Coaxial illumination on the cheap...

[BugEZ]

I wanted to try coaxial illumination of a specimen for some fly eye photos, but did not want to spend big $$ for equipment...

I purchased some inexpensive cover slips and used them as the 2-way mirror.

I used a single white 5mm diameter LED for illumination. I shaded the LED with an aluminum foil tube to prevent direct path "leakage" of light onto the subject

To hold the coverslip in position, I made a construction paper cover with a conical tip that supported the coverslip at 45 degrees to the optical axis.

Here is a photo of the rig. Normally, light is injected in the middle of the optical path for co-axial lighting. For this cheapo rig light is injected very close to the subject. Hence the need for the shade on the bulb.

The cover slip and the construction paper cover that slips over the objective.


The reflector, and LED light. Only the LED with the foil shroud (on the right) is turned on.


A view from the opposite side...

(Edited as I hit Submit\t too soon...)

first photo with a proper subject...


Note that there are no "black holes" in the eyes! I now have the "dreaded X's" Not sure how much improvement that is...

Same fly with my usual treatment...



Cool beans!

Keith

[BugEZ]

link to youtube video showing stacks.

http://youtu.be/PhwKxUUcvlc

[rjlittlefield]

Nice demos!

The X-shaped reflections are due to the combination of a small light source, astigmatic reflections, and PMax accumulation. What happens is that the small light source is distorted into a /-angled bar on one side of focus and a \-angled bar on the other side of focus, then PMax assembles the two bars to produce the full X.

I assume you'd like to see what I'm talking about. Here's a sample, from 0:26 in the video at 200% from a full-screen view.



I expect the small light source is also what's causing loss of detail in that hairy forehead of the fly.

In principle, you can fuzz out the X's and recover a lot of the detail by using the same "half-silvered mirror" trick to reflect a much more diffuse light onto the subject. The goal is to get the light bouncing onto subject from such a large source that from the subject's standpoint, every point on the mirror reflects some bit of the light source.

--Rik

[BugEZ]

Thanks Rik for the description of what is going on relative to the ~ vertical/horizontal bars. Here is a crop from one of the source images, though I think your crop from the movie shows it equally well.



And a crop from the Pmax and Dmap outputs from Zerene...





Rik wrote:

In principle, you can fuzz out the X's and recover a lot of the detail by using the same "half-silvered mirror" trick to reflect a much more diffuse light onto the subject. The goal is to get the light bouncing onto subject from such a large source that from the subject's standpoint, every point on the mirror reflects some bit of the light source.

I think I understand this. Rather than have the LED spotlight shine on the coverslip, make a more diffuse light source that reflects off of the coverslip... That will be worth a try. With the working distance between my 10X lens and the subject, I'll have to get rather clever...

In the meantime, I made a final image of the fly with my diffuse "bucket light" and cover slip/spotlight illumination. The black holes are gone as are the crisp details that show when the coverslip is not inserted in the optical path. This may be about as good as I can make it with my modest budget... The image was processed with Zerene. The eye areas are using DMAP but the whiskers use PMAX.



(edit to fix spelling)
Keith

[fergus]

As a complete beginner, I find this thread both bewildering and fascinating.

But I am very appreciative

Edit for clarity: my comments and appreciation are in relation to the difference that lighting makes to the image. I realise more experienced types might already know this, but for us newbies it quite a revelation,,.

[BugEZ]

fergus wrote:

I find this thread both bewildering and fascinating.

My apologies for the lack of clarity... I was quite excited by the results of the test and did not take the time to explain things all that well...

So better late than never...

Here are a few figures to help explain things a bit better.




Most of us have struggled with the "black hole" in the eyes of our insect subjects. Really, it is just a reflection of the camera lens.

I tried to explain the experiment to my wife this AM and how I had managed to fill the black holes. She is a bit more clever than I and suggested that I may wish to tackle Dark Energy next...

Keith

[rjlittlefield]

With the working distance between my 10X lens and the subject
...
The black holes are gone as are the crisp details that show when the coverslip is not inserted in the optical path.

Right, with a 10X objective, shooting through that slanted glass can cause a lot of aberration. I'm not sure whether the observed aberrations are due to the surface structure of the ommatidia, or whether they're introduced by the slanted cover slip. You might investigate that by using ordinary diffuse illumination but shooting through the cover slip.

