Correcting for Refraction (or not!) in alcohol

[Sam Droege]

All:

We have been playing around with taking photos of insects floating in hand sanitizer inside of glass or plastic cuvettes. See our Flickr set at:

http://www.flickr.com/photos/usgsbiml/s ... 861396270/


This works reasonably well (and the way to go for soft-bodied things) except that we lose a small but noticeable amount of resolution compared to shooting the specimen in the air.

"Fresh Mosquito Pupae lighter_2013-09-15-17.20.03 ZS PMax by Sam Droege, on Flickr"


Below is an example of a specimen in air

"Fly in air test3x detail by Sam Droege, on Flickr


And the same specimen in alcohol in a cuvette

"fly in quartz 3 x detail by Sam Droege, on Flickr"


Not great shots but done just to illustrate the point.

OK, after a lot of talking and reading it turns out that the loss of resolution here is unavoidable because of the different refractive indices of air/glass/Hand Sanitizer and that without some sort of corrective lens it will remain so.

So I looked into the world of underwater photography, and they use acrylic dome ports to correct for the same problem. Except that the person I talked with said that the nearest the specimen can be is 9 inches from the nodal point of the camera which means I can't quite get the degree of magnification I would like. Additionally they thought that the dome acrylic would craze with any extended contact with alcohol and, yes, the dome has to be submersed in the alcohol, which creates logistical issues too.

There does not seem to be a screw on corrective lens to correct for this problem that I can find.

So, am looking for any additional ideas, while not a lot of people would buy such a thing I think that there would be a small market as many of us science guys need to take pictures of things in alcohol or at least water.

Thoughts?

sam

[rjlittlefield]

Sam,

First, let me correct some problems with your posting. Here at photomacrography.net, the markup for links and images uses what's called BBCode. It's conceptually similar to the HTML markup that you provided, but the syntax is slightly different.

I edited your post to use BBCode. The affected code now looks as shown below. Notice the use of url and img tags, contained in square brackets [] instead of angle brackets <>.

--Rik

Code: Select all

This works reasonably well (and the way to go for soft-bodied things) except that we loose a small but noticeable amount of resolution compared to shooting the specimen in the air.

[url=http://www.flickr.com/photos/usgsbiml/9764583566/]"Fresh Mosquito Pupae lighter_2013-09-15-17.20.03 ZS PMax by Sam Droege, on Flickr"[/url]
[img]http://farm8.staticflickr.com/7385/9764583566_61acb385a2.jpg[/img]

Below is an example of a specimen in air

[url=http://www.flickr.com/photos/usgsbiml/9736455115/]"Fly in air test3x detail by Sam Droege, on Flickr[/url]
[img]http://farm8.staticflickr.com/7347/9736455115_c52db2a1ee.jpg[/img]

And the same specimen in alcohol in a cuvette

[url=http://www.flickr.com/photos/usgsbiml/9736465307/]"fly in quartz 3 x detail by Sam Droege, on Flickr"[/url]
[img]http://farm8.staticflickr.com/7442/9736465307_536bee80bb.jpg[/img]

[Sam Droege]

Thanks Rik

In looking back at Flickr. I see that I had not flickred the right button to get bbc code rather than html.

Thanks for the correction.

sam

[rjlittlefield]

Sam,

I'm glad to hear that the formatting issues are under control.

Now, regarding your question about shooting through liquid. I see that the shot in quartz cuvette filled with hand sanitizer is much different from the shot in air, but there are several different kinds of differences and I'm not sure what all the causes are.

The one problem that for practical purposes is unavoidable when shooting through dense medium is what's called "spherical aberration".

The name may be a bit misleading in this particular case, because the aberration actually results from having a plane of thick medium unexpectedly inserted into the optical path. The name makes sense in terms of the ray paths in this case compared to ray paths through spherical lenses, which have the same issues.

Anyway, what happens is that rays passing through the edge of the aperture focus at a slightly different distance from rays passing near the center of the aperture. This results in a loss of contrast for moderately fine detail, even though contrast for coarse detail and for very fine detail near the diffraction limit can be close to unchanged.

However, the degradation due to spherical aberration is often less than people think, and the symptoms do not match what I see in your images. See HERE for what I'm pretty sure is image degradation caused by spherical aberration. Notice that this example is shooting at 5X, wide open at f/2.8, through 2 inches of water. Shooting as you do, stopped down to f/5.0 at 3X, and through a much thinner layer of dense media, I would expect much less severe symptoms. But even in my example, notice that there's a loss of contrast and some levels of detail disappear, but there's no change in brightness or image noise.

