Nikon M Plan BD 40x 0.65 Tested on 5DIII

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

Greenfields
Posts: 116
Joined: Thu Nov 25, 2010 10:54 am
Location: Nottinghamshire, England

Nikon M Plan BD 40x 0.65 Tested on 5DIII

Post by Greenfields »

This test is unreasonable: the objective is not designed to cover a full frame sensor (and it doesn't), and it's not really fair to get StackShot to make one micron steps (it almost manages).

The objective has been tested before by Javier Torrent
http://www.photomacrography.net/forum/v ... hp?t=12669
and ..t=12894, 13789, 14442, 15720 and 16672

It is interesting because 1] despite the relatively high N.A. of 0.65 it is BD (Brightfield/Darkfield) design so does not need a cover-slip, 2] It is currently relatively inexpensive [mine cost £107 on eBay this autumn] and 3] despite its tiny working distance [Nikon say 1.5 mm] the conical shape of the inner barrel makes lighting practicable.

The main limitation of this objective is that it needs subjects which are almost flat and there is a high risk of touching the subject and soiling the front element when setting up a shot.

It is designed for a tube length of 210mm so you don't need a "tube lens" but it does not have an RMS thread: you need an adapter with a Nikon BD thread. Although this is not identical to the Mitutoyo thread, see:

http://www.photomacrography.net/forum/v ... hp?t=12404

In practice I found that this lens fitted my Mitutoyo adapter. You need a "washer" [which can be home-made from black card] to cover the annular illumination channel.

Here is the objective:

Image

Image

I don't know what the small holes are for. They are not threaded and simply go straight through the barrel, spaced at 90 degree intervals. Maybe for the pins of an assembly jig or even G-spanners though I suspect that such objectives were never designed to be serviced. I have seen them on other Nikon objectives from the same period.

Image

The outer barrel unscrewed with very little effort. There was no evidence of a thread locking compound

Image

Outer and inner barrel from rear showing ground-glass diffuser in the outer barrel and its reflective inner surface close to the nose.

The following images have been shot with StackShot in high precision mode set to a slice spacing of 1 micron.

Lighting is by a pair of MT-24 EX flash heads set to 1/8 or 1/4 power.

Diffusion is a ping-pong ball diffuser round the inner barrel of the objective. A notch has been cut in the diffuser which can be turned towards one of the flash heads to produce an adjustable degree of directional light.

I expected the centre of the field to be darker because it might be shaded by the objective - but no. If there is any shading it is almost balanced by the geometrical fall-off in the image brightness with field angle so the net effect is that the image is slightly darker at its edges.

The subject is a sprinkling of slime mould spores (physarum album) from the fruitbody shown in thread (21823) on a photographic grey card (real card - not a modern plastic type hence the texture).

Stacking in Zerene Stacker, PMax.

Contrast has been adjusted to optimise the tonal range in Photoshop CS5 and the images sharpened with two passes of the "smart sharpen" filter, 20% each pass, radius = 1.5 pixels [because there is no real detail at pixel level.]

Image

This is the full frame image. [Warts and all: the banding at the bottom edge is due to imperfect alignment of the stacker by me]. It’s perfectly acceptable at this scale. The image visibly degrades in the corners - but by much less that you would expect for an objective which is probably optimised for an image circle of at most 26.5mm [then Nikon's wide field eyepiece] or maybe only 18mm [then Nikon’s standard eyepiece and the limit of the aberration plots in Nikon’s contemporary patents for microscope objectives]

The full frame sensor has a 42mm diagonal.

Image

This is a 50% [i.e. 2 original pixels down to 1 in the image] crop at of the centre of the image - so only covers the highest quality in the centre of the field and you can see that the detail is already breaking down.

Looking at the table at:

http://www.microscopyu.com/articles/for ... ution.html

A 40 x 0.65 achromat should resolve about 0.42 microns which, at 40x magnification is 16.8 microns.

The 5D2 sensor, with a photosite pitch of 6.41 microns should be able to resolve detail of the order of 20 microns [because you need 2 pixels to resolve a line, then a bit more because the Bayer CFA costs about 30% of resolution and the alignment of the detail won't match the photosite grid].

So: very roughly they match. However: that's for the limiting resolution when the highest frequency subject detail is highly modulated [like black and white bars]. The system should, in theory, be able to capture black and white bars as just noticeable detail. If the subject detail has a lower contrast it will only be captured at lower spatial frequencies.