If I recall correctly, you're using an infinity objective. In that case, a better place to put the beamsplitter is between the objective and the tube lens. In that area each point on the subject is represented by a bundle of parallel light rays that will go through even quite thick slabs of glass without being aberrated. That's actually why microscope manufacturers switched to infinity objectives in the first place -- because it simplified the design of accessories.

--Rik

[BugEZ]

Rik wrote:

I'm not sure whether the observed aberrations are due to the surface structure of the ommatidia, or whether they're introduced by the slanted cover slip. You might investigate that by using ordinary diffuse illumination but shooting through the cover slip.

That is an experiment I tried last night but did not bother to share. The shots through the slip with my normal diffuse lighting only produced rather blurry/foggy images. I did not bother to combine them into a stack. The fog comes from reflections on both sides of the coverslip. It extends from the nose of the objective and some light from my bucket light/diffuser hits the sensor side of the slip and reflects directly into the optical path. Thus the loss of contrast. Even neglecting the loss of contrast, which can be countered to some degree in post-processing, the image is no longer crisp. Here are some crops from zones that transition through the focus zones... Taken with and without the coverslip in place with diffuse light only...
Diffuse light only coverslip in place... Some level setting in Photoshop to diminish the fog...


Diffuse light, Similar settings, similar pose, no coverslip, no Photoshop for fog...



Rik also suggested:

If I recall correctly, you're using an infinity objective. In that case, a better place to put the beamsplitter is between the objective and the tube lens. In that area each point on the subject is represented by a bundle of parallel light rays that will go through even quite thick slabs of glass without being aberrated.

An excellent suggestion. That would also clean up the space between the objective and the subject which is cluttered in my current experiment. Doing what you suggest would require extending the objective through a structurally rigid tube with several holes to allow the light or wires to enter. I don't have an obvious means of doing this on the cheap. The construction paper cylinder/cone was inexpensive. I am not sure how I would match this economy when cantilevering an objective. I'll puzzle over this and see if I can think of an easy way. I have made another sketch that illustrates your suggestion.



Another approach that occurred to me last night is to place the LED directly in the optical path. The LED would obscure the optical path somewhat, a bit like the secondary mirror in a reflecting telescope. You loose some aperture and pick up some artifacts caused by the diffusion of light passing by the obstruction, but you would not have the astigmatism caused by the angled coverslip in an area with an angled ray path. That might work well with some of the tiny LED's available now.




I'll take a few measurements and see if I have enough working distance to test this method. I can easily support a 3mm package size LED. I may have to do some surgery to the plastic wrappings...

Thanks again for the comments!
Keith

[rjlittlefield]

Thanks for the additional images. These make clear that most of the loss is due to imaging through the slanted coverslip, not due to the illumination.

Regarding how to construct a beamsplitter to go behind the lens, I've often toyed with the idea of reworking an ordinary M42 extension tube to make a mirror housing. But that idea has never gotten near the top of the priority list to actually get implemented, and I don't think it will anytime soon.

--Rik

[fergus]

Thanks Keith and Rik - very helpful indeed!

Cheers...
Fergus

[Cyclops]

Pretty impressive!

[BugEZ]

For the last several weeks I have worked in cooperation with forum member elf to test the "method 2" (see schematics above) means of providing coaxial illumination. This method locates the mirror between the tube lens and the objective. It places the mirror where the ray path of the light from the subject is much more parallel to the optical axis. This reduces the astigmatism (see Rik's post above) effect that plagued my "method 1" attempt with the mirror between the objective and the subject.

Elf has manufactured a custom housing that securely holds the objective, mirror and LED assembly, properly positioning them and attaching them to the tube lens. It is in two parts, the Coax housing and the mirror housing. His design is ingenious and the lovely parts he manufactured are well beyond my rather meager construction skills. I am most grateful.

Below is a schematic of the arrangement.



The coax housing has 42mm thread on each end (similar to many extension tubes) to make it easy to adapt to a variety of hardware. Thus, I was able to use adapters from my normal setup.

