brian_caldwell

If you are stopping down the 100mm lens, then try stopping down the 50mm instead. This should drastically reduce or eliminate the vignetting you are getting. Also, macro lenses are not the best to use in this application because they are less tolerant of stop shifting. Its generally better to use a more compact telephoto lens focused to infinity.

Having some experience designing 300-times zooms I would say that 11x zooms don't qualify as "mega". However, I wouldn't be surprised if the 18-200 has some image quality compromises in order to ensure compactness and reasonable cost.

 

There are theoretical reasons why zooms may ultimately suffer in comparison to primes. In particular, there is a theorem in optics that a lens can only be perfectly corrected for a single magnification. Since a zoom lens must have at least one lens group that changes magnification during zooming, there is a theoretical limit to how good a zoom lens can be. You can actually use geometrical ray tracing software (OSLO) to calculate what the theoretical limits are for a given type of zoom lens.

 

In practice, this theoretical limitation is often dwarfed by diffraction. In other words, diffraction will often limit performance long before it is limited by the geometrical optics limitation mentioned above. This means that it is possible to design non-trivial zoom lenses that really are diffraction-limited.

 

It's true that some zoom lenses are better than many primes within their range. This is certainly the case for the 17-35/2.8, and perhaps a few others. I would speculate that the reasons for this are 1)prime designs are old, 2) primes are intended to be more compact or faster or both, 3) the wide angle zooms are allowed to get very large, which is a great help to the designer.

 

Of course, you could always take the 17-35/2.8 design and re-optimize it for a single focal length to improve performance at that single focal length, but then you would have a very bulky and costly prime lens. It's no accident that the 21/2.8 Zeiss Distagon is very bulky and expensive!

True wide-angle zooms must have a negative powered front group, and designs of this type have been limited to a zoom ratio of about 2:1. In extreme cases you can push the zoom ratio to maybe 3.5:1.

 

There is a very special technique for expanding the zoom ratio in ultra-wide zooms. The technology is patented, and it is unlikely you will see anything like it from Nikon or Canon for the next 20 years.

 

There is also a sort of brute force technique, basically add a wide-angle afocal attachment to the front and re-optimize, but it results in absolutely gigantic designs you could never afford. A few TV zooms have been made this way, and its possible to get a 10x zoom ratio with a 90-degree FOV at the wide end, which would be 21-210mm equivalent). If scaled up for DX format the front element would be about 10" in diameter.

Get these:

25-50/4

50-135/3.5

50-300/4.5 ED

"What's the point? You get the same DOF at the same magnification and effective aperture, no matter the focal length or attachments to the lens. I would not recommend the use of anything like a TC or auxillary lens with the 55 Micro - extension tubes and possibly a reversing ring would be OK since the optics are not compromised with added glass."

 

The 55 Micro will be compromised by extension tubes because coma is introduced. The CRC scheme in this lens is designed to correct coma for all focusing distances from infinity down to 1:2 (manual focus version). The same scheme will work just fine down to 1:1, but the secondary helicoid needs to be re-designed.

 

Whether going to 1:1 is best done with a 2x TC or an extension tube is an interesting question. I tried some raytracing using Nikon patent examples, and the TC approach showed slightly better MTF wide open. However, the TC does seem to add a little lateral color, so stopped-down performance should favor the extension tube approach. This is because stopping down will rapidly diminish coma, but will do nothing to lateral color.

 

Reverse-coupling two 55/2.8's focused to infinity will result in exact 1:1 magnification, very fast relative aperture, and extremely sharp imaging near the axis. However, I expect there will be severe vignetting and off-axis aberrations due to the deeply recessed front element in this lens.

Carlos Miami's shot clearly shows that this lens *does not* have good bokeh when used wide open. I'm assuming its a wide-open shot. The bright ring effect is inherent to f/2.8 Tessars.

Alex Lofquist wrote:

"Come to think about it, I believe that I got the best resilts by controlling the aperture of the front (shorter FL lens). It makes sense in that the best location is closest to where the Chief Ray, would cross the optic axis, and that would be near the shortest FL lens. (I expect that the ideal place should be between the two lenses if it were possible.) The only combination that I liked was a reversed Nikkor 35mm f/2 on the front of a Nikkor 85mm f/1.8."

 

Hi Alex:

If you reverse couple two lenses the chief ray will cross the axis only once, and by definition the place that it does this is the aperture stop. You can choose to place the aperture stop either in the front lens or the rear lens, simply by stopping down one and not the other. In virtually all cases the correct place is in the front.

 

Ignoring practical difficulties, placing an aperture stop between the two lenses would not necessarily result in better performance, particularly if a wide angle lens is used in front.

 

One foolproof way to avoid vignetting is to put a pupil relay between the two lenses. Then you could put any lens on the front and still be able to stop down using the rear lens. Unfortunately, it would cost at least $25k to do the design and manufacture a prototoype.

Vivek Iyer wrote:

"Brian, There appears to be a heavy use of glass incorporating Tantalum Oxide (usually clubbed with Lanthanides) from a reference to it by Vivitar in their sales brochure."

