Judging high-NA infinity-corrected objectives

[ChrisR]

FWIW I didn't see a contradiction either.

I think it's reasonable to assume that a
40x NA 0.75 objective on 200mm tube, might perform better than a
20x NA 0.75 objective on 400mm tube, because you're enlarging just a portion of the 20x objective's image.
A 20x NA is harder/more expensive to make, particularly if its image circle has to be the same diameter.

"Theoretically" ie if everything is only aperture-limited and aberrations are insignificant in both, then the images will match. Yes?

[Lou Jost]

"Theoretically" ie if everything is only aperture-limited and aberrations are insignificant in both, then the images will match. Yes?

That's my understanding. But you're right, there might be design constraints that sacrifice central resolution to achieve a wider image circle. And that is what mkzz is saying too. That will have to be answered by experiment.

[mjkzz]

What part of that do you think is self-contradictory? It would be helpful to me if you explained what was the contradiction. Thanks. The statements are based on my (non-analytic) understanding of how infinity optics work, and my experience using my objectives and coupled lenses on different formats from MFT to APS to FF (I use all three sensor sizes). I'd love to see a more technical discussion from someone like Rik.

Re-reading what you quoted there, I should probably not have used the word "exactly", which is a very strong word. But I regularly use tube lenses from 90mm to 400mm with my objectives, and I do not notice a general degradation of the image due to tube lens focal length beyond what you would expect from the different magnifications of the image, though of course no real tube lens is perfect, so some are better than others.

Edited to add this: if you look through this forum, you'll see many beautiful photos by people using tube lenses of different focal lengths than 200mm, both lower and higher, with excellent results.

Whatever, the over all tone in the original post sounds like favoring stretching an objective is better.

I don't understand how you got that from what I wrote.

Please read the following statement of yours.

Using a longer tube lens gives exactly the same image as a shorter tube lens (if both lenses were perfect), just larger. So the CA of the objective is blown up along with the rest of the image.

So a 100mm tube lens on MFT and a 200mm lens on FF give the same image with the same amount of CA in relation to the frame width. But yes, if you are comparing long vs short tube lenses on the same frame sizes, the CA would be more apparent on the more magnified image.

Now, please try to explain the part "So a 100mm tube lens on MFT and a 200mm lens on FF give the same image with the same amount of CA in relation to the frame width.", maybe even post it. You are more experienced person, I think you would catch your error in first few sentences. I give you a hint, why do you think they would give same amount of CA if, per you, CA gets blown up by higher magnification?

If you could not arrive at what's wrong with that, I will post my explanation.

[mjkzz]

FWIW I didn't see a contradiction either.

I think it's reasonable to assume that a
40x NA 0.75 objective on 200mm tube, might perform better than a
20x NA 0.75 objective on 400mm tube, because you're enlarging just a portion of the 20x objective's image.
A 20x NA is harder/more expensive to make, particularly if its image circle has to be the same diameter.

"Theoretically" ie if everything is only aperture-limited and aberrations are insignificant in both, then the images will match. Yes?

Chris, the self contradictory part is with one of the statement of Lou regarding CA on two systems. Please read my reply.

As far as your other questions, I have to admit I have ZERO knowledge about optics, so I will take your word for it. :-)

[ChrisR]

I think the devil lies in the meaning of
"in relation to the frame width".

What "frame"?

If you change cameras, and tube lens to suit, the frame is bigger, the aberrations are the same, relative to the frame size.
If you don't change cameras everything including the subject and all aberrations are enlarged relative to the frame size.

I doubt either of you would disagree, so it's just how the words are interpreted?

[Lou Jost]

the same amount of CA in relation to the frame width

Maybe you are ignoring my phrase "In relation to frame width". The FF image is twice as wide as the MFT image, but it is the same image, just magnified by a factor of two. If a purple fringe on some part of the image has a width of 1% of the frame width on the MFT images, it should have a width of 1% of the frame width on the FF image.

Edit- I didn't see ChrisR's post while writing the above. I agree with him.

[mjkzz]

I thnk the devil lies in the meaning of
"in relation to the frame width".

What "frame"?

If you change cameras, and tube lens to suit, the frame is bigger, the aberrations are the same, relative to the frame size.
If you don't change cameras everything including the subject and all aberrations are enlarged relative to the frame size.

I doubt either of you would disagree, so it's just how the words are interpreted?

Yes, agree, in terms of percentage.

[Lou Jost]

That's what "relative to the frame width" meant; I was trying to be precise. Glad we agree.

[mjkzz]

That's what "relative to the frame width" meant; I was trying to be precise. Glad we agree.

That's what "relative to the frame width" meant; I was trying to be precise. Glad we agree.

Yes, I am glad we agree.

But I think in your original post, it is " in relation to the frame width", English is not my native language, so it did not ring a bell as "relative to the frame width" which definitely means some kind of ratio.

Just to be clear, for those readers who are new to this kind of topic. The absolute amount of CA should be twice as much for an objective on a 200mm tube lens on a FF vs the same objective on a 100mm tube lens on a MFT even though you get the same image. And this is subject to "CA gets blown up when magnification is larger" which is something I do not know as fact but it seems to make sense.

[mjkzz]

That means you can use the 20x 0.75 objective at 40x or even more, using a 400mm tube lens, with no degradation relative to a purpose-built 40x 0.75 used with a 200mm tube lens. This 20x pushed to 40x should outperform 40x 0.65 objectives.

So continue on.

