Zhongyi (Mitakon) Super Macro Lens (1 - 5x)

[mjkzz]

So, yes, telecentricity is very important, particularly when stitching stacked images.

I used f/2 mark which seems to be the sweet spot of this lens.

What % of folks buying this lens are going to do Stack and Stitch? Probably < 0.1%. So indeed telecentricity may be important to you, but is it to the manufacturer, especially if it causes a reduction in performance for other purposes?

You mention f/2 but I think this is an f2.8 lens? Do you mean stopped-down 2 stops?

But I think this lens is so close, moving the aperture MIGHT work, so why not?

I set the aperture at f/2 mark on the lens.

So this is actually an f/2 lens, not f/2.8?

OK, when set at 5x, the focal length is actually 25mm, with 33mm front element, I think the f/0.8 can be true! At 1x, focal length is 85mm, so f/2.8 is also possible.

I set magnification at 3x, I do not know focal length, so all I can say is I set it at f/2 mark.

[Lou Jost]

I'm confused. In your tests on page 1, you started at f/2.8, then f/4, etc. You did not include f/2. Why didn't you start at f/2 on that sweep? Also, your tests showed improvement on stopping down. But now you say f/2 is the sweet spot for this lens. I don't understand.

[mjkzz]

I'm confused. In your tests on page 1, you started at f/2.8, then f/4, etc. You did not include f/2. Why didn't you start at f/2 on that sweep? Also, your tests showed improvement on stopping down. But now you say f/2 is the sweet spot for this lens. I don't understand.

f/2 mark on the aperture ring, it does not mean it is f/2. I think from now on, it is better to use the term f/2 mark instead of f/2.

[rjlittlefield]

mjkzz writes that the working distance is 100mm at 5x. Taking that as a measurement, then best case, if the entire front element is used, the effective f-number on the subject side would be 100/33 = f/3 = NA 0.167 . At the same time, the effective f-number back at the sensor would be 5X higher, so f/15. Those apertures are very similar to the MP-E 65.

At this point, I'm thinking it would be wise to pay attention to how the lens behaves and pretty much ignore the numbers.

--Rik

[mjkzz]

Sorry, this might sound nitpicking, but I would characterize telecentricity as a property of an optical system. A feature sounds like something you can add instead of "being"?

You're right, that does sound like nitpicking to me. Anyway I think if one is assembling an optical system, telecentricity is indeed a feature that can be designed into the system if one wants. If we had two competing optical systems, one telecentric and the other not, I think it is quite natural to say that the telecentric system has a feature that the other does not have.

But it also has limitations on size of objects being imaged. I'm not sure if the front lens is large enough for telecentricity across the full mag range, but if not this might be a problem. Plus won't the lens need to be stopped-down a bit to be telecentric?

Sure, but I do not see that as limitation -- the front element has 33mm diameter, so maybe at 1x, it will not cover a full frame sensor, but at 1.5x, it should be close. At 2x or more, I do not think you are limited at all. To image larger objects, stitching is one solution

[mjkzz]

mjkzz writes that the working distance is 100mm at 5x. Taking that as a measurement, then best case, if the entire front element is used, the effective f-number on the subject side would be 100/33 = f/3 = NA 0.167 . At the same time, the effective f-number back at the sensor would be 5X higher, so f/15. Those apertures are very similar to the MP-E 65.

At this point, I'm thinking it would be wise to pay attention to how the lens behaves and pretty much ignore the numbers.

--Rik

Actually, at 5x, focal length is 25mm :-) The focal length changes for different magnification.

So, I agree, it is better to ignore actual aperture number, but use the mark on the aperture ring instead.

[rjlittlefield]

mjkzz writes that the working distance is 100mm at 5x. Taking that as a measurement, then best case, if the entire front element is used, the effective f-number on the subject side would be 100/33 = f/3 = NA 0.167 . At the same time, the effective f-number back at the sensor would be 5X higher, so f/15. Those apertures are very similar to the MP-E 65.

At this point, I'm thinking it would be wise to pay attention to how the lens behaves and pretty much ignore the numbers.

--Rik

Actually, at 5x, focal length is 25mm :-) The focal length changes for different magnification.

So, I agree, it is better to ignore actual aperture number, but use the mark on the aperture ring instead.

It doesn't matter what the focal length is. If the lens is 33 mm wide, and it's located 100 mm away from the subject, then the effective aperture (working aperture) cannot be any wider than f/3 on the subject side.

--Rik

[mjkzz]

mjkzz writes that the working distance is 100mm at 5x. Taking that as a measurement, then best case, if the entire front element is used, the effective f-number on the subject side would be 100/33 = f/3 = NA 0.167 . At the same time, the effective f-number back at the sensor would be 5X higher, so f/15. Those apertures are very similar to the MP-E 65.

At this point, I'm thinking it would be wise to pay attention to how the lens behaves and pretty much ignore the numbers.

--Rik

Actually, at 5x, focal length is 25mm :-) The focal length changes for different magnification.

So, I agree, it is better to ignore actual aperture number, but use the mark on the aperture ring instead.

It doesn't matter what the focal length is. If the lens is 33 mm wide, and it's located 100 mm away from the subject, then the effective aperture (working aperture) cannot be any wider than f/3 on the subject side.

--Rik

True!

[mjkzz]

OK, insomnia sometimes can be productive

Now, set Sony A7 III to crop mode, ie, APSC mode, to take advantage of much less CA in the center for the lens. The result shows almost CA free stitched image. Now, it is 7x5 tiles or total 35 tiles. But end result is about the same MP count.

Now, Microsoft ICE can stitch it up easily. For last one, it MIGHT be the patterned, out of focus, black t-slot background was tripping ICE, not sure. So I suspended the coin on a tempered glass screen protector for cell phones and it seems to work and less PS work for me to remove the background.

I think I have something reflecting some light on the coin this time.

Intentionally left the XY table NOT well aligned, just to see what kind of issue it can cause, well, not much except jagged edges in the final stitched image shown in ICE image (red arrows)

You can download the final image in TIFF here, to check CA

https://www.dropbox.com/s/aiq7kyfz8iq6b ... .tiff?dl=0


Coin setup



Microsoft ICE result

[mjkzz]

I think this lens is really something, now, I can stitch it with Microsoft ICE without any problem, I do not think it is necessary to make it even more telecentric, so maybe I will withdraw that suggestion to them.

[ray_parkhurst]

What version of ICE is that? Mine looks completely different.

[mjkzz]

[ray_parkhurst]

Different version

[mjkzz]

thanks, just downloaded the newer version.

SnS'ed US Penny, 6x4 = 24 tiles, at 3x and Sony A7 III crop mode, total MP is close to 100MP

Download TIFF file here: https://www.dropbox.com/s/6hsij68i2rs3d ... .tiff?dl=0



Smaller image at 80% quality

[ray_parkhurst]

Looks pretty good!

I uploaded the image to my EasyZoom page. Makes it easier to zoom-in...can I have your permission to make the image public and publish a link?

What aperture did you use for this SnS?

Interestingly, the overall image size using the A7iii in crop mode at 3x mag turns out to be about the same as for my T2i at 2x mag. Makes sense since the A7iii has 5.9um pixels, while T2i has 4.3um.