Circle of coverage at small apertures?greenspun.com : LUSENET : Large format photography : One Thread
Published circles of coverage for LF lenses are apparently standardized at f22. We know that the area of coverage increases as the lens is stopped down, but by what factor? Would it be a percentage of the dismeter? If so, what would the percentage be, or is there a formula, or what?
-- Richard Deimel (Bbadger@aol.com), May 24, 2001
It depends on the physical obstructions to the light path caused by the lens barrel.
-- sheldon hambrick (email@example.com), May 24, 2001.
Richard, that's a good question, I assume that the maximum circle of good definition occurs at f22 for most lenses, and then remains constant as you continue to stop down, otherwise the manufacturers would probably let us in on the other f stops data as well. This is just my assumption, don't know for sure. I've never seen the maximum circle of definition shown at anything but f22, so it could also just a standardized comparision point. You could of course take the test shots yourself at smaller stops and see if there's any difference beyond f22. Anyone know the real answer?
-- Michael Mahoney (firstname.lastname@example.org), May 24, 2001.
Sheldon's answer is correct. It depends on the lens construction, and has very little to do with the circle of acceptable definition. Most modern lenses will vignette before the point where the image quality is totally unacceptable. Their designers engineer them to do that.
If you look through a lens from the back, set to a mid aperture, and slowly rotate it about an axis parallel to the iris, you'll see that part of the iris aperture becomes obscured by the perimeter of the lens at some angle. If you now stop the lens down further, that angle becomes greater before the aperture is obscured. This is the reason why coverage increases with stopping down, and not that the area of best definition carries on increasing.
The point at which vignetting occurs varies according to the optical and mechanical construction of the lens, so there's no fixed formula that can be applied.
-- Pete Andrews (email@example.com), May 25, 2001.
OK, you have convinced me that the expansion of the circle of coverage is limited by the physical design of the lens, but that really doesn't answer my question. As we stop down the lens, the circle expands. Are you all saying that there's no logic to its expansion, but that in each lens it expands at a different rate, determined by the physical design of that particular lens?
-- Richard Deimel (Bbadger@aol.com), May 25, 2001.
I thought it depended on design factors other than physical obstruction designed into the lens, for instance that Dagors have sharply increased coverage at small apertures, while Heliars are supposed to have a more evenly illuminated circle of coverage throughout the range (or so I thought, I haven't actually tested for this).
-- David Goldfarb (firstname.lastname@example.org), May 25, 2001.
You guys are making this too hard. Determining area of coverage is a simple physical problem, not an optical one. If you look at your (for example) f4.5 12 inch lens wide open and hold it at an angle so you can just barely see light through it you will be looking at a tiny sliver shaped aperture, pointed on the ends and bowed in the center. This is, obviously, what a circle looks like when viewed from an angle. Wherever your eye is at this point is the farthest extent that light makes it out on the edges. Even though the lens is wide open, only this sliver-shaped aperture is available way out on the edge. So, while an image is being produced out there, in comparison to the image produced in the center the light fall-off is severe. (And the image may not be too good even taking account for the light fall off.) An image properly exposed in the center will have no exposure to speak of way out in the extreme edges.
Now close the aperture down to f22, 64, whatever, while you are still looking through the lens from an angle. Notice that your sliver of light becomes smaller, and more like a circle (an oval, actually). But notice that now the aperture you have out on the edge is not really that much smaller than what you have in the center. So, at smaller stops the light fall-off will be less in comparison with the center than it is at larger stops.
A lens doesn't actually give you more coverage at smaller stops, it's just that the fall-off becomes less severe relative to the center the more you stop down. f22 is an arbitrary point the lens makers have chosen to define where their arbitrary standards for light-fall off are met, and also happens to be an aperture where lf lenses tend to perform fairly well. The only way to determine what the image circle is for you is to establish your own arbitrary standards.
Optical quality suffers on the edges depending on the lens design, but the point where it suffers has little to do with aperture. That is, you'll have a better image at f22 than f4.5, but light will still be cast on the circle's edges at either aperture, and again you need to establish your own arbitrary standards for when a lens makes an image that suits your purposes.
-- Erik Ryberg (email@example.com), May 25, 2001.
I've always assumed that f22 is the f-stop at which manufacturers quote image diameter, because it's this f-stop that many lenses become diffraction limited.
An interesting example that may pertain to this thread is the Fujinon 250mm f6.7 versus the Fujinon 250mm f6.3. Both lenses are plasmats, the f6.3 is wider than the f6.7, yet the f6.7 is quoted to have the larger of the two image circles. It would appear that image circle depends, at least in part, on the design of the lens.
Certainly, design can have a large impact on the image circle. Compare a Symmar to a Super Angulon, or to a double-Gauss design.
-- neil poulsen (firstname.lastname@example.org), May 26, 2001.
Erik's is the first explanation that really makes sense to me. It's not that the actual circle of coverage varies (that would be limited by the optical design), but that the USABLE circle of coverage will vary. However, having gotten that far, the original question still remains: does it vary by any measureable factor?
-- Richard Deimel (Bbadger@aol.com), May 26, 2001.
Predictable, without any knowledge of the optical construction of the lens? No.
Someone mentioned Dagors and Heliars earlier. These are old fashioned lenses, where the coverage is left to fend for itself, so to speak. Modern lenses have the circle of coverage more tightly controlled by design, since there's no point in having a large circle of light that's unusable due to poor definition. This just increases flare and lowers image contrast needlessly.
A well-designed lens will have its image circle clipped by internal baffles or by barrel design, at the point in the image circle where quality falls below a certain standard.
-- Pete Andrews (email@example.com), May 26, 2001.
I know that's the modern engineer's view, but personally, I'd rather have a soft corner and a little extra flexibility in movement than a vignetted corner.
-- David Goldfarb (firstname.lastname@example.org), May 26, 2001.
Erik's fine (and finally, understandable!) explanation, it seems to me, implies that a data chart might be constructed which would compare a lens' light transmission both in the center and in the corner, as the aperture is reduced. This chart (or its implied graph) would illustrate the lessening edge falloff as the aperture shrinks.
Does this falloff approach zero as the diaphragm size also approaches zero?
-- Wellesley Browning (email@example.com), May 31, 2001.