Over-perfoming objective, or does darkfield exceed brightfield resolution?

[Beatsy]

Sorry, the pic below is just constructed from a quick test shot - but is the source of my question. I've been tweaking the UV illumination setup a bit further, mainly to improve adjustment and add some diffusion to the Convoy torch light source.

This is a single 4-shot pixel shift image shot under 365nm UV illumination, using a Heine condenser, darkfield setting, with an Oly UPlanFL N 40x/0.75 objective. The whole image captured looked like this (the brighter patch in the middle is from not having the Heine fully "into" it's darkfield position).

But on pixel peeping, I was surprised to see (at least) the striae resolved on the A. pellucida diatoms, and fancy I can see dots too. That's not possible! Using 0.61*wavelength/NA to calculate "best expected resolution", it should be 0.61*365/0.75 = 297nm. But the striae are (at most) 260-270nm apart, and the dots are down around the 230-240nm mark. Huh!? I inverted the cropped image to make the dots a bit clearer. They're only visible in a few spots, but DoF is extremely shallow - so to be expected, and I didn't expect to see them at all with a dry objective.

As per the subject title, is this "better than expected" performance from an over-performing objective, or does darkfield improve on the usual (calculated) performance for brightfield? Or something else?

[jmc]

Like you Beats, when using darkfield, I get what looks to be better resolution than with brightfield, and get a similar effect with oblique light. I've read various descriptions of what is going on with some sites basically stating that 'darkfield gives better resolution than brightfield'.

On the Olympus site - https://www.olympus-lifescience.com/en/ ... darkfield/ - it states that "The resolving power of the objective is the same in darkfield illumination as observed under brightfield conditions, but the optical character of the image is not as faithfully reproduced (except when a specially-designed iris diaphragm is utilized to lower the effective numerical aperture with high-magnification oil immersion objectives designed specifically for darkfield microscopy).".

Perhaps constructive and destructive interference from the striae and puncta is making them appear more obvious when using darkfield.

[iconoclastica]

Like you Beats, when using darkfield, I get what looks to be better resolution than with brightfield, and get a similar effect with oblique light. I've read various descriptions of what is going on with some sites basically stating that 'darkfield gives better resolution than brightfield'.

Oblique lighting is like closing the condenser's aperture from one side. It decreases the NA but I couldn't tell to what extent. What it does do is improving the contrast, just like the aperture. I get very similar results either way, but OL looks much better since it mimics 3D and I find it easier to balance with resolution loss.

In my case I read that darkfield too results in loss of resolution - don't ask me where. I never questioned that since it is exactly my own experience, but now I realize, perhaps not necessarily correct. So I am following this with interest.

Beatsy, am I correct to understand that you use UV in transmissing microscopy? How does that translate into a visible image, and doesn't it do damage to your eyes?

[jmc]

In my case I read that darkfield too results in loss of resolution - don't ask me where. I never questioned that since it is exactly my own experience, but now I realize, perhaps not necessarily correct. So I am following this with interest.

I've also seen that said too (that it darkfield reduces resolution), but I have also seen it said that it improves resolution, so there seems to be mixed messages out there. Personally, my I think it improves visible resolution when I compare the same sample with brightfield, but this could be due to the increased contrast it provides.

[Beatsy]

Thanks for the comments folks. Addressing some of the points discussed...

Darkfield doesn't increase resolution, only contrast. I agree with Jonathan, (that it "looks to be better resolution"), but I'm sure that's only because resolution that was already there is just more easily perceived.

Most contrast techniques reduce resolution, albeit some only along one direction (oblique and DIC for example). The main exceptions are dark field and (closely related) annular illumination. Their low-angle oblique rays can "fill" the objective so it achieves it's max rated NA. That's assuming the NA of the condenser and emitted light cones match or exceed the NA of the objective. There is an exception though, when viewing subjects in "aqueous media". The lower RI of water stops very low-angle reflected and scattered light from getting through the coverslip to the objective - even if it's oiled to the slide. Those rays just reflect internally. This effectively lowers the NA of the illumination, so the objective can't perform at full resolution either. This probably relates to the reports of darkfield reducing resolution.

Beatsy, am I correct to understand that you use UV in transmissing microscopy? How does that translate into a visible image, and doesn't it do damage to your eyes?

Yes, 365nm UV transmitted. But I'm not using the oculars as I definitely don't fancy sunburnt retinas. All the light is diverted to the camera, a full-spectrum converted Sony A7r4 when I use UV. I view a zoomed-in image through the electronic viewfinder for focusing. I'll be substituting a monitor plugged into the camera soon. I've got the parts but need some space. Easier said than done...

