Pixels for use at 4-5X on an APS-sized sensor

[rjlittlefield]

I am restarting my microscope project this year and starting to gather lenses. It is a guess I am struggling with.

OK, so now I think you're asking a question like "How far, and by what means, can I drag an objective away from its design specs and still get a good image?"

If that's the case, then I have another study currently in prep that may provide some useful information. One aspect is to explore using a Mitutoyo 5X NA 0.14 M Plan Apo to get 3X, by various means.

Is that more along the lines of your interests?

--Rik

[Blame]

Well yes, in essence.

We (that is my daughter with me sitting back and watching) did a fair amount of experiments with the QV 2.5x equivalent. I know it is not exactly the same but our results did seem to echo the Mitutoyo 5x. This was with the full frame Sony a900 which I think is more difficult to satisfy with ad-hoc tube lenses than APS-c.

135mm vignettes at least a little with any tube lens tried and I recon it is about low as one can go. We tried about 6 135mm primes with the best being the CZJ 135/3.5. It vignetted but I would consider it a good choice for APS-c.

Our best result was a Componon-s 135/5.6 enlarger lens with only the slightest darkening but it could do nothing about the generally mushy corners.

I think the quality area was a circle of about 22mm so for APS-c it would be about right. I doubt you will do better than enlarger lenses at 135mm but I have a sneaking suspicion that a cheap 4 element enlarger lens might be slightly sharper at f/12 with only the center of its designed field of view in use. They have essentially the same design as the CZJ.

For FF I plan to go up to about f/16 with a 180mm lens. My observations so far is that short lenses do better than long ones so I have ordered this one http://www.surplusshed.com/pages/item/l14015.html . It is so small it should mount inside the mitutoyo!

Anyway we have more or less abandoned the mitutoyo for now. The Schnieder D-Componon 40/3.5 offers a lot more promise. The corners hold up a lot better.

Not that I am all that surprised. I have been checking results at http://coinimaging.com and discovered an interesting thing. A lot of non-microscope lenses when mounted reversed on their own were found wanting because they offered poor sharpness at the magnifications that their apertures justified. However they all seem to offer close to diffraction limited resolution when stopped down to sharpest (often f/4.7) and the magnification was sufficiently raised. If a tube lens could be found that maintained that resolution then they would all be winners.

My guess is that, with a tube lens, for 0.1 NA and under there are a lot of 50mm lenses that will excel while there is a fair supply of 0.14 NA f/3.5 microfilm lenses that will challenge. I doubt that the 10x 0.28 NA mitutoyo has a lot to worry about though.

Anyway, with the D-Claron having a quality area that almost covers at 2.25x and f/8 I recon I have a real problem choosing the right tube lens. I am stocking up with enlarger lenses between 50-135mm and 50mm camera lenses. When my daughter comes back from university in the summer we are going to play.

[rjlittlefield]

Thanks for the additional info.

What you're saying makes sense to me, and your results agree with my expectations.

In most cases, microscope objectives do not work well when dragged outside their design spaces. Even under best conditions, their circles of high quality image do not extend far beyond the "field number" for which they were designed, and usually that is APS-C or even smaller. If you alter the focus arrangement by changing the objective-to-subject distance, then the corners degrade. If you alter it enough, say by using a 5X NA 0.14 infinite objective as finite at 3X, then even the center will degrade due to spherical aberration. Likewise for the equivalent 40 mm f/3.5 landscape lens, reversed. Stopping down is a powerful cure for spherical aberration because SA goes as NA^4, but of course the downside there is that stopping down loses resolution because of diffraction.

If you want to work at low magnification, then your best bet is to use a lens that's designed to do that. Reversed enlarging lenses used by themselves fit that description.

Second best is to use a lens combo in which each component is working near its design point. In the case of an infinity corrected microscope objective or a reversed landscape lens or a microfiche projection lens, that means using it in conjunction with an appropriate converging lens ("tube lens", "rear lens") so as to maintain the as-designed focus arrangement between subject and lens. Unfortunately then the edge/corner quality is controlled by the off-axis effects discussed in FAQ: Stopping down a lens combo. I have no idea how to compute those. Using a physically short converging lens makes good sense because then the pupil locations will match better, but still I think it's a matter of "try it and see".

--Rik

[Blame]

Well yes, eventually one has to get ones hands dirty. Still, there is a planning stage. Between the 1x of a decent macro and the - well call it 3.5x for the 5x times mitutoyo... err anyway between those magnifications there doesn't seem to be a clear solution bar the Canon MP-65 and it has its own limitations. Not least of which is cost and owning a canon camera.

Buying a mitutoyo 2x seems excessive and while reversing an enlarger lens should give good results the truth is it doesn't deliver. Here is the MFT curves for the well regarded Componon-s 50/2.8 https://www.schneideroptics.com/pdfs/ph ... 8_50_2.pdf

As you can see, while it is advertised as handling an enlargement of 3:1 it doesn't do it very well. And why should it when a 144x96mm photograph is too small to show the blur? It is in truth optimized for 12:1 or more because that is where any blur will show. Near as makes no difference, optimized for infinity. Of course the longer focal lengths really are optimized for 6:1 or so but with so much unneeded coverage they ain't that sharp.

