Ever since I witnessed the beauty of diatoms under a half-broken Olympys BX40 microscope, I have been seeking better image quality. My only knowledge on Diatoms go no further beyond a Wikipedia page and that they are diverse in shape.
As the stereotypical resolution-crazed amateur, I started with a handful of cheap Plan Achromat objectives. These came with the broken microscope I purchased., and they are honestly just fine. Chromatic aberration (CA) is accounted for in 2 wavelengths, typically blue and red. My goal was to obtain a set of Olympus’ UPlanFL objectives for Phase Contrast Microscopy.
The letter “U” stands for “universal”, “Plan” means it is flat-field corrected, and “FL” refers to fluorite glass being used. Fluorite objectives are also called “semi-apochromats”, they also account for CA is 2 wavelengths. However due to the unique properties of fluorite glass, they have better properties than achromatic objectives, making them the choice for a wide range of microscopy methods.
Each manufacturer has their own cryptically coded and annoying abbreviations. There exist no standard and I would argue a standard will never exist as such mechanisms are often patented and trademarked. For example, Nikon took a page out of Sigma’s book and trademarked the term “Noct (nocturnal)” for their large aperture lenses, f-stops at 1.2 and above presumably gets this designation. There is currently one “Noct” lens you can buy by selling a couple vital organs, and another that is planned. If we include the original “Noct-Nikkor”, that is a whopping 3 of them. At least their Z-mount lenses are called “S” rather than having “S-line” written on, that would make them sound like products you will find in the naughty area of downtown.
Rants on unnecessary codes aside, I quickly obtained a phase contrast condenser from a kind Thai surplus seller. I purchased 2 objectives, a 40x Achromat Ph2 and a UplanFL 10x Ph1. Time for more abbreviations, “Ph” refers to the phase annulus that must be dialled to on the condenser. The annulus being deployed seems to correlate with the numerical aperture (NA) of the objective. Later on, I acquired a UPlanApo 40x oil immersion objective for an unbelievable price. I was told that apochromatic phase contrast objectives are extremely rare and very expensive. They are said to be non-standard, Olympus’ objective selector sure did not list such an optic.
Gathering up my small army of objectives, I began to play around with a “centric and pennate” diatom comparison slide, made by the kind British Diatomist Klaus D. Kemp. I have previously imaged this slide with my 40x achromat and obtained very good results.
The background is gross and the typical phase contrast haloing is present, but this is nothing to be worried about, as a simple background swap gets rid of all the gunk. The achromat has a respectable NA of 0.65, yielding very dimensional results with nice levels of contrast.
My 40x apochromatic objective has an NA of 1.0, it requires oil immersion, it has to be better, right?
Eeeeeek, that is well, just bad. Contrast levels are higher, but it is overall just horrible. There is no detail in the diatoms, there is a myriad of stacking errors and other weirdness. Usually, people are quick to blame the inefficiency of software, their cat and basically anything but themselves. The problem however is usually the brain (often silly, in my case) behind the camera. A quick analysis led to several points:
- Sharpening artefacts being included as detail → turn the sharpening to zero
- Lighting errors, shadows being cast which is an indication of light source’s position → tune the condenser
- Phase contrast effect is lacking → I am stupid, I used the Ph2 annulus!!!
- Detail lacking → change Zerene Stacker’s Dmap settings
Upon my attempt to fix everything, here is the new stack.
Alright, this looks like a high NA objective. The problem unfortunately remains, I still think the low NA achromat yielded better results! Zooming all the way in, the achromat clearly has a far better definition of edge, it is also cleaner. Despite that, the oil immersion 40x is clearly sharper, boasting far superior resolution. At that moment, I was happy with the result.
No good drama comes without more pitfalls. “The higher the NA, the better”, right? Well, this is an utter fallacy. What do I actually want? The best scientifically useful image, or simply the best looking one? If I was to take photos for myself, I am going with the latter!
Klaus D. Kemp’s arrangements simply look amazing under the microscope. Here are 100 unique diatoms arranged into a circle. This requires a lot of skill, patience, DIY mind, and advanced techniques. There are microscope slides dating to as early as the 1850s which feature such amazing skills, I can only respect the diatomists behind these slides.
Look at the colours, are they not a thing of beauty? Even though the resolution is low, the colours are vibrant and pleasing. The saturation slider is untouched. These colours are actually interference colours, diatoms themselves should be colourless after being made into preserved microscope slides. Thomas Young’s “Double-Slit” experiment demonstrated the wave-like properties of visible light, which ties into the existence of interference light. A very simple, every-day example are soap bubbles, they exhibit amazing chromatic patterns. Apple’s iPhone X and more recent models display this on the expensive box that houses the phone. Olympus has an explanation and tutorial here.