And to be scientific about it, making all parameters the same except the part we compare, lets assume the 20x 0.75 produce same amount of CA as a 40x 0.75, pushing the 20x 0.75 up to 40x 0.75, according to your assumption about CA being enlarged as mag goes up, it means the 20x 0.75 is at disadvantage as it would produce twice as much CA regardless of camera used.

OK, maybe a 20x 0.75 has less CA, but how much less? to break even with 40x 0.75, it has to have less than half according to your assumption.

Anyway you play it, it seems it all depends which objectives you pick for 20x 0.75 and 40x 0.75 and this is getting rather arbitrary to a point it is meaningless.

Plus, the longer focal length tube lens tends to be longer, so it is even harder to operate (vibration, space, acquisition of longer focal length tube lens, etc, etc)

[Chris S.]

I don't see why this conversation has gotten so involved. Personally, I liked Lou's answer.

If you take an objective designed for a 200mm tube lens, but mount it on a 400mm tube lens on the same camera, you effectively crop the field of view and enlarge that crop to fill the sensor. This should not add chromatic aberration per se (unless the 400mm tube lens has flaws that add some), but we're likely to get a clearer look at that CA, so it may seem more apparent.

On the other hand, since some objectives demonstrate more CA toward the periphery of the image circle, and since the 400mm tube lens uses a more central portion of the image circle, CA might actually be less visible with the longer tube lens.

If we use a shorter tube lens than 200mm, much of the above applies, but in reverse. Given a 100mm tube lens on the same camera, we're doing the opposite of enlarging--shrinking the projected size of any element in the view field, even though the CA is still there, it might be less apparent.

On the other hand, since in this case, we're using a larger portion of the lens' image circle--and outer portions often have more aberrations--these aberrations may seem worse with the 100mm tube lens.

For photographers doing this, the size or the image circle and quality-falloff rate as one moves away from center become important, and unfortunately these characteristics are not adequately published on. For those who like to use objectives at lower than nominal magnification, this information is crucial.

In a portion of Lou's example, he adds a variable: Comparing two camera formats, one with a sensor half the size of the other. A 100mm tube lens on the half-size sensor provides the same field of view as a 200mm tube lens on the full-size sensor. In this case, we'd expect the same CA (again, unless one tube lens adds more CA than the other).

To my mind, for this particular discussion it just confuses things to mention the NA of the objective. Nothing I've written above involves the NA. This said, NA is crucial to the original point higher in the thread. It's easy, but a bit naive, to think of microscope objectives primarily by their specified magnification, with other specifications largely ignored. Numerical aperture (NA) is arguably a more important spec, since, with a well-designed lens, NA determines resolution. How well the lens deals with distortions, such as CA, probably comes next--those of us who have APO objectives seem very fond of them.

Is seems that NA has come up in this conversation due to the availability of low-cost 20x/NA 0.75 objectives. As we know, 0.75 is a very high NA for a 20x optic. By way of comparison, my Mitutoyo 100x has an NA of 0.70. So theoretically, the 20x/0.75, on a 1000mm converging lens, has a chance of providing the same field of view as my 100x/0.70 Mitty on its nominal 200mm tube lens, while slightly outresolving the 100x Mitty. Would it actually work better? I'd bet against it--but I don't know. We'd be pulling the 20x so far out of its design intent that otherwise hidden issues might appear. This said, it might happen to work fine. If anyone cares to perform this experiment, that someone won't be me--a 1,000 mm lens could easily cost $10,000. For far less, one can buy a Mitty 100/0.70. Or the Mitutoyo 100x/0.9 for about $6200. Also, a 1000mm lens would be very bulky--yuck. And the 20x/0.75 lenses discussed have short working distances--yuck again!

This said, the original thought of running the 20x/0.75 on a 400mm converging lens to operate as an effective 40x/0.75 has much more merit than the example I gave above, as good, inexpensive 400mm lenses are not uncommon. The limited working distance would still be there. Still I'd enjoy seeing the results of such an experiment.

Cheers,

--Chris S

[Pau]

In the very near future I will be able to compare the Nikon 20/0.75 with a 2X magnification changer against a native 40/0.75 although it will not be a good test: modern infinite plan apo vs old finite plan fluorite; direct projection vs afocal with compensating eyepieces...apples to oranges.

At first tests the 20/0.75 is brilliant: APSC excellent coverage at 20X, no visible CAs and it seems to outresolve the 18mpx sensor. As a bonus I have got it working well with my weird DIC equipment

The results I expect are: similar resolution but less CA with the Nikon 20X working at 40X. Unfortunately I don't own the adequate microscope for it and I can't use the trino head.

[mjkzz]

If you take an objective designed for a 200mm tube lens, but mount it on a 400mm tube lens on the same camera, you effectively crop the field of view and enlarge that crop to fill the sensor. This should not add chromatic aberration per se

So you are saying amount of CA does not change when magnification gets higher! If this is the case, sure, Lou's answer is better in this case. But this is contradicting what Lou is saying that CA does get blown up as magnification goes up and this invalidates all arguments based on that.

[ChrisR]

This has fallen down to "if"s "but"s and "maybe"s - depending on how much of one type of aberration or another there is, and which one(s) matter to a given user in a particular application.

A really good 20x would beat an average 40x.

Any NA 0.75 objective "should" resolve the same detail on subject. Beyond that, "It Depends".

[mjkzz]

hahaha . . .

I think it is becoming pointless at this point, it all comes down to every single instance of setup . . . learned a lot from this thread. thanks all.