The coax housing and mirror housing are pictured below. Elf made the mirror housing from black plastic to help minimize internal reflections. The outside is from aluminum for strength. The mirror housing fits nicely into the smooth bore of the coax housing. (Another pat on the back to elf!) The second photo shows how the LED was a perfect fit!






The opening in the mirror housing was sized for a 30mm cover-slip. I chose to purchase quality German cover slips and not the inexpensive ones I used in my testing of the first method. Though they cost 10 times as much, they are of excellent quality. Money well spent. The photos below show the cover-slip being glued to the mirror housing. I chose to use simple Elmer's glue as it would be easy to remove from the plastic housing if any design iterations were required.





After the glue dried I added flocking to block/absorb internal reflections inside the shiny aluminum housing. I flocked the obvious spots as shown in the schematic above. I used Walmart black felt. It is the best flocking material of all the flocking materials I have tested. I then installed the assembly on the lens.

Below is the arrangement I have used for the last several years with my Oly 10X infinity objective mounted to my 200 mm tube lens. There is an iris in the middle that I have not adjusted for several years, but inertia has kept it there, even after I decided that "wide open" was my best practice. The iris has 42mm thread so all I had to do was replace it with the coax housing.



And below the lens with the coax adapter replacing the iris...



I was quite anxious to test the arrangement and made the test image (single frame) shown below using the side lighting only and not the internal LED.....



This showed great promise as the image was excellent, much better than through the nose mounted inexpensive coverslip.


I then wired up an LED that interfaces with my existing LED light arrangement. It allows me to use the existing "side light" from the bucket light or the coax LED, or both...



As you can see in this photo (below), ample light is cast upon the subject.


I then turned off the side lights and used the internal LED only...



This was terrible! There were way too many internal reflections producing a fogged image. I shared these results with elf. He provided wise counsel and suggested some improvements to the internal flocking arrangement. His suggested changes are below...



The photos below show the second try at internal flocking...

First the objective adapter before the new flocking scheme...
Then after, making sure the flocking extends to the OD of the objective's optics.


Flocking was added (not shown) to prevent reflections from the black plastic surfaces on the objective side of the mirror housing (shown below).


On the tube lens side the initial flocking was only the crescent shaped chunk that forms to the mirror housing ID. A facing piece was placed on the shiny surfaces of the mirror housing as shown...


An annular ring of flocking was added to the tube lens side as elf showed in his schematic....


With the improved flocking package installed I gave the coax adapter a second try...

First a reference photo using side light... As you can see some time has passed and this doli has lost much of its color. Still a fine subject for a lens test...



The side light image (above) remained acceptable.

Then a test with the LED only...


The image was quite good. Much of the fogging in the previous coax attempt was gone. I call this a success.

Unfortunately, my stacking computer is in the shop so I won't be able to make any stacks for a week or so. I will post some proper stacks when I am able to process photos again.

Once again, I wish to thank elf for his assistance. He is an excellent collaborator!

Keith

[Chris S.]

Keith, thanks for bringing us along for the ride on this. Experiments shared, we all get to learn from. And a tip of the hat to Ed ("Elf") for his handiwork.

If you don't mind, I'll share a few thoughts, as making a through-the-lens illuminator has been on my to-do list for some time, so I've noodled on it a bit. Also, I have several commercial illuminators of this type that I've been able to dissect and reverse engineer.

My sense is that you might want to make the LED light source quite a bit larger, to avoid what appears to be a nearly point-size light source--albeit a point source delivered through the lens. I suspect you could do this with a diffuser, a light-spreading lens, or both. This would probably require moving the LED farther away from the mirror housing, which would require that a bigger hole be bored for it. The professionally-made units I’ve looked at do this with lenses. Perhaps this also collimates the light, which may provide some benefit. They also seem to place an iris (aka, in this case, "field diaphragm" between the light spreader and the mirror, which, I believe, permits the user to adjust the diameter of the light source.

The approach that you and Ed have taken seems roughly similar to a few off-the-shelf products, such as Anchor Optics mirror/beamsplitter 45 degree adapter for kinematic mounts.



