Hello all,
I thought some of you might find this comparison interesting.
Resolution of diatoms can be improved by using shorter wavelength illumination. The following examples were all shot with an Olympus 100x/1.4 UPlanSApo objective and a Sony A7riv camera converted for full-spectrum imaging (for the UV capability).
The wavelengths and filters used were white light (no filter), blue (450nm cutoff), violet (405nm bandpass) and UV (365nm bandpass with a 365nm LED light source).
The diatom is 150µm long and each version of it is a ~160-image focus stack. All frames were captured using 4-shot pixel shift and merged into true-color RGB images using PixelShift2DNG, then converted to monochrome after stacking.
Note: the 450nm filter blocks wavelengths longer than 450nm but passes a spectrum extending to violet at the short end. That may explain why the result is so similar to the 405nm bandpass version (closer than I expected anyway).
wavelength comparison V2-4k.jpg
These are 100% clips from the centre of the right hand edge of each version.
wavelength 100pct clips V2-pmn.jpg
Cheers
Beats
Hello Beats,
the increase in resolution at 365 nm is of course very noticeable. But if you adjust the contrast of the blue image slightly to that of the UV image, the difference becomes somewhat less pronounced. In the blue image, there appear to be image distortions (periodic darker areas along the left edge of the image) that are almost completely gone in the UV.
Vergleich Blau UV365.jpg
Hubert
Hi Hubert,
Thanks for your response. I agree the increased contrast makes it *look* clearer but resolution isn't improved. It's even a little degraded in some areas.
On the rows of pores that are more closely spaced, the lighter gaps between the pores have shrunk with the higher contrast such that the pores are pretty much touching. In the original they are separated more but that's less easy to see due to lower contrast.
I played the trade-off to favour resolution, yours favours clarity. Neither are wrong, just choices made...
Cheers
Beats
Dear Steve,
thank you for sharing this comparison with us.
I think it will help a lot to understand and to find a way and compare it with their own results.
best
anne
Hello Beats,
Thanks for your effort in comparing.
Doesn't it make more sense to compare just a single image?
I think stacking also affects the resolution and may create artifacts.
Ideally without any processing.
As far as I understand, doesn't UV light require smaller stacking steps and therefore more overall?
Regards, Frank
Hello,
ZitatI think stacking also affects the resolution and may create artifacts
Yes, that is also my practical experience. This stacking problem is increased when the source images to be compared do not have the same contrast. In this case, the program may have trouble imaging identical object areas with equal sharpness, i.e., in the same focal plane.
Hubert
Hi Frank,
Thanks for replying.
"Doesn't it make more sense to compare just a single image, since there should be a single point?"
That's a little complicated. The RI of the mountant (Darax) increases at shorter wavelengths. That squeezes a bit more into an objective's depth of field and effectively shifts the plane of focus when the wavelength of light is changed. So you'd need to find the same focus depth again which would be quite difficult given the super thin DoF at high NA.
Besides, I'm not "doing science", I simply wanted to compare results in the final images after following my usual process (stacking, contrast enhancements and light sharpening) which I always use when photographing mounted diatoms.
"I think stacking also affects the resolution and may create artifacts."
This is true with image alignment enabled (x/y shifting and scaling) but I disable them. The stage of my BX61 runs straight along the optical axis and I'm careful to ensure the brightfield illumination is exactly central too. So no shifting or distortion in the stack.
There may be minor artefacts caused by the low contrast and "unsharpness" of small details as the stacking software might interpret different colours a little differently. But as above, my goal is detailed stacked images, not science, so I can tolerate these tiny imperfections.
"Ideally without any processing."
Haha - good luck with that! ;D
Here's an example (though it's a stacked image, not a single shot).
A stack of Surirella gemma mounted in canada balsam. Shot in brightfield with a 100/1.4 objective under UV light
s-gemma-100-ps4-uv-pmn.jpg
After processing for "visibility"
s-gemma-100-ps4-uv_1_1_1-pmn.jpg
"As far as I understand, doesn't UV light require smaller stacking steps and therefore more overall?"
Yes, shorter wavelengths give a shallower DoF. But the increased RI of the mountant at shorter wavelengths also squeezes a bit more into that DoF. I dodged the whole issue by setting the focus step size to suit UV then focus bracketed the other colours with the same setting.
Hope that clears things up.
