fern spores

[iconoclastica]

The quest for more and better fern spore photos continues. I am getting better results now, and faster too. I changed the way I prepare my slides, got better and more practical understanding of the NA, did a lot of experimenting, and perhaps the new condenser helped too.
Here is one of the more recent resuts:

Dryopteris aemula, TLM, Nikon Fluor 100x/1.30

Stacks like this one are suitable for 3d-fication too. I like the anaglyphs for contrary to stereopairs they don't impose size limits. Some do well in colour, others look better as B/W anaglyph:

Dryopteris aemula, TLM, Nikon Fluor 100x/1.30 - red-cyan anaglyph

Smaller surface details disappear in TLM. With episcopic lighting they show up better, but reflections make it hard to get proper contrast. I ttend to use the transmitted light path as main light source and turn in the incident light just as much as necessary to bring out the tiny surface details. Sometimes I change the balance in favour of the transmitted light near the end of the stack. The cell walls are nearly perpendicular to the image plane anyway and don't reflect much usable incident light:

Dryopteris filix-mas, TLM + ILM, Nikon M-Plan 100x/0.90

The ILM makes the spores look solid, a quality I don't like much. Sometimes tey look more like asteroids than like delicate micro-propagules...

I tried to reduce the specular reflections from the epi-illuminator by adding a polarizer, but that had no effect. By diabolic inspiration I swapped the polarizer with a green interference filter. With surprising effect:

Dryopteris filix-mas, TLM + ILM, Nikon M-Plan 100x/0.90, epi+GIF & transmitted + blue filter

The intricate surface details are best revealed without the hefty colours . The lines of the surface network are to my best measurements about 270 nanometer wide. It 's a pity the outline wouldn't come into focus. Don't know why.

Dryopteris filix-mas, TLM + ILM, Nikon M-Plan 100x/0.90, epi+GIF & transmitted + blue filter

[Adalbert]

very nice 3D

[Lou Jost]

These are fantastic. Congratulations on your journey, it must have taken quite a lot of work to get to that level. I hope some day I can do as well, but I may not have enough years left....

[rjlittlefield]

I really like image #3, the combination of transmitted and reflected light. (One niggle: I don't see any color fringes or get any stereo effect with red/cyan glasses, so I'm thinking the stereo icon at upper right is a mistake.)

The effect of the green interference filter in images #4 and #5 is so surprising that I wonder if it's more artifact than helpful. At https://www.photomacrography.net/forum/ ... 89#p126689 , user Asha points out that in some cases "an incoherent optical system can behave mostly like a coherent system." I'm wondering if the green filter used with these spores has a bandpass narrow enough that it's revealing what is essentially a speckle pattern of the illumination rather than ordinary surface structure of the subject.

--Rik

[iconoclastica]

One niggle: I don't see any color fringes or get any stereo effect with red/cyan glasses, so I'm thinking the stereo icon at upper right is a mistake.)

Apologize. So often I forget to check off that icon that I started yesterday with a script to do it for me. But not here, yet...

The effect of the green interference filter in images #4 and #5 is so surprising that I wonder if it's more artifact than helpful. At https://www.photomacrography.net/forum/ ... 89#p126689 , user Asha points out that in some cases "an incoherent optical system can behave mostly like a coherent system." I'm wondering if the green filter used with these spores has a bandpass narrow enough that it's revealing what is essentially a speckle pattern of the illumination rather than ordinary surface structure of the subject.

Having seen the laser speckle recently (https://www.photomacrography.net/forum/ ... 34#p285034), I didn't recognize it as such here. For one thing, the laser speckle remained tack sharp almost independent on the focusing. Ther pattern here had the same shallow dof as any other detail. In fact, I used it to sync the focus of the camera with the eye pieces. Also, it wasn't random in that I could locate specific highlights at the same place repeatedly.
Have a look at these SEM photos. Photo 'd' especially shows that something is going on between the coarse structure. Wouldn't that be the same?

