, in that the resolution test for a lens determined the CoC value for that lens when computing other values, for example, the HD for a given set of values. No. CoC is not a property of the lens at all. It is a measured or estimated tolerance of how blurred an image can be before the viewer, looking at a print, feels it is out of focus. The minimum blur spot size for a lens is normally quite a lot smaller than the CoC used in depth of field calculations. I'm not sure I follow this. Do you mean that the CoC is not a base level of resolution or acuity, but only a semi-useful construct, and that the blur spot is a more meaningful measure of lens acuity? CoC is not a measure of lens acuity *at all*. It is a measure of how badly out of focus something can be allowed to get before it seems unsharp to a person viewing the image under some assumed set of conditions. CoC is a useful construct for calculating DOF, not lens resolution. If you have a lens whose blur spot diameter is equal to the Circle of Confusion for the sensor size you're working with, then *everything* will be borderline unsharp, even what ever is in best focus. Normally, the CoC will be several times the lens blur spot size if the CoC is reasonable and the lens is suited to the format. For this discussion, I'm holding print size at 8x10, lens focal length and all other parameters constant. The only variable will be sensor size -- 35mm and the D100 2/3 equivalent. That's an unusual choice, because you won't get the same field of view if you use the same lens focal length for 35 mm and for 2/3 frame. If you do this, the short answer is that the D100 will give you only 2/3 as much DOF as you would get with a full-frame sensor, at the same FL and same f/number. (This is because you have to enlarge the D100 image more). On the other hand, if you changed the lens focal length to keep the same field of view, so the D100 lens is 2/3 the focal length of the 35 camera lens, then the D100 would give you 3/2 times as much DOF as the 35 camera. (There's one lie in the above. Real DOF tables usually give distance from the film plane, not distance from the lens front principal plane. This makes the exact calculations messier, but there's little practical difference between the two methods for non-macro photography). Well, two things: I'm used to measuring distance to subjects with a tape from the focal-plane mark on the camera body, so that does apply to me, especially when I'm using extension tubes and/or macro lenses; Ok. I'd suggest you build your own DOF tables using a spreadsheet. It's not really that hard. If you're calculating things in terms of distances from lens front principal plane to subject, and from rear principal plane to film/sensor, you can use classic lens formulas directly. If you want to know subject-to-film distance, you need to add together the lens-subject distance plus the lens-image distance *plus the spacing between the two principal planes within the lens*. You may have to guess at the latter number. Also, note that some modern lenses do not focus by moving the glass as a unit towards or away from the sensor, they move elements within the lens like a zoom. For these lenses, the focal length may change from the marked value when focused closer than infinity, and the usual DOF calculations will have some error. and ... the 64-dollar question -- Does all that add up to an adjustment in the HD when sensor size changes and all else is held constant? There seems to be a circle of confusion regarding the answer, and I'm not smart or experienced enough to figure it out, but you seem to be. Yes, see above. If you keep the lens FL constant, thus narrowing the FOV when sensor size decreases, the DOF also *decreases*. On the other hand, if you make lens FL decrease along with sensor size, so FOV remains constant, then the DOF *increases* with decreasing sensor size. Dave |
#34
|
|||
|
|||
hyperfocal distance
"M Barnes" wrote in news:q9-dnS6e4c3M3H7dRVn-
: Hm. Okay, I think I get it. How is the CoC computed for lenses then? And for film body/lens combinations? CoC is the diagonal divided by a number. This number can be choosen somewhat arbitrary, but it is in the region 1500-2000 I think. One standard value is 1740. Someone else said 1525. Both are probably valid choices. It all depends on your criteria for when you think a picture is sharp. /Roland |
#35
|
|||
|
|||
hyperfocal distance
|
#36
|
|||
|
|||
hyperfocal distance
|
#37
|
|||
|
|||
hyperfocal distance
"M Barnes" wrote in news:saKdnehg9qGnxn7dRVn-
: You just saved me a weekend with a tape measure in the back yard. Thanks. /Roland |
#38
|
|||
|
|||
hyperfocal distance
Dave Martindale wrote:
M Barnes wrote: Dave Martindale wrote: I was under the impression that [CoC and resolution] were related, in that the resolution test for a lens determined the CoC value for that lens when computing other values, for example, the HD for a given set of values. No. CoC is not a property of the lens at all. It is a measured or estimated tolerance of how blurred an image can be before the viewer, looking at a print, feels it is out of focus. The minimum blur spot size for a lens is normally quite a lot smaller than the CoC used in depth of field calculations. I'm not sure I follow this. Do you mean that the CoC is not a base level of resolution or acuity, but only a semi-useful construct, and that the blur spot is a more meaningful measure of lens acuity? CoC is not a measure of lens acuity *at all*. It is a measure of how badly out of focus something can be allowed to get before it seems unsharp to a person viewing the image under some assumed set of conditions. CoC is a useful construct for calculating DOF, not lens resolution. OK, got it. I