In comparison, your cuvette image is significantly brighter than the air image, with way more pixel noise and some pretty dramatic halos. It's a stretch for me to blame those on spherical aberration. I'm suspecting that you have something else going on, but without having hands on your equipment it's hard to tell what. Offhand I suspect that your cuvettes are not as good as they need to be for this purpose, and maybe that hand sanitizer isn't quite perfectly clear either.

To get a handle on avoidable versus unavoidable problems, you might consider running some trials with your specimens sitting behind windows that are known to be optically smooth. See http://www.photomacrography.net/forum/v ... 479#102479 for a test setup that I've found helpful for running experiments.

--Rik

[Sam Droege]

Rik:

Interesting and it seems likely that spherical aberration is not the problem, good dialog though. I think the haloing in the picture is just a light problem...I had the flash firing too high.

The shots were taken through fused quartz cuvettes which are designed to be the best transmitters of light, though not necessarily optically the best (though I am not sure that they wouldn't want them optically high quality since they are measuring precise levels of light passing through chemical mixes). A new quartz cuvette goes for $100+ easy.

Note that for the tests I used (at least in some of them ) were in alcohol not Hand Sanitizer to see if the HS was an issue...it was not.

That said, I was concerned about the glass and modified an old photo stand so that I could shoot vertically into an open petri dish of alcohol, sans glass interface. I didn't publish the pictures because they were basically exactly the same sort of fuzzing (the colors change too, but that may simply be due to some sort of wetting phenomenon).

I tried to hunt down the link I found that talked about refraction differences causing loss of resolution, but I couldn't. The point I took home (which could be actually Missing the point) was that some of the pathways coming from the image were being bent at such an angle that they were "lost" to the opening of the camera lens. This has to be a misinterpretation, but for lack of any other explanation it seemed in my mind to be reasonable.

sam

[rjlittlefield]

How thick a layer of alcohol are we talking about?

--Rik

[Sam Droege]

Just covering the specimen so the upper most parts would only be a few mm.

[ChrisR]

What happens when you stop down? SA should get better, whereas some kind of light scattering/dispersion would not change, possibly.
I notice you have much blacker backgrounds in this first shot, compared with many of the flikr examples. Is that Photoshop, or a light mask?
I was thinking you might need something like:

With the tubette just far enough into a black tube to not mask light to the critter.

[rjlittlefield]

It's definitely not spherical aberration at f/5 and a few mm of alcohol. Even with 10 mm of alcohol, I calculate that the maximum added aberration is less than 0.007 wavelength, about 40 times less than what's needed to be visible.

Here is an example to illustrate. This is a high contrast target, chrome-on-glass, Edmund #58-198, diffuse backlighting, photographed using Canon T1i and MP-E 65 at f/5.0 and 3X, using a test setup similar to HERE through no water, 13 mm of water, and 40 mm of water. Single shots, EFSC, 1/30 second. These are actual-pixel crops.



After I shot them, I noticed that the first two shots were slightly off perpendicular to the face of the setup, resulting in a bit of color fringing but no obvious astigmatism. I tried to correct the tilt for the third shot, but I think I overshot and put the fringes on the other side.

Numerical musings... Group 7 Element 4 is 181 line pairs per mm. I calculate cutoff at 550 nm under these conditions as 272 line pairs per mm with MTF=22% at 181 lp/mm. What I see here looks real close to diffraction limited, even through 40 mm of water. That's consistent with another calculation that says worst wavefront error = 0.03 lambda in this case.

--Rik

[Chris S.]

Sam, do you have any shots of the fly in the cuvette without alcohol or sanitizer? Based on your petri dish test, we'd predict that the air-filled cuvette should perform similarly to air alone--have you confirmed this?

Elaborating on something Rik pointed out, the pixel noise is much higher in the images of the flies in alcohol and hand sanitizer within the cuvette, compared with the air/no cuvette image. It looks similar to what we would see if an image were underexposed and pulled up in post, or shot at a much higher ISO setting, where the sensor's dynamic range is lower, and noise is amplified. Can you elaborate on the cause of this? (Was the ISO the same in all shots? Were all images shot to the right of the histogram? Does the cuvette and/or its contents block or divert some of the light?)