So I believe that there is a real case for stacking lenses for low magnifications. If the lenses are physically small enough, stacked tight to each other and a black paper stop mounted between them then the optical path should stay reasonable.

Our plan is to aim for 2.5x to 3x with the 40/3.5 microfilm using its internal stop, 1.5-2x with a 50mm and maybe a paper stop and 1x using a matched pair of 4 element enlarger lenses of round about 75mm with definitely a paper stop.

Given that enlarger lenses often work best at a little under infinity we may well bend things a bit by trying the combo at about 20% above the magnification calculated by ratios. After all you can generally get away with increasing magnification with a finite lens and once combined a pair of lenses becomes a finite lens.

Like all plans it won't survive contact with the enemy but its a starting point.

[TheLostVertex]

What I've done here is to set up an f/11 lens and capture its image in two different modes: 1) direct projection onto a DSLR sensor, and 2) magnified projection, roughly 10X onto the same DSLR sensor.

Not being very(at all) familiar with the USAF test chart, I have been reading trying to figure out your methodology for your calculations. So please correct me if my understanding is wrong.

For mode two you had Group 1 Element 3 resolved, giving you 2^1+(3-1)/6 = 2^8/6

For mode one Group 0 Element 4 resolved, for 2^0+(4-1)/6 = 2^3/6

You then wanted to know the difference in resolution for the two images, resulting in your value of 2^5/6 or 1.78.

I have not looked closely at using the USAF test charts before reading this thread. I am still fuzzy on using it to calculate lp/mm, such as in the link you posted to your film vs dslr test.

[rjlittlefield]

For mode two you had Group 1 Element 3 resolved, giving you 2^1+(3-1)/6 = 2^8/6

For mode one Group 0 Element 4 resolved, for 2^0+(4-1)/6 = 2^3/6

You then wanted to know the difference in resolution for the two images, resulting in your value of 2^5/6 or 1.78.

I think you have the right idea. If you plug exactly what you've written into a calculator or spreadsheet program, you'll get wildly wrong answers because the order of operations will not be what you intended. More parentheses are required, as in:
2^(1+(3-1)/6) = 2^(8/6)
2^(0+(4-1)/6) = 2^(3/6)
2^(5/6) = 1.78

I am still fuzzy on using it to calculate lp/mm, such as in the link you posted to your film vs dslr test.

Well, someplace you need an absolute reference.

If you're buying a small target, such as HERE, then the manufacturer will tell you the numbers for each element. If you want to check, it's pretty simple to match those up against a micrometer slide.

With any sort of projection system, you have two options: you can measure the target and multiply by the magnification of the optics if you know that, or you can measure the image itself.

In this particular thread I've chosen to measure the image directly by optically superimposing it on a micrometer slide (aerial projection). In my Film vs DSLR page I measured the image indirectly by using an overlay in Photoshop whose scale factor was determined by computing from the pixel count and the camera manufacturer's specification of sensor size.

--Rik

[TheLostVertex]

More parentheses are required, as in:
2^(1+(3-1)/6) = 2^(8/6)
2^(0+(4-1)/6) = 2^(3/6)
2^(5/6) = 1.78

Yes, this is indeed what I meant. What I imagine the equation to look like, and what I typed are two different things though

With any sort of projection system, you have two options: you can measure the target and multiply by the magnification of the optics if you know that, or you can measure the image itself.

In this particular thread I've chosen to measure the image directly by optically superimposing it on a micrometer slide (aerial projection). In my Film vs DSLR page I measured the image indirectly by using an overlay in Photoshop whose scale factor was determined by computing from the pixel count and the camera manufacturer's specification of sensor size.

--Rik

I was under the impression that I would need to know the chart size or magnification, but was unsure of where to go from there. So after reading HERE I get:

LineFrequency/Magnification

Looking back at your Film vs DSLR page specifically, this image I tried to cross check your original finding of 125lp/mm for the olympus lens to make sure I understand completely:

Bar size in group -2 element 2 is ~0.04mm projected
Bar size of group -2 element 2 for the chart specs is ~1.785mm
Magnification=0.04/1.785=0.0224
The last bars resolved are group 1 element 4, giving a line frequency of 2.83
2.83/0.0224=~126lp/mm

Since I am so close to your value I assume I have the right idea now, barring any glaring misunderstandings.

[rjlittlefield]

I tried to cross check your original finding of 125lp/mm for the olympus lens to make sure I understand completely

The crosscheck for lp/mm looks OK.

The calculated magnification is wrong because the actual target was a custom-printed one about 4X larger than the number that you used. This doesn't affect the calculation of lp/mm in the image because the reference was in the image and the physical size of the target cancels out when you calculate the size ratio between elements.

--Rik