need to do more reading to keep up, but I'm pretty sure I understand. If you have a lens whose blur spot diameter is equal to the Circle of Confusion for the sensor size you're working with, then *everything* will be borderline unsharp, even what ever is in best focus. Normally, the CoC will be several times the lens blur spot size if the CoC is reasonable and the lens is suited to the format. Ie., the lens acuity is much sharper, theoretically, than required to produce an image that appears sharp to the viewer. This allows for aberration, aperture, DOF, etc., etc. For this discussion, I'm holding print size at 8x10, lens focal length and all other parameters constant. The only variable will be sensor size -- 35mm and the D100 2/3 equivalent. That's an unusual choice, because you won't get the same field of view if you use the same lens focal length for 35 mm and for 2/3 frame. If you do this, the short answer is that the D100 will give you only 2/3 as much DOF as you would get with a full-frame sensor, at the same FL and same f/number. (This is because you have to enlarge the D100 image more). Yes, exactly. To start with, I'm looking for this artificial condition as a learning aid, for me. On the other hand, if you changed the lens focal length to keep the same field of view, so the D100 lens is 2/3 the focal length of the 35 camera lens, then the D100 would give you 3/2 times as much DOF as the 35 camera. (There's one lie in the above. Real DOF tables usually give distance from the film plane, not distance from the lens front principal plane. This makes the exact calculations messier, but there's little practical difference between the two methods for non-macro photography). Well, two things: I'm used to measuring distance to subjects with a tape from the focal-plane mark on the camera body, so that does apply to me, especially when I'm using extension tubes and/or macro lenses; Ok. I'd suggest you build your own DOF tables using a spreadsheet. It's not really that hard. If you're calculating things in terms of distances from lens front principal plane to subject, and from rear principal plane to film/sensor, you can use classic lens formulas directly. If you want to know subject-to-film distance, you need to add together the lens-subject distance plus the lens-image distance *plus the spacing between the two principal planes within the lens*. You may have to guess at the latter number. That's what I've done with the N2020. It won't be hard to create a new one for the D100. and ... the 64-dollar question -- Does all that add up to an adjustment in the HD when sensor size changes and all else is held constant? There seems to be a circle of confusion regarding the answer, and I'm not smart or experienced enough to figure it out, but you seem to be. Yes, see above. If you keep the lens FL constant, thus narrowing the FOV when sensor size decreases, the DOF also *decreases*. On the other hand, if you make lens FL decrease along with sensor size, so FOV remains constant, then the DOF *increases* with decreasing sensor size. Ah, that's the 128-dollar question I was just about to ask. I thought I understood, but wanted to make sure. If I create a shot with the N2020 using hyperfocal focusing, then switch to the D100 body and adjust -- say using a zoom -- FL to keep the shot the same, I would need to, or could, move the near focal point outward to compensate for the increased DOF. Thanks. This helps a lot. I appreciate your patience and knowledge. I pretty much figured that FL and FOV, being on opposite sides of the dividing line in the formulae cited by others, would signify a reciprocal relationship. Of course, if Nikon would come out with a full-frame sensor, I could forget all this. Until the medium-format DSLRs come out .... |
#39
|
|||
|
|||
hyperfocal distance
Roland Karlsson wrote:
M Barnes wrote: CoC is neither depending upon the pixel size nor the lens. It is a pure aestethic meassurement; how unsharp do human beings accept out of focus parts to be before thinking they are too unsharp. And tests have shown that the CoC is approx. the diagonal divided by 1500-2000 for most people and most pictures. I think I'm beginning to click into this concept from other discussions. I've never paid much attention to it, but read a Nikon tech paper years ago that got me off on the wrong track. I've had to root out the synaptical assumptions and replace them with new ones. Very painful. Not like finding out Santa was really mom and dad, but still painful. Also, I believe the OP was asking if it's required to adjust the "f" in your formula due to the _apparent_ focal length adjustments required by fractional-frame sensor form factors, n'est-ce pas? Yes he did. And, yes he has to. If the sensor has a cropping factor of 1.6, than the hyper focal distance is increased with a factor 1.6. Got it. Thanks. |
#40
|
|||
|
|||
hyperfocal distance
Roland Karlsson wrote:
CoC is the diagonal divided by a number. This number can be choosen somewhat arbitrary, but it is in the region 1500-2000 I think. One standard value is 1740. Someone else said 1525. Both are probably valid choices. It all depends on your criteria for when you think a picture is sharp. The older I get, the more forgiving I am, and the more lenses I need stacked up in front of me to see any difference at all. But I do like to play with DOF, hyperfocal focusing and macros, sometimes combining foreground macros with infinity backgrounds, so this is a technical point that has some interest to me. Obviously I need to do more reading, but thanks for the help. The two numbers you cite helps explain why there are different values in the tables I've seen on various Websites. |
Thread Tools | |
Display Modes | |
|
|