I'm still mystified as to the clear visibility of those too-closely-spaced features from an objective that shouldn't be able to resolve them. Couple of extra things that might be factors...

1. That whole image shot is a 40 megapixel square. I think that's quite a bit higher resolution than most people would use for photomicrography (isn't it?) so perhaps this simply hasn't been noticed before. The 40x mag doesn't exactly make the features jump out.

2. The diatoms are mounted in Zrax which has a high RI of 1.7. Maybe that's doing the opposite of what water does, allowing even lower-angle rays to exit the coverslip. But that still doesn't explain how the objective can resolve them.

An interesting conundrum. But I'm stuck for ideas now...

----

And finally, I might as well use this slot to show the recent UV illumination improvements. I've reduced the fringing quite significantly, but what remains is entirely due to out-of-focus elements "diverging" in the hollow cone of light emitted by the Heine. The further OOF, the wider the divergence. I think I'm stuck with that now (when using the Heine) but it's acceptable. I've also got much smoother diffusion with very little light loss, using the small swing-in diffuser from an old Zeiss HBO 100w halogen lamp. It's fitted on the front of the Convoy UV torch. Here's another A. pellucida shot with this setup. Heine doing annular illumination with a 100/1.4 objective. Everything oiled to the slide and a 16-shot pixel shift image (160 megapixels for the full square, but cropped way in for this single diatom image). Second image is just a crop into the first to show the dots. Focus wasn't perfect (DoF is excruciatingly shallow), but I'm happy with it so far. More tweaking to follow...

Interesting little error, or malformation, in the dot pattern near the end. Never seen that before (not that I've seen many A. pellucida dots anyway, until now).

[Beatsy]

Aha! After further musing, I think I might have the answer based on a thing I didn't think to mention. Namely...

The 40x lens isn't corrected for UV, so after focusing first in visible-light, I have to move the lens as much as 100-150 microns closer to the specimen after I switch to UV, to focus the image again. Since the image (rays) emerge from the coverslip as an inverted cone, the front element of the objective moves onto a narrower section of that cone. Therefore it accepts more of the outer regions (the higher NA rays?) and effectively operates at higher NA itself (but shorter working distance).

Sounds plausible to me - so it's probably completely wrong. But I can't think of anything else.

[Scarodactyl]

Yeah, if the focus is closer the NA should be higher in this case.

[jmc]

Aha, that is interesting about the closer focusing increasing NA. With the Zeiss monochromat objectives which are just quartz lenses, there is a big focus shift as function of wavelength, so the NA presumably changes with wavelength as well. Next time I have one on the microscope I'll have to check and see which way focus goes as I the wavelength changes.

Oh, and epic photo Beatsy - much better than I am getting at 365nm with my setup. Not jealous at all

[Beatsy]

Aha, that is interesting about the closer focusing increasing NA. With the Zeiss monochromat objectives which are just quartz lenses, there is a big focus shift as function of wavelength, so the NA presumably changes with wavelength as well. Next time I have one on the microscope I'll have to check and see which way focus goes as I the wavelength changes.

Hopefully you have to focus closer, or effective NA will be reduced. It will be great to get confirmation (or not) from another observation or two. Look forward to hearing about your findings.

Oh, and epic photo Beatsy - much better than I am getting at 365nm with my setup. Not jealous at all

Thanks, but I think this is a clear case of "gear matters". I've got modern objectives and a pixel-shifting camera in play here plus "AI" tools for noise reduction (but not sharpening - it messes up near diffraction-limited images at pixel level). That all helps immensely. In addition, I'm only working at the soft(ies) end of the UV spectrum too, and optimised for only that. I couldn't even touch the shorter wavelengths you (and your gear) can handle, so we're not exactly comparing like with like. I'd suggest your 365nm results are significantly compromised by the needs and restrictions of being able to "go shorter" with the same setup. But just like you, I'm not at all jealous of that either

A related aside/rant - the phrase, "gear doesn't matter" is oft-repeated in photography circles and it rather winds me up! I know it's supposed to be a gross simplification, aimed at beginners, but it's frankly pure nonsense. I've given up arguing the point though and just use a stock answer now - "OK, if gear doesn't matter, show me a photograph taken without any".

[jmc]

Aha, that is interesting about the closer focusing increasing NA. With the Zeiss monochromat objectives which are just quartz lenses, there is a big focus shift as function of wavelength, so the NA presumably changes with wavelength as well. Next time I have one on the microscope I'll have to check and see which way focus goes as I the wavelength changes.

Hopefully you have to focus closer, or effective NA will be reduced. It will be great to get confirmation (or not) from another observation or two. Look forward to hearing about your findings.