 

Hi Vivek:

In the prescription I have none of the glasses have an index greater than 1.7, and by my reckoning the two crown glasses used are Hoya LAC9 and LAC12. IIRC, the early Vivitar lenses used alot of Hoya glass. I wouldn't be surprised if Tantalum were an ingredient in one of these glasses, but I don't know for sure. In the Hoya catalog crown glasses with an index higher than 1.72 are given a name beginning with "TAC" instead of "LAC", which perhaps indicates a higher percentage of Tantalum relative to Lanthanum. However, I see no reason why Tantalum wouldn't be used in smaller amounts in the "LAC" series.

 

At any rate, Tantalum, like Lanthanum, can't be used to make "ED" glass because it gives the wrong sign for the partial dispersion. As a result, using Lanthanum/Tantalum containing glasses in the positive-powered elements generally results in worse secondary color correction than you would get using more normal glasses such as BK7. However, in many cases the benefits of these high index crowns far outweigh minor secondary color problems.

 

To make "ED" glass you need Fluorine and/or Phosphorous.

In my experience - both raytracing and taking photographs, the conventional wisdom that you should leave the reversed lens wide open is completely wrong. If you want to stop down, then you should use the shorter EFL front lens, not the longer EFL rear lens. You will get less vignetting and less image degradation due to lateral color and other problems. Its very easy to experiment with both approaches. I've had success (no vignetting, good corner sharpness) when reversing wide angle primes (20, 24, 35) onto longer primes (50, 85, 105, 135, 200) so long as I stop down the shorter lens and not the longer one.

 

The reason is related to the pupil matching thing that Bjorn mentions. Basically, when you reverse-couple two lenses and stop one of them down, the lens that you leave wide open is having a huge stop shift forced upon it. Longer lenses, such as a 200/4 or 105/2.5, are much more tolerant of this stop shift than a short lens like the 24/2.8. Zooms are almost always very intolerant of stop shift, which is why I would avoid them in reversed-lens work.

"And, oh yes, you can use this adapter with any other macro lenses that you have. There is not loss of transmission due to the 3 element glass in it (but there will be light loss due to the extension). It is put there to correct for spherical aberrations at close distances. The glass used in the 90/2.5 Series 1 lens includes ED grade lenses. They were quite expensive when they were new. "

 

Hi Vivek:

I've always liked this lens even though it does have flare problems.

 

IIRC, E. Betensky once told me that the patent prescription for this lens was the actual production design (U.S.P. 3,942,875). If so, then the glasses used are the usual assortment of lanthanum crowns and lowish-index flints, which are typical for double-Gauss designs. This is actually the opposite of what you need to do to correct secondary color, which only goes to show that ED glass just isn't necessary for this focal length.

 

Also, I would guess that the 1:1 adapter corrects for coma and not spherical aberration, since the former tends to be much more of a problem for a modest aperture lens focused away from its designed conjugates.

"Coating the front surface will improve transmission a little but will have no effect on flare. The latter because light reflected from the front surface, um, goes away, doesn't bounce around incide the lens."

 

Flare due to ghost images is most often caused by a pair of reflections. The front surface can certainly participate in the creation of ghost images. When designing a lens for minimal ghosting you have to consider all possible pairs of reflections, and this increases the complexity of the design process a great deal.

Although perhaps not applicable to rangefinder optics, one of the best uses for aspheric surfaces is to correct distortion. In this application the aspheric tolerances are very loose, so the chance of success is high. On the other hand, using an asphere to correct spherical aberration results in very tight tolerances.

'Here's where I'm hoping the Zeiss ZF 50 f/1.4 will shine. With a 58mm filter ring size, they should be able to make an f/1.4 lens that doesn't have to have so much curvature of field and distortion to gather the light.'

 

I'm afraid the ZF 50 f/1.4 will suffer the same several percent of barrel distortion that all fast 50mm SLR lenses do. The reason is related to the fact that the working distance on these lenses is actually a bit longer than is comfortable for the design. Some distortion is allowed into the design to maintain the necessary working distance while keeping image sharpness high. Slower 50mm lenses work well with the required SLR working distance, which is why they have less distortion.

 

Of course, if you scan film or shoot direct digital then correcting distortion during post-processing is both straightforward and extremely accurate.

"Just wondering if the distortion gets lower with smaller apertures (I mean, higher f-numbers) or not."

 

Not. The two aberrations that don't change when you stop down are: distortion and lateral chromatic aberration.

The 18-200 is DX format only. Therefore, distortion is irrelevant. Why? Because you can completely correct it with software, and the effect on image sharpness is undetectable.

 

If the lens could be used with film the distortion thing might have some relevance.

I like this lens a lot. As other users report, sharpness is extremely good, which is why it works well with teleconverters. I use mine with a D1x. For $550 I would grab it fast.

The 24/2.8 is optically better than the 24/2.0 mainly because the latter has more color fringing. Although I have both of these lenses, I always use the 17-35/2.8 when I want maximum image quality at 24mm.

