Glorioptychus callidus - a most undulant diatom

[rjlittlefield]

In the rocking stereo I can't get the image to compute in my brain...It doesn't appear as a 3-d object but as a rubbery thing moving in multiple directions at once......

Yeah, I get that too. It's a Necker cube kinda thing. If you start by assuming the deeper parts are at the front, it does the squirming malarky. The bump at 11 o-clock is at the front. I find if I look at that first, the animation settles into the correct "interpretation".

I had that same "rubbery" impression the first time I saw the gif. That persisted long enough for me to consider posting a video of writhing black soldier fly larvae and saying "it looks kinda like this".

But I think that was associated with my off-axis viewing position at the time. When I moved to on axis, it settled down to being a rock-solid 3D impression and has stayed that way. I cannot get the rubbery impression again even by moving way off axis.

However, for me the depth by rocking strongly prefers to be inverted, compared to depth by stereo. In rocking view the bump around 11:25 o-clock is consistently in back. I can move it to front by changing tabs so as to hide the rocking image, staring at the place where I know the bump will appear, and changing tabs to show the rocking image again. That impression will persist as long as I only look at front bumps. But as soon as I shift my attention to the back bumps they flip forward, and then I'm stuck inverted until I do the tab-switching thing.

I cannot flip depth by conscious effort while the image is continuously visible and rocking. However, if I pull the gif into Photoshop so that I can briefly stop the rocking while the image remains visible, then it's easy for me to flip the depth while the image is stopped, so that when rocking resumes it has the correct depth.

I have no solid theories for why any of this happens.

--Rik

[Sym P. le]

My only question is is there a right side up or up side down?

[rjlittlefield]

My only question is is there a right side up or up side down?

There is certainly a correct front-versus-back. Front should be the side that was facing the objective. I assume that's what Beatsy meant when he wrote

The bump at 11 o-clock is at the front.

An interesting thing about this particular subject is that when I zoom in close on the Larger version, I clearly see that there are two levels to the structure. On the 11 oclock bump and all other forward-facing bumps, i see a crisply rendered coarse mesh with a separate finely structured surface a short distance behind it. On the backward-facing bumps, the coarse mesh is blurred out so much that I don't get a clear impression of where it is with respect to the fine structure. I get the distinct feeling that on the backward-facing bumps, I'm seeing the coarse mesh through the finely structured layer that is acting like a sort of privacy glass. But I don't know if that matches physical reality, or maybe it's an effect that is introduced by other parts of the subject partially occluding the large-NA objective. I would love to see an SEM's view of this thing!

Here's a crop, zoomed on the large image, showing both sorts of regions:

By the way, I'm working on a new viewer that makes it easier to get this sort of view. Visit HERE, paste in the image URL https://live.staticflickr.com/65535/54017232877_d6ce57f9a0_o.jpg, and click the Load Image button. Press-and-drag to pan, mousewheel to zoom.

--Rik

[rjlittlefield]

When I moved to on axis, it settled down to being a rock-solid 3D impression and has stayed that way. I cannot get the rubbery impression again even by moving way off axis.

OMG! I accidentally clicked on the tab that shows the gif, and it went back to being rubbery! But this time I was prepared to think about it, and I see what's happening: my brain has decided that all the bumps are facing front, and if that's the case, then some of them are moving one way while others are moving another. Even more interesting, though I swear I've had no interesting chemicals and I feel fine, at this time I can freely pop it back and forth between rigid and rubbery. It's quite trippy!

--Rik

[Beatsy]

My only question is is there a right side up or up side down?

There is certainly a correct front-versus-back. Front should be the side that was facing the objective. I assume that's what Beatsy meant when he wrote

The bump at 11 o-clock is at the front.

An interesting thing about this particular subject is that when I zoom in close on the Larger version, I clearly see that there are two levels to the structure. On the 11 oclock bump and all other forward-facing bumps, i see a crisply rendered coarse mesh with a separate finely structured surface a short distance behind it. On the backward-facing bumps, the coarse mesh is blurred out so much that I don't get a clear impression of where it is with respect to the fine structure. I get the distinct feeling that on the backward-facing bumps, I'm seeing the coarse mesh through the finely structured layer that is acting like a sort of privacy glass. But I don't know if that matches physical reality, or maybe it's an effect that is introduced by other parts of the subject partially occluding the large-NA objective. I would love to see an SEM's view of this thing!

Hi Rik,

Yes, the sharper bumps protude toward the lens.

It was difficult to tell front from back on this one as the seive plate(s) with the small pores are weirdly positioned. Typically, the "coarse mesh" is on the outside (front) and backed by a thin seive plate with small holes in. In this case, it appears the x,y plane of the sieve exactly bisects the form, with the bumps sitting in front of it (closest to the lens) and the hollows behind. Confusing!

Bernd recently posted a relevant colourised SEM image over on the German Mikro forum. It's relatively low res and doesn't show fine detail, but it does show the same effect as this optical image. That is, the mesh of the forward-facing bumps is sharp while it looks "privacy screened" in the backward-facing hollows. It at least made me confident that I captured what was actually there.

Link here, you'll need to scroll down to see the SEM image. https://www.mikroskopie-forum.de/index. ... =49063.139

Cheers
Beats

[Pau]

Just an (unsure) complementary point:
To me the own geometry of the diatom and/or the image features have something perverse . Despite being sure (easy for a microscopist) of that the better resolved zones are in the frontal plane and the others with resolution somewhat killed by spherical aberration must be deeper, the apparent geometry of the less resolved bumps strongly suggests that they are protruding towards the observer. Even in the crops posted by Rik for his last stereo the misleading optical effect remains until you actually see the stereo, of course. And this also happens to me even with the Bernd SEM

Beatsy, great work again!

