Micro flyeye

[patta]

View through an insect eye. The hexagons are ommatidia lenses; the color blobs are the images that each one forms.
.
The bug's point of view: Too late to fly away! He's gonna smash me and put my skeleton in front of his creepy macro rig!
And that's exactly what happened.
.

.
Aperture about f/20, no stacking for now
Micro-fisheye objective, inside a prepared fly eye shell. Field of view about 160 degrees; point of view, about 1,2 mm from focus
The hexagons are about 30 micron diameter. The hand is 1/2 foot long.

Subject: human hand ... ok, eye of a common sirfid, likely Eristalis tenax
I'm slowly writing down the details for the setup, please be patient, the introduction keeps inflating https://patta107285337.wordpress.com/micro-fisheye/
If you know this type of picture being taken before, please tell me, so I'll spare to write down all the details and learn from what's been done already , maybe get sharper results.

[patta]

Ok I've changed the title, so it is clearer what this grainy and aesthetically horrible photo is about.
I wrapped up the project for now (40 days of afterwork), maybe will restart in January/February.

Giorgio

[Scarodactyl]

Bizarre and very cool!

[patta]

Thanks for the look! If you prefer a less cruel flower...
.

[Sym P. le]

Very interesting. I didn't understand the hand image but I can see it now that you've posted the flower.

[rjlittlefield]

This is very interesting work!

One minor correction...

Your description says:

View through an insect eye. The hexagons are ommatidia lenses; the color blobs are the images that each one forms.

The concept of "images that each one forms" seems not quite right.

The image formed by the lens in each ommatidium in a fly's eyes is real and inverted, just as it would be for any short simple lens. If we were seeing those images, then each color blob would show an inverted image of the hand.

See pages 28-29 of "Eyes to See: The Astonishing Variety of Vision in Nature", by Michael F. Land. Here is a small extract from the book:

So, I think that what we're seeing in your image is really a blurred view of the outside world, the hand or the flower, seen through a hexagonal mask formed by the borders of the ommatidia. I expect that your preparation of the fly's eye has destroyed each ommatidium's ability to form an image of its own.

That said, the overall effect seems pretty faithful to what would appear on the fly's neural retina, after the first couple of layers of neurons have un-done the optical inversions and stitched together the information from neighboring ommatidia.

Land's book has a fine account of how all this works. I strongly recommend it.

--Rik

[Sym P. le]

I don't know what's more astonishing, that someone would try to put a flies eye in front of a microscope objective or that someone would know that someone has tried it before!

[patta]

.....The image formed by the lens in each ommatidium in a fly's eyes is real and inverted, just as it would be for any short simple lens. If we were seeing those images, then each color blob would show an inverted image of the hand.

So, I think that what we're seeing in your image is really a blurred view of the outside world, the hand or the flower, seen through a hexagonal mask formed by the borders of the ommatidia. I expect that your preparation of the fly's eye has destroyed each ommatidium's ability to form an image of its own.
...
Land's book has a fine account of how all this works. I strongly recommend it.

--Rik

Thanks again for the comments! I've read one article by Land, but skimped the book... I'll go back to it!

Yes, my preparation of the fly's eye was quite crude and not really delicate... I had to brush away all internal structures, cone and baffles, since they were soft tissues. However the bulk of the refractive power should still be in the hard shell, the front curvature of each ommatidia; under normal microscope, they still produce an image, pretty sharp too; I've measured the focal length at about 120 um; ommatidia diameter is about 30 um. Below I'll brag some excuses on why the single ommatidia images are not very clear in the photos:

Mostly is narrow aperture and poor focus: it was f/20 nominal on the relay, and (estimate) about f/10 object side; not much resolution available for an image inside a 30 um hexagon. At high aperture (low f#) each ommatidia would project an image with higher resolution but much larger than the ommatidia itself thus all small images would overlap, with global result much worse than the images posted. Stopping down reduces this overlap. In the eye of the live bug, each image is contained by dark baffles (that I had to brush away).

The focusing point: in the center I've put it not exactly in the image, but a bit shifted toward the ommatidia lens, so the black contour of the ommatidia become a bit focused too (otherwise it will be more blurred); away from the center, the **** eye is approximately spherical (~3mm diameter; I cut a cap of 2mm) so the focus wanders across the field: I tried out different microscope objectives to find the one with more or less the same field curvature, with the best focus surface a sphere of radius matching the eye, but think I'm still far off. So the only quick way to get all in one shot was to close down the aperture.
I wrote to vendors of the big 4 microscope manufacturers, "what is exactly the field curvature of your objectives? I need it for a project...", you may imagine the answers.

It is possible to focus at infinity (hand, flower) but then the hexagonal mask get totally blurred: even at the closest aperture (f/32 nominal) the effective entrance pupil is about 0.15mm diameter (objective fl 2mm), much larger than the ommatidia.

The other excuse is the relatively narrow FOV of each ommatidia, about 12°; while the flower or the hand are much larger, about 120° (they were right in front!); so each ommatidia pictures just a piece of finger or petal; it is (barely) visible that such pieces (the color blobs) are inverted, on the "wrong" side of the finger or of the petal.
The photo has been taken by a fisheye objective, with the entrance pupil (point of view) in the center of the eye dome; so each ommatidia looks at a different direction.

So, I believe that the images can be improved with better skills, optics and patience. I'll try to post all the setup details and thinkering, but in a while, I've exaggerated the last weeks and now I have to look after the wife and the real jobs .

Giorgio

[patta]

I don't know what's more astonishing, that someone would try to put a flies eye in front of a microscope objective or that someone would know that someone has tried it before!

You wouldn't believe how may people did it before! Looks that scientists have battled for centuries on how the fly's eye should work (and they still don't agree)

[rjlittlefield]

Thanks for the additional info.

