Pushing the 7.5x Mitu down to 1.8x on MFT

[Lou Jost]

Beatsy recently posted about finding a 105mm tube lens that worked on his Mitutoyo objectives without vignetting on a full frame sensor. That was interesting especially because many of us had noticed corner degradation already with longer focal length lenses; we all generally concluded that this was because the image circle of the Mitutoyo lenses wasn't big enough. But Beatsy's result shows that the image circles may very well be big enough, and it was in fact something about the design of the majority of tube lenses that caused the vignetting.

This raises the question- how far down can we go? And what lenses to use? The answer depends partly on what sensor size we are using, but we can take that out of the equation by asking what's the largest field of view (FOV) we can image without vignetting. This number (measured on the subject side, not the image side) doesn't depend on the sensor size, and it determines everything else.

We can figure this number out using Beatsy's results. Beatsy found a 105 mm lens that does not vignette his 10x Mitu. That would be m= 10*(105/200)= 5.25. His FF diagonal is about 43.3mm so the FOV of the Mitu (on the subject) is 43.3/5.25= 8.2mm in diameter, much larger than the 4.3mm we see when we use the 10x with its normal 200mm tube lens. Of course the quality must drop off away from the center, but still, that's nice to know. How much of that extra space is useful remains to be seen: it will depend on the subject and our fussiness about quality. But it is worth exploring to see how far down we can go.

I use mostly MFT sensors, which are exactly half as wide as Beatsy's FF sensor. If he could find a 105 mm lens that does not vignette, this implies that there could be a 52.5mm lens that would not vignette on MFT (at that focal length, the 8.2mm FOV would fill the MFT diagonal). The same relationship would be true for his other Mitutoyo objectives; if the 105 worked for him, a 52.5mm lens could work for me. There are no lenses of that focal length, so I tried lenses of 50mm and 55mm.

The first one I tried was the Olympus 50mm f/2 macro lens for Four Thirds. This is a very good lens. I tried it on my 10x and 7.5x Mitutoyos. It showed very slight corner vignetting but good sharpness except right at the corners and extreme edges. Here are the results for the 7.5x, which is being pushed down to m=1.875, FOV = 12mm, first without any adjustments or sharpening (Zerene PMax):



Below is a 100% crop from the center:


Below is a 100% crop form the bottom left corner:


And here is an adjusted version, with vignetting corrected using the Photoshop vignetting correction tool:


I also did a version in high resolution (Olympus pixel-shifting to obtain 50Mp resolution); Here is a crop from that version, with no adjustments or sharpening:


More adventures in pushing down tomorrow....

[ChrisR]

Solid results.

The Nikon 10x NA 0.25 CFI has a large image circle, around 5-6mm, and I'm not sure than one of the other cheapies (BE?) isn't a tad wider. They're way less cost than a Mitty. Good point about the very corners often not being looked at in an image. Casting an eye down the image galleries,
- apart from butterfly wing shots -,
it's not so easy to find an appealing image where high corner resolution is a important.
One can "eyeball" to see the potential image circle. View a phone ruler, white bars ona black display ("Swiss army knife" app) through the objective, just as though using it as a magnifying glass.
Hold the subject and objective still, but move your eye about. If you can see 5mm, that's the maximum you're going to get. Then comes the "but how good"!
I've been playing with a 32x LWD objective. I can see 2mm. That's 64mm on sensor. There's CA but it corrects,...Hmmm.

[mjkzz]

Wow, great results.

You beat me to it, I was about to test "pushing down" thing with my GH5, but both of my Canon 50mm (1.4 and 1.8 ) are not very good for tube lens -- too much space from rim to the front glass, causing objective being mounted too far away.

Even that, I mounted a Olympus MSPlan 20x 0.40NA, it does have vignetting but not like port hole. Also mounted a Mitutoyo QV 2.5x (which requires 100mm tube lens), seems to have little vignetting. I believe most of it are due to being mounted too far away from front glass.



setup: GH5 -> Viltrox EF-MFT flange adapter, empty one, no glasses -> Canon EF-50mm f/1.4 -> Mitutoyo QV 2.5X


this is the olympus 20x pushed down to 5.56X, it is not pretty, but you get the idea


Mitutoyo QV 2.5 0.14NA pushed down to 1.25X

[Lou Jost]