Unfortunately, at 10x, I am wasting at least 75% of the frame. Lucky for me, a 20x high NA objective was on its way.
Admittedly, I abused the saturation slider. +75 in Capture One, the colours are just not there. At higher NA, interference colours are lost. What I do get is far superior resolution, an NA of 0.3 at a mere 10x simply holds no chance against 0.75 in the realm of resolution.
What about merging these two? Maybe I can magnify the lower resolution exposure by 200%, overlay it on top of the 20x exposure, and add in the colours! Turns out to be a good idea.
I do not see any penalties on image quality, the halos can be cleaned up with easy masking. I will discuss the technique in future posts, I need to experiment with it some more.
At this point, I wanted the most out of my 40x and I am on the looks of a 100x. The 40x results are still sub-standard. Nearly one month of learning, I decided to put my knowledge to the test. Here are my current understandings:
- Double immersion gives the best results. This is when immersion oil is dispensed on top of the condenser and stuck to the bottom of the microscope slide. This makes cleanup an even bigger pain, but I am happy to deal with that.
- The condenser must be centred properly. I simply use my 4x objective, close the condenser aperture (lowest NA), and put the polygon (hexagonal I believe) in the centre
- The phase annulus must be centred properly. I was always puzzled by the antenna on my condenser. Turned out I am too lazy to even read a manual, they are used to centre the annulus. Doing so yields far better results. I simply use my 10x UPlanFL at Ph1, slap a 50mm onto my Z6 and place it on top of the camera port. I then use the antenna to centre the annulus. It is not necessary to use a centring telescope, you can simply look into the tube lens. My light source is bright even at lower intensities, so I prefer to use the camera.
- Oh, and clean the slide prior to imaging, saves some whinging about image pollution.
I received a “Diatomace Prize Medal” slide from Britain, it featured some nice diatoms of the “Cocconeis” genus, set at different angles. The stack was good actually, it looked alright?
I can see the overall structure and shape, it is however very soft. Too soft from my experience. Well, turns out, I had the NA of my 40x objective set at 0.5! I laughed at how stupid I was. I already cleaned up the oil, I was set for bed, oh well.
My 40x objective has an internal iris that allows the adjustment of NA. Despite how soft it is, I actually like the results. It is clean, all the crucial detail is there… I do want more detail though! If I wanted to use my 40x at 0.5 NA, I should just stick to that achromat and buy my kidney back. I set the NA back to 1.0 and ran another stack.
Oh brilliant, this looks worse than the 0.5 NA stack, what have I done?
Fair to say, it is objectively worse. Maybe “it is SHARPER”, but so what? There are too many problems here. Again, I analysed the stack and the results, and came up with solutions.
- The seven diatoms are on drastically different focal planes, stacking them together generates a lot of computational errors → split stacking
- The “rim” of the diatom and the “centre” are often on different focal planes → split the stack into “rim” and “centre”
- 2 of them seems displaced, as they span several focal planes → more split stacking
- Phase contrast halo → replace the background and mask out the halos
- The small holes present in the diatoms are maybe hallow, or maybe they are structural. Scanning electron microscopy images suggest both → leave them there
- “Olive like” rendering due to the stack → rigorous selection of start and finish images for stacking
I categorised the 0.5 NA stack into 7 segments, and selected the images for stacking, corresponding to each of them. The Zerene Stacker Pmax results were all subsequently discarded due to odd-looking artefacts that resemble over-sharpening.
This looks far better. After about 10 hours of processing, I finalised the photograph.
I believe that I have done enough to respect the original Diatomist and this high power objective that landed in my hands because of destiny… alright I will stop. This is indeed a slide from 1867, the quality is astonishing. Unfortunately, I am only using one-ninth of the frame — the centre out of nine equal quadrants. A 100x UplanFL is common, but ones for phase contrast are not. They are usually too expensive too.
There are always improvements. My purchase of the Z6 was for my photomacrography and frankenmeasurescope setups, which feature low power metallurgical objectives. I never saw myself buying a microscope and using higher NA objectives. For example, the 40x matches to a monochrome fullframe sensor of 28 megapixels (MP), the Z6 has a 24MP colour back-side illuminated (BSI) sensor and an optical low-pass filter. The 20x at 0.75 NA is even more powerful, 64 MP, monochrome! If I buy a Z7 or an a7r4, or a smaller sensor camera with high MP, I will get more details.
I plan on sharing my microscope stacking setup in the future and write more articles about microscopy. Stay tuned and thanks for reading!