Cheers
Beats
Hello Steve,
I realize that you're working at the highest level of light microscopy here.
Of course, the structures should be recognizable even when converted to black and white, so some image processing is obviously required. Thank you for your impressive example without any processing.
I simply thought you could find the same section with the same sharpness in the stack of 160 images and compare them.
But that probably won't provide any new insights, as I understand it now.
Thanks again for showing us this comparison.
Best regards, Frank
Great stuff, Beats. That first couple of images of the whole Tryblionella valves are stunners. Any chance of a stereoset/anaglyph?
Best,
René
Hello Steve,
a very impressive comparison.
Tx
Gerhard
Dear All,
Presumably in this case, the limiting factor isn't the wavelength of light, but rather the combined numerical aperture of the objective and the dry condenser seems to be the limiting factor.
Microscopy – resolution comparison between visible and UV light:
https://jmcscientificconsulting.com/microscopy-resolution-comparison-between-visible-and-uv-light/ (https://jmcscientificconsulting.com/microscopy-resolution-comparison-between-visible-and-uv-light/)
Helmut
Hello Beats,
how did you take the UV pictures? The Axiocam 506 has a detection range of 400 - 700 nm.
Sincerely
Alexander
Zitat von: Aljoscha in Juli 22, 2025, 20:04:24 NACHMITTAGSHello Beats,
how did you take the UV pictures? The Axiocam 506 has a detection range of 400 - 700 nm.
Sincerely
Alexander
Hi Alexander,
It was momotaro who mentioned Axiocam 506. I use a Sony A7riv converted for full spectrum with a custom 365nm LED light source shining up into the condenser from underneath.
Cheers
Beats
Hello Beats,
Thank's a lot for sharing your valuable experience with us.
Very lucid comparison. Great.
May I ask you some questions?
Which bandpass filter was used for 365nm, and which kind of projection method (direct,photo eyepiece, afocal...)?
What about empirical values about different objektives used in the 365nm range? Are there other properties (except NA) having a visible impact on the resulting image?
And last but not least, did it make sense to use a non converted camera (with bayer sensor) in combination with 405nm?
Greetings Holger
Hi Holger,
In answer to your questions...
Which bandpass filter was used for 365nm,
I use a 1.25 inch (31.75mm) Baader U "Venus" filter which has a bandpass around 365nm. Unlike the cheaper 365nm filters, it blocks all other wavelengths including IR. This Baader filter is the last piece of glass in front of the camera sensor (so it can block any fluorescence generated by the UV too).
and which kind of projection method (direct,photo eyepiece, afocal...)?
Direct projection onto the sensor. This only gives a 24mm image circle so I generally use APS-C crop mode (26 megapixels with Sony A7riv). But for a slightly taller FoV, I use full frame (61 megapixels) and crop to a full-height square in the centre. This captures the whole circular FoV, but with black corners.
What about empirical values about different objektives used in the 365nm range? Are there other properties (except NA) having a visible impact on the resulting image?
The most important thing is that objectives must be corrected down to 365nm wavelength for good results. My Olympus 100x/1.4 UPlanSApo is the only one I have that's suitable. I also have an Oly 60/1.42 SPlanApo which appeared to work at first, but the image is only useable from the very centre of the frame. Away from the centre the image is awful. That one works great with 405nm though - which is handy for larger diatoms needing a larger FoV (less magnification).
And last but not least, did it make sense to use a non converted camera (with bayer sensor) in combination with 405nm?
Yes, it makes sense, but I only have the converted A7riv dedicated to this task (and the Bayer filter is still in place). However, the Bayer filter and conversion from RAW to (say) TIFF will steal resolution. That's because only one of the four colour channels (RGGB) in each pixel is actually exposed to light of the corresponding colour. The other three are extrapolated (guessed) from neighbouring pixels.
The remedy is to use 4-shot pixel shift to ensure every channel in each pixel is actually exposed to light of the relevant colour so Bayer decoding isn't needed. I reluctantly chose to have my A7riv converted purely because it supports pixel shift.
This is important for UV imaging too. UV gets through the Bayer filter with different intensity in the different colour channels. R and B capture the most signal, G not so much. Note, it's still UV light hitting the sensor as the Bayer cells don't change the colour/wavelength of the light.
Cheers
Beats
Hi,
This is a load of elaborate information!🙂
It's much appreciated!👍
Holger