Your 2.5x was one of the victims of my tests yesterday. I had the same problem--- I wanted to use it on my 70mm Sigma macro lens but it has a recessed front element and it was hard to insert the objective close enough to the front element to avoid vignetting. I managed to do it though, by putting the objective on a 42mm mount and then inserting it deep into the lens and then putting a large step-down filter on the lens. You can also sink lenses by stacking step-down filters in an inward-facing cone or alternate inward and outward-facing rings to get exactly the distance you need from the lens front. I have used this technique to plavce waterhouse stops accurately at the telecentric distance for lenses with recessed front elements. See my post:
http://www.photomacrography.net/forum/v ... ure+stepup
But in this case I am using one of the rings as a washer, not screwed into place, and the objective + 42mm ring is also not screwed into anything but held firmly in place by the washer, the lens, and the step-down rings that act as a collar. Here are pictures:




Using your 2.5x on the Sigma 70mm macro, I got beautiful images at m=1.75 with absolutely no CA and only very minor corner light falloff, though not as bitingly sharp as a 10x pushed down. There was some image degradation at the edges and extreme corners but it would not be noticeable in most subjects. I want to re-do that image because of lighting inhomogeneity, but here it is, with only brightness adjusted.

Full image:


Center at 100% below:


Center right edge at 100% below:


Upper right corner 100% below:


A good 80mm lens would give 2x exactly, and would use only the sharp parts of the image circle. Unfortunately I don't have an 80mm lens. I think a 50mm lens would include large blurry sections at the corners with that objective, but again, these might not be noticeable for bugs.

[Lou Jost]

With the 7.5x on the 50mm Oly macro lens at m=1.875, about 70% of the diagonal across the frame is very sharp. So I would say the good field of view (on the subject) of the 7.5x Mitu is at least 8mm in diameter. That number is independent of sensor size.

For a given sensor you can figure out what lens focal length (f) you need by this formula:
8= [diagonal size of sensor]/[f/200 *7.5]
[(f/200) *7.5]*8 = [diagonal size of sensor]
f = [200/(8*7.5)]*[diagonal size of sensor]
f = 3.33*[diagonal size of sensor]

So with that objective on MFT, it should be possible to get a completely sharp image using a lens >-=72mm focal length.This is more or less confirmed by my very good (but not perfect) results with the 70mm Sigma macro lens.

For APS (Nikon) you should use a 94-95mm lens.

For full frame you should use a 143mm lens or longer. Note that Beatsy got good results with a 105mm lens on FF, so my numbers are conservative.

[Lou Jost]

One other thing: the 7.5x + 70mm lens was almost telecentric, with Zerene scaling factors differing between successive frames by only 1 unit in the fourth decimal place. I re-assembled the stack with scaling turned off and there was no loss of detail in the finished stack. That means one could use this for stitching very easily.

[lothman]

shouldn`t a lens from a slide projector be ideal acting as a tube lens. The have no moving parts and can be easily machined?

[Lou Jost]

Most are really cheap, but maybe a very high-end one would work. I don't know. They are not exactly optimized for infinity as an ideal tube lens should be, but they are close..

[ChrisR]

I've tried a couple of projector lenses as tube lenses, including a supposedly very good zoom one.
Terrible CAs, !

[Beatsy]

Interesting tests Lou. My minor discovery was pure luck really - thanks for methodically taking it forward. Those Mitties sure take a pushing! Could be getting into territory where copy variation becomes an issue though...?

[Lou Jost]

I wish we knew more about how the tube lenses play a role in this, either hindering or helping...

[mjkzz]

Thanks Lou.

Found a left over moth wing from maybe 2 years ago, so I tried with that 20x Olympus MSPlan 0.4NA and with that Canon EF 50mm f/1.4 again. I think I am pushing it too far as the corners are really bad. I do like the center. Equipment: GH5, Canon EF-50 f1.4, Olympus MSPlan 20x 0.4NA with f=180mm, so, it is about 20*50 / 180 = 5.56x

Now, maybe I will take Beatsy idea further to get a Pentax 50mm SMC, see, I really like the fact that those Pentax lenses are made of metal, feels solid, on the other hand, the Canon ones (50mm and 100mm) are plastic and wiggly.

cropped out some artifacts due to magnification


center 100% crop


lower middle 100% crop


right middle


upper left corner 100% crop, really bad (or I just did not do a good stack)

[Lou Jost]

It may be that your step size should be bigger, there seem to be fuzzy strips in the stack. Edit--I meant "smaller"!!!

I do have a Pentax 50mm SMC but can't find the step-down ring I need to mount the Mitu on it... I will keep looking....

[mjkzz]

oh, thanks Lou, yeah, the step size was 3.125um, I will try larger step if the wing is still intact :-)

[Lou Jost]

Sorry, I meant to write that the step size should be smaller!!!