Oh, and epic photo Beatsy - much better than I am getting at 365nm with my setup. Not jealous at all

Thanks, but I think this is a clear case of "gear matters". I've got modern objectives and a pixel-shifting camera in play here plus "AI" tools for noise reduction (but not sharpening - it messes up near diffraction-limited images at pixel level). That all helps immensely. In addition, I'm only working at the soft(ies) end of the UV spectrum too, and optimised for only that. I couldn't even touch the shorter wavelengths you (and your gear) can handle, so we're not exactly comparing like with like. I'd suggest your 365nm results are significantly compromised by the needs and restrictions of "going shorter" with the same setup.

A related aside/rant - the phrase, "gear doesn't matter" is oft-repeated in photography circles and it rather winds me up! I know it's supposed to be a gross simplification, aimed at beginners, but it's frankly pure nonsense. I've given up arguing the point though and just use a stock answer now - "OK, if gear doesn't matter, show me a photograph taken without any".

Will add it to the list of experiments for when I can tear myself away from my fascinating literature review work

Yes, you're right, gear does matter, so it is a bit of stretch to think that I can match your setup at 365nm. Skill also matters, and I still think I have a lot to learn with regards to microscopy and optimizing things.

Love the imperfection in the dot pattern. I've never seen that before either.

[Beatsy]

Oops, sorry. I was editing my post as you quoted it in your response.

Ooh, literature review. I'm not jealous about that either - but really not in this case

[iconoclastica]

A related aside/rant - the phrase, "gear doesn't matter" is oft-repeated in photography circles and it rather winds me up! I know it's supposed to be a gross simplification, aimed at beginners, but it's frankly pure nonsense. I've given up arguing the point though and just use a stock answer now - "OK, if gear doesn't matter, show me a photograph taken without any".

That phrase is used in conjunction to the (emotional) content matter of the photography (or it should be). 't Has nothing to do with the technical qualities.

[jmc]

Aha, that is interesting about the closer focusing increasing NA. With the Zeiss monochromat objectives which are just quartz lenses, there is a big focus shift as function of wavelength, so the NA presumably changes with wavelength as well. Next time I have one on the microscope I'll have to check and see which way focus goes as I the wavelength changes.

Hopefully you have to focus closer, or effective NA will be reduced. It will be great to get confirmation (or not) from another observation or two. Look forward to hearing about your findings.

Oh, and epic photo Beatsy - much better than I am getting at 365nm with my setup. Not jealous at all

Thanks, but I think this is a clear case of "gear matters". I've got modern objectives and a pixel-shifting camera in play here plus "AI" tools for noise reduction (but not sharpening - it messes up near diffraction-limited images at pixel level). That all helps immensely. In addition, I'm only working at the soft(ies) end of the UV spectrum too, and optimised for only that. I couldn't even touch the shorter wavelengths you (and your gear) can handle, so we're not exactly comparing like with like. I'd suggest your 365nm results are significantly compromised by the needs and restrictions of being able to "go shorter" with the same setup. But just like you, I'm not at all jealous of that either

A related aside/rant - the phrase, "gear doesn't matter" is oft-repeated in photography circles and it rather winds me up! I know it's supposed to be a gross simplification, aimed at beginners, but it's frankly pure nonsense. I've given up arguing the point though and just use a stock answer now - "OK, if gear doesn't matter, show me a photograph taken without any".

Finally managed to get an A. pellucida image I was happy with. Using 365nm light, a 63x Leitz Pl Apo NA 1.4 objective (on my Olympus BHB microscope), a Watson Quartz Cassegrain condenser, a 2.5x Nikon CF photoeyepiece, and my monochrome converted Nikon d850 camera (a 10 shot average using the multiple exposure setting). Light source was a 365nm LED torch, and I used 2x 365nm Edmund Optics 10nm bandpass filters stacked together. Cropped from the original image and the crop was resized from 1600 pixels across to 1000 for sharing here.

[Beatsy]

Deftly dotted and a nice clear result. Always difficult to visualise the scale at these high mags though. Here, you'd need a row of >4000 dots set out at this spacing to barely reach 1 mm. Boggling!

[jmc]

Deftly dotted and a nice clear result. Always difficult to visualise the scale at these high mags though. Here, you'd need a row of >4000 dots set out at this spacing to barely reach 1 mm. Boggling!

Cheers Beats. Yeah it's mad isn't it. I just did some measurements on the spacings. Striae spacing of 255 nm and a puncta spacing of 216 nm, which is in the same ball park as the values I have seen in the literature for this diatom. At just over 200 nm apart, that is approx 250x less than the diameter of a typical human hair (50 microns), 40x less than the diameter of a red blood cell (8 microns), and more like the size of a single virus.