This is a $50 USD item, and appears to work with a 50/50 beamsplitter (50 percent of the light is transmitted through, 50 percent of the light is bounced) available for $38 (stock # AX43736). I suspect—but don’t know—that such a beamsplitter might work better than a cover slip. Edmund Optics, Thorlabs, and other companies also offer 50/50 beamsplitting plates and cubes, sometimes for quite reasonable prices. Of course, what I’m terming a reasonable price may or may not coincide with the “on the cheap” focus of this thread. Still, some folks might well like to know of components they can purchase, rather than fabricate. A difficulty I see with the Anchor Optics items is that they are meant to mechanically interface with Anchor Optics kinematic hardware, which is more suitable to an experimental optical bench than a working photomacrography setup.

Another—potentially more practical--hardware approach might be found in cubes for cage systems, such as Thorlabs C4W and C6W (about $60).



These cubes can easily mount a beamsplitter at 45 degrees, and are threaded to connect with SM1 components—a wide variety of which exist, including adapters for most microscope objectives, and adapters to many converging lens setups. Edmund Optics also has similar products, though at somewhat higher prices.

A thing I like in some commercial axial illuminators I've looked at is that the 50/50 mirror (beamsplitter) can be flipped out of the way when not needed.

On a last note, I’ll venture further afield: When I build my axial illuminator, I’ll want to be able to drop in and rotate a polarizing filter between the light source and mirror/beamsplitter. This will permit cross polarization, so long as another polarizing filter (aka “analyzer”) is placed between the beamsplitter and converging lens. X-pol is important to me, but this will also permit reflected DIC imaging, so long as I stick a single Nomarksi or Wollaston prism in the right spot. Reflected DIC is conceptually much simpler than transmitted DIC, and requires just one prism per image, whereas transmitted DIC requires two. I've purchased several Nomarki prisms for this purpose, and—sooner or later—will give it a try. Of course, reflected DIC, unlike transmitted DIC, offers little benefit to photographers who work with most living things, such as insects or protists; but it can be a real boon when photographing subjects for physics research, material science, and the like.

Cheers! And again, way to go, Keith; you do, you share--excellent!

--Chris

[Pau]

Keith, this is a nice and interesting project.

In the second (more logical) approach I see an important fault: light collimation. To get a good illumination the light bundle that enter the objective rear focal plane must be parallel or almost parallel and uniformely filling that plane, while your LED emits fairly divergent light.
What you're trying to implement is episcopic illumination (o epi-illumination in short) and is very usual in microscopes for industry and material science. You will benefit to see diagrams of this kind of systems at microscopy sites like: http://micro.magnet.fsu.edu/primer/java ... index.html (but it isn't necesarily so complex)

To deal with reflections, both from the optical components and specular reflections from the specimen, crossed polarization as Chris S. suggests will be very important in most cases.

[BugEZ]

Chris S and Pau,

Thanks for the comments and feedback. The review of commercially available beam-splitters is especially helpful. And the notion of a splitter/mirror that can be flopped out of the way is intriguing. I see great advantage to getting the mirror out of the optical path without having to alter the setup so much... And the point source comments are also helpful. I contemplate withdrawing the LED a few mm and diffusing the beam a bit to broaden the source of light it currently provides. Perhaps shining it upon a 10mm diffuse target inside the housing. I look forward to performing and sharing the results of that testing, but first I want to characterize the "as is" configuration. I don't anticipate any polarization experiments, though I can see how polarization might help block the internal reflections.

I am not certain what fraction of light the cover-slip/mirror in my test article throws toward the subject. I am guessing 20% (??). There is probably an equation to predict this and I'll have to do some on-line searches... I do know that I need to dim my solo LED to match the exposure times (~1 second of "light on" during an exposure) I use with my normal side illumination scheme. There is a trade between the reflected and transmitted light. A 50-50 mirror/splitter would attenuate much of the light returning to the tube lens from the subject. In a dual mode, (side and coax lighting) I would need to crank up the side lighting to compensate for the reduced transmission through a 50-50 mirror.

Anyway, this is interesting stuff to think about and great experiments to be able to perform. Again, thanks to elf (Ed) for his help and making it possible for me to experiment.

One final note, I have received the smaller LED's I want to try for "method 3" where an LED is placed directly in the optical path...That should be very interesting as well.

Keith