This may be important for two reasons. First and most obviously, the pixel noise may be contributing to reduced visual quality. Second and not-so-obviously, the apparent size of a light source can greatly impact the resolution of reflective subject features. With reflective subjects (or reflective features on otherwise non-reflective subjects), small light sources can utilize (or fill) less than the entire aperture of the lens. This has much the same effect as stopping the lens down--you gain depth of field and lose resolution. But unlike stopping down, the effect need not be even across the entire subject--less mirror-like portions of the subject will be effected less. To my eye, your in-cuvette images have some visual hints suggesting they might have been illuminated from a narrower angle. So I wonder if the cuvette or its contents might be blocking or redirecting part of the light, making the source effectively smaller?

BTW, I'm glad you raised this question. I'd looked at your cuvette images with particular interest, as have been planning to try shooting some small spiders in hand sanitizer (they would arrive in alcohol). I saw the "small but noticeable" loss of resolution, and wondered if it could be avoided. My intent was to use not a cuvette, but a small "aquarium" made from a cover slip on the front, silicon caulk for the sides, and a slide for the back.

By the way, Sam--welcome to the forum!

--Chris

[Sam Droege]

Chris / Rik: Good conversations.

I have not tried a specimen simply in the air in the cuvette...will do that when I get back in the lab (am on travel right now). We take all pictures with a black background or a black ground of black clay (effectively it comes out gray in pictures), when the back ground is black we will push it to 0 0 0 in PS.

The ISO would have been the same between the two pictures (1-200), but the light dynamics in the cuvette are definitely different with the wall of the cuvette partially reflecting light resulting in an overall grayer background, that I can't really push to pure black. I wish the cuvette's were bigger (1 cm square) as we have significantly decreased the reflective light by putting a piece of black plastic on the floor of the cuvette. I will try stopping down, but above f8 or so I start losing resolution big time (actually, having taking pictures at high fstops the resulting pictures do seem very similar to the alcohol problem). It would be nice to get purer black background, will run some more test, but note that the petri dish pictures had the same loss of resolution and the setup I had gave a near perfect black background.

Note, one cool thing to do is use a dab of HS to fix the specimen in place in a petri dish and then use alcohol to cover the specimen in a uniform way. If the specimen is expendable then you could also use water.

sam

[TheLostVertex]

I have not tried a specimen simply in the air in the cuvette...will do that when I get back in the lab (am on travel right now).

If you are going to reshoot to test, then perhaps you can cover all your bases at the same time. Take your subject, mount it with a pin so that it may be suspended from the pin(pin above the subject). Then try taking a stack without the cuvette, slide the cuvette from under the subject without altering its position(still mounted to the pin), and then a stack after filling the cuvette with alcohol(again without unmounting the subject). Hopefully then you can get 3 images that are fairly close to aligned with each scenario.

Post processing them all the same and without and dramatic contrast adjustments would help to see any actual optical differences as well, I imagine.

[Sam Droege]

Steve

This sounds like a good plan. ..

sam

[g4lab]

I am not certain why you would want to immerse these specimens.

In gemology we immerse transparent stones in order to suppress surface reflections of the facets. This allows us to see inside and examine inclusions. I don't think you are trying to look inside your specimen.

To accomplish that we try to match the refractive index of the stone as closely as possible and there is a list of liquids that can be used.

If you have a fused silica cuvette the Refractive Index is 1.46 . The RI of ethanol is about 1.36 hardly higher than water which is 1.333. On the other hand white mineral oil usually in the neighborhood of fused silica. So is carbon tetrachloride, and glycerol. All these liquids have much higher surface tension than alcohol , although ethanol and propanol are miscible with glycerol and might make it wettable enough to use.

Most gemologists do not have a horizontal nor inverted microscope available. So they often perform immersion examinations through the meniscus of a liquid surface. This is definitely less than optimum and if someone starts doing much of this kind of work they will often buy or build a horizontal microscope or an immersion cell with an optical quality bottom.

http://www.gemologyproject.com/wiki/ind ... rsion_cell

http://www.szfable.com/e/products/Gem_M ... copes.html

Most spectroscopy cuvettes I have had the chance to examine (a large number) are more than adequate optical quality to shoot through.

I have been watching for such cuvettes because the ones you refer to with approximately one cm square area in plan view , also are not quite big enough for larger gemstones and the tweezers that hold them. There are however "colorimeter cuvettes" that are as large as 60mm. When they get that big they are usually just glass but it is good optical quality and water white. (which is a whole other difficult thing to find)

http://www.hellma-analytics.com/kuevett ... ubmitted=t

[TheLostVertex]

I am not certain why you would want to immerse these specimens.

My guess would be for photos of preserved soft bodied insects, and/or the possibility of posing insects in a suspension of hand sanitizer.

I found the information and links for the use of an immersion cell very interesting, as I have little knowledge of gemology. Makes me want to try it