"Any fast lenses that consist entirely of spherical lenses will show coma to some degree. You should just live with it. Slight underexposure would mitigate the phenomena."

 

Akira:

Coma is virtually never present in any photographic lens, regardless of speed. You are probably thinking of oblique spherical aberration. The latter can be reduced with aspherics, but even better is to use a reversed telephoto construction and forget about compactness. Its amazing how good lenses can be if you don't restrict their size too much.

Series E patents that I've found are all assigned to Nikon, not Kiron.

"I know I have to keep the reversed lens wide open to avoid vignetting. "

 

This is widely believed, but is in fact completely false. The best way to avoid vignetting is to leave the primary lens wide open and stop down using the reversed lens. Reason: the primary lens is generally has a larger entrance pupil and can withstand stop-shift to a much greater degree than the shorter focal length reversed lens. I've verified this by both ray tracing and experimenting with numerous combinations of lenses.

 

Simple, compact designs are best for the primary lens: 200/4 AIS, 135/2.8, and 105/2.5 are all good choices in my experience. I would avoid using a macro lens or a zoom lens as the primary lens because these may be less tolerant of the extreme stop shifting that necessarily takes place with the lens reversal technique. Remember that the lens reversal technique works extremely well with two lenses corrected for and focused at infinity. Macro lenses offer no advantages here.

 

One caution here regarding AI vs. AIS: the AI versions of the longer focal length lenses may have a nice stiff feel to the helicoid, but the AIS helicoid is in fact a far superior mechanical design that will resist focus creep. Years ago I bought a 200/4 AI in pristine condition just to do ultra macro via lens reversal, but found it completely useless for the purpose. I then got an AIS version of the same lens and found it completely free of focus creep despite the fact that its helicoid had a much looser/less-damped feel.

 

If you use one of the above-mentioned primary lenses you can use virtually any shorter focal length prime lens for the reversed lens. I would avoid zooms since they will generally exacerbate the pupil matching problem. However, as always, you should let careful experimentation be your final guide.

I agree with Vivek. I've always felt that the 50/1.2 and 50/1.4 were very similar at like apertures. But the 50/1.2 has the bonus of having a unique "look" to its images at f/1.2. I also think that the 50/1.2 makes a more logical pairing with the 50/1.8 or even better the 55/2.8 in case you want to get two different 50's. As a general purpose lens, I even think the 50/1.2 has some advantages over the 58/1.2 Noct because it lacks the Noct's field curvature.

"Cool stuff for Bob and Brian:

 

"Oblique spherical aberration"

 

 

"Definition:

 

A type of aberration in which a symmetrical light patch is formed at points that do not lie on the lens axis. When the image moves away from the center of the field, the patch increases in size, aiding in the formation of coma. Also known as abaxial spherical aberration." "

 

Kelly:

Some people will claim that any assymetrical off-axis aberration is "coma". Technically this is not true, and is just an oversimplification of what is really going on.

 

Of course, the distinction is really only important to lens designers, who really have to know what sorts of aberrations are present in order to correct them. I look at transverse ray aberration plots constantly in the course of doing design work, and there really is a big difference between coma and oblique spherical aberration, in terms of how the aberration arises, what effect it has on images, and how you fix it.

 

High order aberrations are not the sort of thing that can be settled or understood with a simple definition pulled out of cyberspace. Even excellent optical texts such as Warren Smith's "Modern Optical Engineering" only scratch the surface of this topic. Its the sort of thing you really only get comfortable with after working in the field for a decade or more.

"ALL the worlds finest lenses will show abberations far off axis, when the lens is wide open, under harse nightime shots like these. The abberations are a mixture, usually a blend of several types. There will almost never be just one type, and none of all the rest. These simple examples are for college texts, and not complex blends.Coma and astigmatism"

 

A well-designed fast double-Gauss lens will have virtually zero coma or astigmatism at its design magnification. However, it will certainly have a ton of saggital oblique spherical aberration unless you use one or more aspherical surfaces to correct it. This is one case where the corner aberrations really are almost purely of one type and not a complex mixture. For many other lens types your comments are correct, however.

"Doesn't look like coma. Coma should extend along a line between the point and the center of the image - and it should be Coma shaped!

Could be tangential astigmatism or possible (but less likely) out of focus highlights modified by the shape of the iris which is distorted by vignetting and/or oblique viewing near the edge of the frame."

 

Bob:

You're right that the aberration is not coma. However, its not astigmatism either.

 

The aberration that causes these butterfly-shaped blobs is saggital oblique spherical aberration. Oblique spherical aberration is very similar to ordinary spherical aberration except that it is zero on-axis and increases as you move off axis. Tangential oblique spherical is normally absent due to vignetting at wide apertures. So you were on the right track with regard to vignetting. These are the classic aberrations of virtually all double Gauss lenses.

"That aperture used is the key to the "surprises" here. Try comparing them wide open, especially at f2.8. :-)"

 

As I'm sure you're aware, lateral chromatic aberration (CA) is independent of aperture. So, the aperture used cannot be the "key" to at least some of the surprises.