Here are the same spores shot with more aggressive lighting. [the background is also different - it’s a white "grey card"] Getting an accurate white balance was very difficult because the background is translucent enough to take up and diffuse the brown hue through the translucent spores.

Image

Then a crop at 100%

Image

It does appear that detail can be captured at a pixel level in the centre of the frame if the lighting generates sufficient contrast in the subject.

I don't know what the little spheres are. Perhaps calcium oxalate crystals from the membrane enclosing the fruitbody ?


Back to the original test, this time with 100% crops:

Image

Image


The crops show that the image detail starts to deteriorate away from the centre more rapidly than we are used to from lower magnification objectives: for example the 10x 0.25 or 0.28 lenses. I'd suggest that this is because the good 10x objectives mentioned in these pages out-resolve the sensor in the centre of the field, sometimes by a significant margin. Although their performance does fall off with field angle, it does not become clearly visible until a field angle at which the sensor resolution [which is constant across the field] and that of the objective are balanced.

Stackshot and 1 micron steps:

When scrolling through images there is consistent evidence of nutation [nodding] from frame to frame. It's very consistent: one frame slightly to the left and the other back again. So consistent that away from the centre and perpendicular to the direction of nutation the trails from "stuck pixels” in the stack form two parallel tracks like this:


Image

Image is a 150 x 150 pixel crop enlarged to 400% using nearest neighbour setting.

In case you are wondering: this does not arise from two separate photosites: all of the "stuck pixels" in the image produced double trails.

Another observation which may have the same cause is that the position of the focal plane seems to rock back and forth in consecutive images in the same way.

Fortunately, none of this appears to confuse Zerene.

Henry
Feel free to edit my images.

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Henry, this is a very interesting thread.

To cross-check the numbers, I ran them through the MTF diffraction limit formulas copied HERE. Those formulas produce a cutoff wavelength of 16.9 microns per cycle on sensor = 0.42 microns per cycle on subject, assuming lambda = 550nm. This is reassuring to me because the words used at http://www.microscopyu.com/articles/for ... ution.html have never made clear to me what they mean by "smallest resolvable distance between two objects".
When scrolling through images there is consistent evidence of nutation [nodding] from frame to frame. It's very consistent: one frame slightly to the left and the other back again. So consistent that away from the centre and perpendicular to the direction of nutation the trails from "stuck pixels” in the stack form two parallel tracks like this:

Image
This part puzzles me. I can easily understand that the rail could be consistently left/right/left/right, for example due to alternate stick/slip on each of the rails. What I don't understand is why this should result in parallel tracks that are horizontal as shown in the crop. If the frames are alternating left and right, then a single stuck pixel should result in two clusters that are positioned left and right of each other, with a clean gap between them. Combined with any motion at all on the other axis, this would result in two vertical streaks positioned side by side, not two horizontal streaks positioned one above the other. What I see in this crop looks like a consistent up/down nodding (actually more like NNE/SSW), combined with a more or less continuous drift on the horizontal axis.

What have I missed?

--Rik

soldevilla
Posts: 684
Joined: Thu Dec 16, 2010 2:49 pm
Location: Barcelona, more or less

Post by soldevilla »

We can not forget:
we're talking about 0,001 mm displacement. That's about 100 times the thickness of a paper. Get manufacture a mechanism that is repeated systematically below 0.01 is not easy task. Here are small defects accumulate in the parallelism of the guide rods, in the eccentricity between the shaft, which is coupled to the flexible coupling, and the thread. The shift may also be due to a small angle, not a theoretical 0º, in the parallelism between the guide rails and the platform, and even between the optical axis of the camera and tripod engagement platform.

This produces small movements but added in the image on the chip. Y dead pixels or dust spots on the chip is not displaced ...

All this can easily be avoided by applying a flat in the images automatically before Stack :roll: ... :D

Greenfields
Posts: 116
Joined: Thu Nov 25, 2010 10:54 am
Location: Nottinghamshire, England

Post by Greenfields »

Thank you Rik,

I can't explain the observation, but soldevilla makes some good points.

In case it's relevant, I am aware that my Stackshot has a larger than average backlash. [Cognisys, as always very helpful, have confimed that the value they use is an average value].

To work around the backlash when stacking with very small slice spacing, this is my process:

1] Decide what the start points and end points of a sequence look like on the LCD screen.

2] Draw the carriage right back until I am certain that I have travelled further than the backlash.

3] Move forward without allowing any reversing to actually set the start and end points.