I am surprised to hear that the ommatidia retain focusing power.

Is the cornea wet on the inside, or does the cross-section of cornea for each ommatidium vary in thickness so as to form an ordinary lens? I had assumed from cast skins of other groups that the fly cornea would be so thin that it would not have significant refractive power unless filled with fluid.

Mostly is narrow aperture and poor focus: it was f/20 nominal on the relay, and (estimate) about f/10 object side

Image and object side f-numbers should be simply related by magnification, at least at image center and at focus.

So, is your system running at 2X overall in the "plane" of focus?

--Rik

[patta]

Thanks for the additional info.

I am surprised to hear that the ommatidia retain focusing power.

Is the cornea wet on the inside, or does the cross-section of cornea for each ommatidium vary in thickness so as to form an ordinary lens? I had assumed from cast skins of other groups that the fly cornea would be so thin that it would not have significant refractive power unless filled with fluid.

Mostly is narrow aperture and poor focus: it was f/20 nominal on the relay, and (estimate) about f/10 object side

Image and object side f-numbers should be simply related by magnification, at least at image center and at focus.

So, is your system running at 2X overall in the "plane" of focus?

--Rik

Hello again
I think I did the same preparation as Leeuwenhoek and Land: cleaned eye cornea suspended over water. Just that they took a small flattened section, while I kept a large spherical cap, mounted over a droplet of water. So the back of the cornea is filled with water.
As we see from all macro the outside surface of the single ommatidia lens is curved (tried to measure it under microscope with incident light... 50 micron radius? It is strongly curved anyway, and looks precisely spherical);

Luckily for us, the cornea is also the main skeleton of the eye, it is "thick" (40 micron?) and robust, so it keeps its shape; inside of the ommatidia should be a bit curved too, making a double convex lens (not a meniscus!). But since it is suspended on water, I think the power of the rear surface is reduced. Even if it was a meniscus, having it filled with water restore the refractive power. There is another lens on the ommatidia, crystalline/cone, that needs to be removed for practicality; but luckily it has little refractive power; the same as for human eye (about 2/3 of the power comes just from the front surface of the cornea, 1/3 from the crystalline).
I looked at a small piece of cornea under microscope; dry on both sides, it has focusing power and produce images. Immersed on both sides with water, seems no power (or very small).

I found out afterwards that there are many researchers trying to manufacture an artificial insect eye, but they struggle in manufacturing the small ommatidia. The real ommatidia from an insect look so well made (and cheap), that I have dreamed about a business of farming flies to use their eyes as objectives.
https://iopscience.iop.org/article/10.1 ... aaffb5/pdf
See Figure 41 in particular

The imaging system is a relay: a pre-war 100x microscope objective in the front and a 90mm consumer macro lens as relay.
I still didn't figure out how it works exactly (and thank you for publishing ancient posts about pupil ratios...).
The macro was at about 1:1; I got 4x total magnification in the center (by pixel count over the ommatidia size), so the objective should be working at 4x.

The macro lens is on Canon camera, so I get (and posted) the nominal f# at infinity; I assumed that at 1:1 it gets doubled:
Relay Nominal f/20 -> relay effective at 1:1 f/40
Objective magnification 4x -> effective object-side f# = 40/4 =10
That should give 5 micron resolution, more or less matches with the details in the photo (where there is focus).

I still have to sort out properly all those details!

Giorgio

[iconoclastica]

I don't know what's more astonishing, that someone would try to put a flies eye in front of a microscope objective or that someone would know that someone has tried it before!

Haven't we all tried this once (and failed completely)?

[MarkSturtevant]

That is unique. There are various chemical agents to dissolve soft tissue while doing minimal damage to the cuticle. Dilute bleach and Potassium hydroxide being among them.
I haven't checked to be sure, but a cast skin might retain lenses.

[patta]

Worker's day! Luckily this is an hobby for me.

Long break, it was winter and I was short of fresh "optics". Now they are flying around again.

I followed the last suggestion from Mark, cleaned the inside with weak sodium hydroxide, it worked well. Also ammonia worked.
Other big improvement, a stable way of mounting the eye skin, so the focus is better centered and I could use larger aperture; you get sharper flowers. And found a new (actually ancient) apo lens with field curvature right there. Yes a lot of dirt and scratches... and still no focus stacking.
The whole image is like a fisheye, now about 130 degrees wide for the circle. The single ommatidia images are narrower, and point more or less in the right direction.

About the old issue, if the skin itself has refractive power, it is not clear. Is the back of the ommatidia flat, convex or concave? I tried to observe it directly, with scarce success.
For exuviae cast skin, it make sense to be concave, as the new lens grows inside it; so the eye skin of an exhuvia should have small/no focusing effect by itself.
Here a link of an old paper, where they have carefully imaged the bee's eye, also in cross-section. The result is actually unclear. Varela & Wiitanen, The optics of the compound eye https://rupress.org/jgp/article-pdf/55/ ... 67/336.pdf

Edit: a more thorough discussion of different insect eyes - in German and ancient, but still better than any subsequent publication:
Exner "die physiologie der facettirten augen" year 1891
https://archive.org/details/diephysiolo ... 9/mode/2up
Illustrations at the end.
It provides a definitive answer on whether the cornea has refractive power or not!
Answer: for some insects yes, for other not.

[patta]

For the pixel peepers in all of us, here the max that I managed this round with the sirphid eye. Probably more can be squeezed with larger aperture (but shallower focus), more magnification on sensor and better alignment. The ommatidia are about f/2.5 or f/4 and should manage some further.
The scale is the size of ommatidia and intermediate image, about 30 micron; 1 pixel ~0.7 micron.
i'm super happy about those results, now I just need to catch more flies and clean the eyes properly.