[Beatsy]

...better resolved zones are in the frontal plane and the others with resolution somewhat killed by spherical aberration must be deeper...

I suspect that's what it is after all. I just re-oiled the mount and checked visually. The finer pores are behind the coarse mesh in all cases - as one would expect. The (slight) loss of resolution must cause the stacking to mush up in the deeper parts. Interesting that the SEM shows the same appearance though.

Cheers

[rjlittlefield]

...better resolved zones are in the frontal plane and the others with resolution somewhat killed by spherical aberration must be deeper...

I suspect that's what it is after all. I just re-oiled the mount and checked visually. The finer pores are behind the coarse mesh in all cases - as one would expect. The (slight) loss of resolution must cause the stacking to mush up in the deeper parts.

Can you show us actual-pixels crop of what a single frame looks like, when focus is set to slice through a slightly tilted section of a back-facing bump?

Or better, corresponding crops of several successive images.

Or provide access to the original stack so I can explore it myself?

Interesting that the SEM shows the same appearance though.

Indeed, very interesting!

Perhaps Bernd could provide a full resolution image?

--Rik

[Beatsy]

Hi Rik

Pau had it right! SA is the issue, it's worse than I first thought. And I take back my implied slanderous remark about the stacking (Zerene) mushing up the deeper parts

The coarse grid is definitely not as distinct in the deeper parts, although the fine pores don't look as bad as the grid. It seems the two layers mainly interfere with each other more, if that makes sense.

Top row of the image is the middle of the 11 o-clock bumps, nearest the lens. Bottom part is a similar position on the 2 o-clock depressions, further from the lens. Focus is shallowest on the left, deepest on the right. Z-spacing between frames is roughly 2-3 microns (much more than the stacking frames).

Pixel for pixel version here https://live.staticflickr.com/65535/540 ... 4a01_o.jpg

Quite disappointing to see this really. I'll have to limit myself to 40x stacks of much thicker diatoms that have overlapping details. Sigh.

Cheers

[Pau]

I'll have to limit myself to 40x stacks of much thicker diatoms that have overlapping details. Sigh.

Don't do so, please. the whole image is glorious. The pores are also resolved in the deeper parts.
Only, we live in a physical world with physical limitations

[Lou Jost]

Hi Rik

Pau had it right! SA is the issue, it's worse than I first thought. And I take back my implied slanderous remark about the stacking (Zerene) mushing up the deeper parts

The coarse grid is definitely not as distinct in the deeper parts, although the fine pores don't look as bad as the grid. It seems the two layers mainly interfere with each other more, if that makes sense.

Top row of the image is the middle of the 11 o-clock bumps, nearest the lens. Bottom part is a similar position on the 2 o-clock depressions, further from the lens. Focus is shallowest on the left, deepest on the right. Z-spacing between frames is roughly 2-3 microns (much more than the stacking frames).

PS040000DS400509--pmn.jpg

Pixel for pixel version here https://live.staticflickr.com/65535/540 ... 4a01_o.jpg

Quite disappointing to see this really. I'll have to limit myself to 40x stacks of much thicker diatoms that have overlapping details. Sigh.

Cheers

This problem would not arise if the diatom was mounted in oil. The system would then be homogeneous. You'd lose contrast but maybe you could recover some of that by using astrophotography techniques and file formats (SharpCap live stacking and FITS files).

[Beatsy]

This problem would not arise if the diatom was mounted in oil. The system would then be homogeneous. You'd lose contrast but maybe you could recover some of that by using astrophotography techniques and file formats (SharpCap live stacking and FITS files).

Hmm, interesting idea. I'm so wedded to the "high RI mountant" thing, I hadn't really considered it. Diatoms would almost certainly get crushed with a coverslip floating on oil above them. Or did you mean only oil, with no coverslip?

Having said that, I have some DPX mountant with RI of 1.519. That's near identical to immersion oil for all intents and purposes (RI 1.5185). Perhaps that would work instead?

Diatoms have an RI of about 1.43, so there would be *very* little contrast due to that small difference. Short wavelength light (like 450nm blue or UV-A at 365nm) increases contrast though - so there might be extra leeway there.

I feel (yet) another bout of experimenting coming on...

Thanks for the suggestion.

Cheers

[hkv]

Absolutely stunning! I could not see the stereo though. I never can... Must be something wrong with my eye-brain coordination.

[rjlittlefield]

Pau had it right! SA is the issue, it's worse than I first thought. And I take back my implied slanderous remark about the stacking (Zerene) mushing up the deeper parts

No worries. I've seen Zerene Stacker create some quite puzzling effects. Some initially puzzling effects are now well understood, for example fine hairs routinely get removed when there is crisp detail in another layer behind them. But NA 1.40 is a different world, and in any case I'm certainly not going to get upset over the mere mention that it might have done something weird.

The coarse grid is definitely not as distinct in the deeper parts, although the fine pores don't look as bad as the grid. It seems the two layers mainly interfere with each other more, if that makes sense.

The part about "making sense" is where I'm having trouble.

If the structure is uniform, then I cannot understand why the fine detail of the sieve plate would retain a lot of contrast, while the coarser structure of the main grid would become almost invisible, at all focus increments shown.

Added to that, we have the SEM image which also shows a fuzzing out of the main grid in the concavities.

I still have the feeling that we don't fully understand the physical structure.

--Rik

[Lou Jost]

"Diatoms would almost certainly get crushed with a coverslip floating on oil above them. Or did you mean only oil, with no coverslip?"

I don't know...I was thinking pure oil.