-----------------------

I had a fresh look just now. The broad-band blue filter has a similar effect, even though it's way to big for the epi-illuminator. The Nikon L900C-filter (colour balancing?) suppresses it. Once I know where to look I can see it without the filters too. I took two new pictures, only difference is presence of the filter and ISO setting:

no filters, unedited, blue-channel only

GIF, unedited, blue-channel only

Could it be that the contrast in the colour image is too low for the PMax stacking process?

[rjlittlefield]

I could locate specific highlights at the same place repeatedly.

Do you see the same pattern of light and dark with both the blue filter and the green filter?

If the patterns are the same in both colors, then I would be inclined to trust that they directly represent structure of the subject. By "directly represent", what I mean is that we viewers can correctly interpret shape-from-shading in the same way that is more or less reliable with everyday subjects at human scale.

But if the patterns are not the same in both colors, then I am more inclined to suspect that they are formed by interference. This is a complicated interaction between illumination and subject structure, so while the patterns would be telling us something about the subject, the usual inferences of shape-from-shading don't work. In other words, if we were to look at the subject in blue, in green, and in SEM, we might see fine structure in each case but only the SEM image would be directly interpretable as physical structure.

Could it be that the contrast in the colour image is too low for the PMax stacking process?

PMax is not affected by the absolute amount of contrast, because it's always just looking for maximum contrast no matter how small that might be.

However, there can be some subtleties in play. With default settings, PMax identifies the sharpest bits by looking at local differences in image luminance, which is just a particular weighted average of the RGB values, dominated by green. This works fine in the usual case where all three channels carry similar information. But if the three channels carry conflicting information, then you might get better results by selecting Options > Preferences > PMax Settings > "Use All Color Channels In Decisions". That changes the decision strategy to look at the sum-of-squared-differences across all three channels. Still, even that option assumes that the R, G, and B images have the same focus points. There are no settings to get around that assumption inside Zerene Stacker. In principle you could work around it by separating the three channels to produce three monochrome stacks, PMax'ing each stack separately, then recombining the monochrome outputs to produce a full color image. In practice, I do not recall anybody reporting that they got significantly better results from that more complicated workflow, and I do not know why not.

In any case, I think the key question is whether blue and green show same or different patterns. I would be inclined to judge that from individual best-focused images since it seems like there is less to go wrong with those.

--Rik

[iconoclastica]

I took two pictures focused to show an area of the micro-features. Both with epi-illumination only, one with the GIF filter and one with double blue filters (cut from a rectangulare photographic filter).
Then I laid a profile along a line of about 500 pixels with rather well defined features, as judged by eye:

The locations of the peaks are almost (not entirely though) identical, the relative brightnesses vary, the GIF image has more contrast than the blue one. The latter is clearly visible too in the desaturated images. The difference image shows all dark greys: mostly L<15 with extremes up to 25%. The difference pattern is frequently concentric around the highlights.

100% crops; the difference has been exaggerated and inversed.

I'd say there is a change of contrast only, no translation.

[iconoclastica]

Running PMax with the option "Use All Color Channels" makes hardly any difference in my test case.
Running PMax on a stack that had been converted to grayscale versions of the blue channel neither is an improvement. The result is essentially the same as the full colour version, but hazier, less defined, as ofthe the blue channel is in normal pictures.

[rjlittlefield]

Thanks for the new images and careful investigation!

To me these look like very good evidence that you're seeing real physical detail and not some misleading interference effect.

But now I am curious why the green filter produces so much higher contrast.

I wonder if the objective is just that much better corrected at the GIF wavelength?

--Rik

[iconoclastica]

It's a Nikon CF M-Plan 100x DIC 0.90 (achromat), of which I have no further information excepth the catalogue's prop:

CF M Plan DIC objectives are designed for reflected
Normarski DIC system and standard polarized light These
strain free objectives help minimize the depolarization
effects. They also offer a remarkably high EF (extinction
factor). The reflected Normarski DIC system is used to
gain surface relief that is not attainable with brightfield.

[rjlittlefield]

I do not know that particular objective.

But there is a long history of using green filters to get best performance from achromats, and it strikes me that dry NA 0.90 would be prime territory to see benefits.

It could be useful to know whether using the GIF gives similar improvements in sharpness on other subjects.

--Rik