4] Draw the carriage well back again then move it forward to approach the start point [but not cross it]

5] Start stacking - but check that the first shot look right. Stackshot usually starts at the right place, but often does not in which case I go through the process again.

One thing I have not tried is adjusting the Stackshot motor's speed. It's currently set at 600. Maybe a lower speed would have an advantage with a small slice spacing.

soldevilla:

I understand what a "flat" means in astrophotography, but how. in practice would you make one for photomacrography ? I presume that you would then subtract it from each image before stacking ? Can Photoshop do this or would you need an astro image editor ?

Henry
Feel free to edit my images.

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

I can't explain the observation,
If you have any reason to investigate more deeply, one source of information is the XOffset and YOffset values computed by Zerene Stacker's alignment process and recorded in its console log and project files (*.zsj). That's the way I got the data on what High Precision really means. See HERE for detailed discussion of what the values mean.
Greenfields wrote:Maybe a lower speed would have an advantage with a small slice spacing.
At nominal 1 micron focus step, the StackShot's motor will only be advancing by 2 microsteps. I'd be pretty surprised if any of the "speed" parameters have any detectable effect in that regime.
To work around the backlash when stacking with very small slice spacing, this is my process:
That's the same process I use. In Zerene Stacker there's support for automating it, via the "Prerun distance" parameter in the StackShot controller configuration panel.
In case it's relevant, I am aware that my Stackshot has a larger than average backlash.
I would not expect that to have any effect on the one-way operation that you've described. As far as I know, the variations in backlash are just a matter of tolerances in the thrust bearings and mating of the lead screw and nut.

There are a couple of procedures for calibrating backlash, described HERE. But even when properly calibrated the correction is not exact because the rail does interesting things like twitch significantly from side to side when the direction is reversed. My usual recommendation is to not worry about calibrating backlash, but just know that it exists and work around it with prerun. I also recommend shooting a few extra frames at front and back, not because the equipment does anything wrong but just because it's hard for me the human to tell exactly where the start/stop points should be.

--Rik

soldevilla
Posts: 684
Joined: Thu Dec 16, 2010 2:49 pm
Location: Barcelona, more or less

Post by soldevilla »

I did some tests strongly reducing the speed of my StackShot and have found improvements when I photograph at x20 (the maximum that I can use, I have no more powerful lenses).

A flat can efficiently remove all the dead pixels and dust spots on the chip, it allow to do disappear those "worms" of points, clearly visible on blur backgrounds.

The operation to apply a flat is the division, and Photoshop does not this way of applying a layer. But Gimp can, I did tests with Gimp to automatically recover a correct color balance (the flat was a Median of the original image). Maybe we can find an automatic process to apply a flat before stacking, I'll try ...

seta666
Posts: 1071
Joined: Fri Mar 19, 2010 8:50 am
Location: Castellon, Spain

Post by seta666 »

Hello,

This is one of my favorite lenses indeed, and while others may think it is difficult to use I always get good results with it; even with not so flat subjects like this bee tongue

40x with 5D mkII

Image

Full size

http://farm8.staticflickr.com/7162/6695 ... 1495_o.jpg

stereo

http://farm8.staticflickr.com/7145/6695 ... f97b_o.jpg

In my opinion is the best high magnification lens you can get at low cost

Regards
Javier

enricosavazzi
Posts: 1474
Joined: Sat Nov 21, 2009 2:41 pm
Location: Västerås, Sweden
Contact:

Post by enricosavazzi »

I suspect - but don't know for sure - that the holes are for a centering tool. The tool has pins that press sideways through the holes against an internal element or group of elements until it is centered according to a collimating instrument. The groups/elements are then blocked in place by tightening a ring that presses on the element mounts in the axial direction.
--ES

Greenfields
Posts: 116
Joined: Thu Nov 25, 2010 10:54 am
Location: Nottinghamshire, England

Post by Greenfields »

That fits.

I received different Nikon objective with similar holes. It was not the item ilustrated and had a broken front element but the seller refunded the payment at once. He did not want it back.

I tested it in case it was an example of how well a damaged lens could perform but the image was not usable so I disassembled the objective out of interest and even managed to match is with a Nikon patent. Build quality was impressive. Shame about the damage.

The elements were stacked together in cells and the whole column retained by a screw-on cap, all inside the outer barrel just as Enrico suggested.

Henry
Feel free to edit my images.

Post Reply Previous topicNext topic