Close-up filters

[ray_parkhurst]

I'm familiar with so-called "close-up filters", and own a couple types, but am not sure about their optical properties. Someone over on dpreview is claiming that there is no "light loss" when using them. This means that they are able to increase magnification while not reducing effective aperture. Is this correct?

[Lou Jost]

A reversed close-up lens on a main lens is basically a lens combo. As a lens combo, the aperture of the rear lens doesn't matter if the front lens is fast enough.

[ray_parkhurst]

Yes, makes sense. The "close up lens" becomes the objective and dominates the aperture equation.

[rjlittlefield]

This means that they are able to increase magnification while not reducing effective aperture. Is this correct?

Yes, the above is correct.

But these two are not correct. I suspect a couple of serious typos.

As a lens combo, the aperture of the rear lens doesn't matter if the front lens is fast enough.

The "close up lens" becomes the objective and dominates the aperture equation.

With respect to "close up" lenses, the rule is that aperture of the front lens doesn't matter if the front lens is fast enough.

In other words, in that case the aperture of the rear lens is what limits.

The other case is when the aperture of the front lens limits, but then the rule is that the aperture of the rear lens does not matter, if the rear lens is fast enough. This is typical when you're stopping the front lens.

--Rik

[Lou Jost]

Rik, those weren't typos, I am genuinely confused. I reached that conclusion by analogy with coupled lenses.

I thought a close-up filter is basically the front lens of a combo? In that case the EA is determined by the aperture of the front lens and the focal length of the second lens, according to the standard formula, as long as the rear lens is fast enough not to be limiting.

Is this analogy not correct?

[rjlittlefield]

Rik, those weren't typos, I am genuinely confused. I reached that conclusion by analogy with coupled lenses.

I thought a close-up filter is basically the front lens of a combo? In that case the EA is determined by the aperture of the front lens and the focal length of the second lens, according to the standard formula, as long as the rear lens is fast enough not to be limiting.

Is this analogy not correct?

The analogy with combos is appropriate in general, but your analysis is wrong in this specific case. The problem is that you're treating the combo as if it were stopped in the front, when it's actually stopped in the rear.

Close-up filters are so wide that they do not provide the limiting aperture. The limiting aperture remains the one inside the rear lens, so this combo is stopped in the rear, not the front. Adding the close-up filter does not affect the exit pupil (because the limiting aperture is behind the added component), and because the exit pupil remains the same, so does the EA.

In the rare case that a close-up filter provides any limiting aperture at all, that manifests as vignetting in the corners of the image. Over all the non-vignetted part of the frame, the limiting aperture is inside the rear lens, typically the adjustable iris of a conventional camera lens.

--Rik

(Edited for clarity)

[Lou Jost]

Of course! That makes good sense, the rear lens will be the limiting aperture if the front lens is wider than the rear lens (which it usually is). Thanks for the correction.

[ray_parkhurst]

All makes sense, and the key is the tube lens is the limiting aperture. So taking this a bit farther, if you add a close-up lens, and simultaneously reduce tube lens extension such that the magnification is the same, then you will actually increase the aperture.

I guess the limiting factor here is the quality of the close-up lens, and how it interacts with the tube lens, resulting in final image quality.

[rjlittlefield]

So taking this a bit farther, if you add a close-up lens, and simultaneously reduce tube lens extension such that the magnification is the same, then you will actually increase the aperture.

I think you have in mind reducing the extension while physically changing nothing else. In that case, then yes, usually you'll end up with a larger EA because the exit pupil will be moved closer to the sensor.

There is one notable exception to that idea. If the rear lens is telecentric on the sensor side, like some lenses designed for MFT, then there will be no change in EA because the exit pupil is located at infinity, so a few cm of movement makes no difference in how the exit pupil looks to the sensor.

--Rik

[Lou Jost]

There is one notable exception to that idea. If the rear lens is telecentric on the sensor side, like some lenses designed for MFT, then there will be no change in EA because the exit pupil is located at infinity, so a few cm of movement makes no difference in how the exit pupil looks to the sensor.

That's an interesting observation. So using extension on an MFT lens doesn't increase diffraction?

[elimoss]

If the lens is telecentric on the sensor side, then the angle of light rays through the exit pupil, from the perspective of the sensor, doesn't change much with extension. If the incident light rays don't change, the EA doesn't change.

So then it does stand to reason that the increase in diffraction due to extension would be much less than a lens that is not telecentric on the sensor side...

[rjlittlefield]

That's an interesting observation. So using extension on an MFT lens doesn't increase diffraction?

If the lens is truly telecentric on the sensor side, that should be true.

This counterintuitive result is the flip side of the situation illustrated at https://www.photomacrography.net/forum/ ... 989#249989 , panel 3.

There, I wrote that

Another odd feature of telecentric [on the object side] lenses, which I realized only very recently, is that their subject-side NA is not changed at all when you refocus the lens by changing extension. In other words, putting the lens closer or farther from the subject does not change the angular width of the entrance cone.

Here is an illustration of that effect, showing the same lens and aperture operating at 1.7X and at 3X, refocused by extension:



This behavior contrasts sharply with ordinary lenses, where placing the lens closer to the subject makes the entrance cone have a wider angle.

The effect can be easily understood in one way by concentrating on the entrance pupil. With a telecentric lens, the entrance pupil is at infinity, so of course moving the subject by a short distance closer or farther has no effect on how the subject sees the entrance pupil.

But on the other hand, if the angle of the entrance cone does not change, then it must be that the width of the entrance cone, at the lens, does change, and in exactly the right way to keep the angle constant. In other words, as you change the extension so as to focus closer, somehow the optical system automatically "stops down" by exactly the right amount to compensate for the change in lens-to-subject distance. Looking at the behavior this way simply makes my head hurt!

In the very simple system animated above, it is straightforward though tedious to work through an analysis based on scaling of triangles to confirm that indeed the width of the cone at the lens is exactly what it needs to be to keep the entrance cone angle fixed.

For the more complicated system diagrammed earlier as "Setup 4", I would not want to work through a detailed analysis based on triangles! Nonetheless the argument about "entrance pupil at infinity" seems simple and foolproof, so I'm quite confident the triangles would work out OK also, if only I slogged through the analysis carefully enough.

Here is a telecentric-on-the-sensor-side illustration, built using WinLens3D's "Winlens Library 2002\Telecentric_lenses\WLTC001.SPD".



Notice that with the longer extension, the ray bundles inside the lens get wider, while the EA remains fixed at 10.0006. (EA is labeled "Effective F/Nos", found in the lower right corner of the upper left panel.)

From a causal standpoint, what's happening is that as the subject gets closer to the fixed-diameter entrance pupil, the entrance cone angle gets wider. That is, adding extension behind the lens simultaneously increases the magnification and the subject-side NA. Slogging through the details, it seems miraculous that the magnification and NA increase together at exactly the right rate to keep EA constant. But if you turn the problem around and think about the exit pupil location being at infinity, suddenly the conclusion seems simple even though the details still look miraculous.

Another way of "explaining" the same conclusion is to use the standard equations involving pupil ratio.

In particular, f_e = f_r*((m/P)+1), where P = pupil ratio as rear/front diameters. A lens that is telecentric on the sensor side has an infinitely large exit pupil, located at infinity, so P=infinity, m/P=0, and f_e = f_r independent of magnification.

This still makes my head hurt, but not as badly as it would if the analyses were inconsistent.

--Rik