Diatoms from Williams Bluff, Oamaru, NZ

[Beatsy]

Hello all,

I found a 1.8g lump of diatomite at the back of a drawer, processed it, and made some open strews to search. The cleaned material, from Williams Bluff, Oamaru, NZ, was a car crash of broken frustules, radiolarians and spicules but it didn't take long to pick out several well-preserved diatoms to mount and photograph too.

This coincided with some ongoing tinkering to improve my diatom images, so I used a few of these pickings as test subjects. I'm pretty close to consistently producing a "look" that I'm content with. Finally! Well, content until the urge to tinker returns Here's a few recent results in order of largest to smallest forms.


Triceratium favus - outside view


Triceratium favus - inside view (different diatom than above, but close to the same size)


Auliscus and A. stoeckhardtii var.subpunctata


Glyphodiscus grunowii var. lacunosus (thanks to Michel Haak for the ID)


Cheers
Beats

[deBult]

Nice "clean" look.

Maarten

[Jürgen Boschert]

What a marvelous detail in these photographs!

Thanks for sharing.

[bill2penn]

Hi Beats,

Beautiful images!
They almost look like they are drawings from Schmidt's Atlas but more detailed.

I guess I may have to go back to look at my William's Bluff samples that I collected over 20 years ago. Most of them were not that great as I remember. Did I send you some of them? I don't remember.

Best,
Bill

[Alex H.]

Hi Beats,

Those are beautiful photos. I find the images of Triceratium favus especially stunning. They look like first-class illustrations.

Would you mind sharing a bit about your imaging technique?

Best regards,
Alex

[Beatsy]

Thanks for your kind and encouraging comments everyone. Much appreciated.

Bill: I think you did send me the Williams diatomite, in the first parcel of samples I got from you many years ago. I always split diatomite into at least two portions to insure against cleaning disasters, which explains how this crumb got detached from the crowd and hid itself for years. A pleasant little surprise. Also a useful test of recent imaging and cleaning changes to compare with previous methods (using identical material). Improvement was verified in both cases!

Cheers
Beats

[Beatsy]

...
Would you mind sharing a bit about your imaging technique?

Hi Alex,

I don't mind at all.

There are some non-imaging things that also contribute to improving resolution and image quality and most of the techniques I use are specific to my setup. For now, to avoid writing an essay, I'll just list the key equipment and general processing steps involved, but I'm happy to answer more specific questions later.

1. Diatoms are cleaned, selected, and attached to a 0.17mm coverslip (or thinner if the diatoms are thick). Then they are mounted in Darax which has a high refractive index of 1.7+ (to improve contrast).

2. I use a custom 365nm UV illuminator positioned over the illumination port, under the condenser on my Olympus BX61. Short wavelength light increases resolution. Using an Olympus UPlanSApo 100/1.4 objective this setup resolves features down to ~150nm apart (compared to ~240nm using white light).Both condenser and objective are oiled to the slide. I use brightfield illumination only, because all contrast-enhancing lighting methods (e.g. DIC, Phase, COL etc) compromise resolution.

3. My camera is a Sony A7riv that has had the IR/UV filter removed from the sensor. The Bayer matrix is still in place to allow for colour photomicrography and I just drop an IR/UV blocking filter over the illumination port in the base of the microscope for that (otherwise images would look fogged due to high IR output from the 100W halogen lamp).

4. I focus stack all my diatom images using a WeMacro Micromate attached to the fine focus knob. At this magnification and resolution, step size of the focus movement between shots is 100nm or 0.1 µm. The capture process is automated, so I just set start point and number of shots, then leave it going. The camera is tethered to my PC so images are saved directly to disk and displayed in Capture One Pro. I use the exposure evaluation histogram (a RAW histogram without camera profiles or colour correction applied) to ensure exposures are as hard to the right as possible for maximum dynamic range. You can get ~2 stops more exposure compared to what the histogram on the back of the camera shows as the maximum.

5. Note: 365nm UV gets through all the Bayer matrix filters (R,G and B), but most through R, a bit less through G and least through B. The filters don't change the colour of the light, it's still UV light hitting pixels. I don't want captures interpreted as colour images by the usual RAW decoding though, so I use 4-shot or 16-shot pixel-shift mode. That means 4 or 16 exposures at each focus bracketing step. After everything has been captured, the groups of 4 or 16 images are combined and converted to DNGs using PixelShift2DNG. After converting to monochrome and turning off colour correction and camera profiles in C1, the images are exported as 16-bit TIFFs for stacking. The conversion to mono combines R, G and B values in each pixel into one averaged luminance value. This greatly improves signal to noise ratio and increases dynamic range ~1.5 stops.

6. Stacking: I use both/either Zerene Stacker or Helicon Focus. I much prefer to use depth map stacking (DMAP in Zerene or Method B in Helicon). However, if a diatom has large areas with little detail, this doesn't work well - so I switch to pyramid max mode instead (PMAX in Zerene, method C in Helicon). This doesn't get the smeary artefacts but does boost the contrast quite a lot, which can make some fine detail look a bit "crunchy".

7. The final stacked result (16-bit) is cropped and exposure-adjusted in C1, then processed in Affinity Photo 2 to enhance the visibility of small details. This involves...

7a. A very light denoising in Topaz Denoise 6 - if necessary.

7b. Use frequency separation to create a greyscale high-pass layer over a (now slightly softer) low-pass colour layer (although there's no colour here anymore it's just shades of white). The radius of the filter area is lowered until the smallest details are only *just* showing in the greyscale image. Now, select this high pass layer and run it through the tone mapping persona. Overall tone mapping (HDR) is turned off, but I crank local contrast to 50% or more. This enhances contrast of small details *only* so the overall image stays more natural-looking.

7c. flatten the image (combine the layers), select the diatom(s), cut them out and place in a new layer. Put a solid off-white background under the diatom layers. Add a scale bar. Flatten the image again and return to Capture One

8. Last tweaks to exposure, then export as jpg as needed.

That sounds like a lot of work, but it isn't really. Sure, it takes a fair bit of elapsed time (over an hour for deep, 16-shot stacks), but I can do something else during most of it.


Cheers
Beats

Edit: I forgot to mention that I don't do any sharpening at all. In fact, with 16-shot pixel shift images stacked using PMAX, I will often lightly blur the full-size image before export. Usually Gaussian blur with a 0.8 pixel radius. Since the 16-shot images are more oversampled, no detail or resolution is lost but the overall "look" is much cleaner and the crunchy, pixelated effect on small contrasty details disappears.

[Rene]

As Bill says, drawing-like and as pleasant as the Schmidt atlas, while pushing brightfield microscopy to the extreme limits. Respect!

Thanks Beats,

René

[sushidelic]

Hey Beatsy,

those are phenomenal!

Best regards,
Michael