Anemia

Images made through a microscope. All subject types.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

iconoclastica
Posts: 486
Joined: Sat Jun 25, 2016 12:34 pm
Location: Wageningen, Gelderland

Anemia

Post by iconoclastica »

Reviewing my fern identifications from Central-America in the 90-ies, I had a look at their spores too. At the time, I didn't have the equipment to do so. I was struck with their variety and beauty. These are tetrahedral spores, the less common type so, with surface ornamentations of ribs that form triangular shapes. Some are spiny or have auricles. Six different types exist. Also, these spores are exceptionall large, 70-160 µm, double the size of ordinary ferns.

Although I photographed them routinely, except for experimenting with a monochrome filter, the quality of the result varied. Too much. In general, I couldn't find dMap settings that worked well. There was quite a lot of contrast from the deeper layers brought to the front, often in the form of irregular concentric rings. It was like there was not enough information in the originals to give the stacking process the foothold it needs. nearly all the images required heavy retouching, especially in the upper layers. Some of the resulting images have essentially been manually stacked. Closing the condenser aperture sometimes helped - the subject does afford to loose some resolving power.

Optical contrast sometimes was an issue too. These spores are best photographed in water. Alcohol and glycerin makes them too transparent. I found it hard to get a balance between too much water (and floating objects) and too little water (quickly drying out --> currents). Possibly a mixture would work better.
Attachments
A. hirsuta 100x proximal face
A. hirsuta 100x proximal face
A. hirsuta 100x distal face
A. hirsuta 100x distal face
A. phyllitides distal face 100x Green IF recouloured
A. phyllitides distal face 100x Green IF recouloured
A. oblongifolia distal face 100x Green IF recouloured
A. oblongifolia distal face 100x Green IF recouloured
A. guatemalensis distal 100x
A. guatemalensis distal 100x
A. speciosa 100x disal face
A. speciosa 100x disal face
A. "Pastinacaria" 100x
A. "Pastinacaria" 100x
A. flexuosa (Venezuela) distal 40x
A. flexuosa (Venezuela) distal 40x
A. "Pastinacaria" 40x
A. "Pastinacaria" 40x
--- felix filicis ---

rjlittlefield
Site Admin
Posts: 23563
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Anemia

Post by rjlittlefield »

iconoclastica wrote:
Wed Sep 28, 2022 3:06 am
In general, I couldn't find dMap settings that worked well. There was quite a lot of contrast from the deeper layers brought to the front, often in the form of irregular concentric rings. It was like there was not enough information in the originals to give the stacking process the foothold it needs. nearly all the images required heavy retouching, especially in the upper layers. Some of the resulting images have essentially been manually stacked.
Your images are lovely!

As I understand your description, these spores show a lot of internal features but you only want to render the surface.

In that situation I do not know any way to make the pictures automatically.

There is a similar problem in photographing insect eyes, trying to render surface details while omitting reflections of the environment. An example of that is discussed at https://www.photomacrography.net/forum/ ... hp?t=11271 .

As noted there, it can help a lot to use "slabbing" as a first step, to reduce the number of frames while still capturing each part of the desired surface in one or two slab outputs.

--Rik

iconoclastica
Posts: 486
Joined: Sat Jun 25, 2016 12:34 pm
Location: Wageningen, Gelderland

Re: Anemia

Post by iconoclastica »

rjlittlefield wrote:
Wed Sep 28, 2022 9:30 am
As I understand your description, these spores show a lot of internal features but you only want to render the surface.
That would explan it indeed. I learned later today that in some species the ribs are hollow while in others they are filled with a spongy medulla. I will check that once I got that paper downloaded from jStor.
If the deeper spore content is the culprit then I could try to empty them by an osmotic shock. I did so once accidentally by suddenly adding ethanol. I wasn't impressed by the unnatural appearance of the empty shells, by perhaps I'd give that another chance.

Wouldn't it be possible to edit the depthmap and rerun the stack with it?
--- felix filicis ---

rjlittlefield
Site Admin
Posts: 23563
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Anemia

Post by rjlittlefield »

iconoclastica wrote:
Wed Sep 28, 2022 1:35 pm
Wouldn't it be possible to edit the depthmap and rerun the stack with it?
I think I am not understanding the idea.

In principle one could re-render the stack given an edited depthmap.

But then how would the depthmap get edited, without doing something equivalent to manual retouching?

--Rik

iconoclastica
Posts: 486
Joined: Sat Jun 25, 2016 12:34 pm
Location: Wageningen, Gelderland

Re: Anemia

Post by iconoclastica »

This is an original depth map:
original depthmap
original depthmap
This is an edited variant:
smoothed depthmap
smoothed depthmap
This is a very simple editing result. First I applied a big range (100) gausian blur in order to get the general shape only, i.e a slightly flattened hemisphere. Since spores are generally simple forms, I suppose i would opt for a mathematical model if this wasn't just to show the idea. Having subtracted that from the original, I applied a second blur on the remainder with the ranges chosen to show the ridges but hide the noise.

When I choose the smoothed depth map as external DM, the result of the process is surprisingly similar to the original one:
the original result
the original result
the new result
the new result
The DMap-stacking process generates a new depth map, but one that is limited to the centre of the image:
the newly saved depth map
the newly saved depth map





??do depth maps of n layers need exactly n levels of grey or even specific values? Editing will of course introduce many intermediate or differing lavels of grey.
--- felix filicis ---

rjlittlefield
Site Admin
Posts: 23563
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Anemia

Post by rjlittlefield »

I am confused about what you've done. As far as I know, there is no facility in Zerene Stacker for working with an external depth map. Depth maps can be exported, but not imported.

There is an external depth map mask, specified as "Use external mask" at Options > Preferences > DMap Settings. But that only allows finer control over where the depth map is computed from imagery versus smoothly interpolated from neighboring values. Essentially the external mask image augments the contrast threshold slider, allowing areas to be forced "black in preview" and thus interpolated. Its primary use is to get a clean margin of low-contrast foreground against featureless background, a situation that is otherwise vulnerable to OOF halos intruding into the background.

So now I'm wondering: did you mistake that external mask as being an external map, or did you do something else which has yet to occur to me?

--Rik

iconoclastica
Posts: 486
Joined: Sat Jun 25, 2016 12:34 pm
Location: Wageningen, Gelderland

Re: Anemia

Post by iconoclastica »

After that explanation a certain amount of bafflement only seems reasonable :P
I indeed had misunderstood completely the function of "Use external mask" in the DMap preferences section.

Yet the idea I thought I was realising seemed too simple not to try it. I had both the edited depth map and the saved aligned images. A few lines of pyton do the rest:
DMap result based on edited depth map
DMap result based on edited depth map
It's not really an improvement, to say the least. Tried and rejected, as far as I can say ...
--- felix filicis ---

rjlittlefield
Site Admin
Posts: 23563
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Anemia

Post by rjlittlefield »

Thanks for the clarification. A lot of the confusion is my fault, due to never writing any detailed documentation about the mask. I apologize for the lapse.

But let's not give up too quickly on your idea...

Looking at your original depth map, what I think I see is that brighter corresponds to foreground, so that troublesome errors in the depth map are spots that are darker then they should be. My eyes are drawn to those spots of intense dark in the middle of lighter gray. Those represent places where the DMap algorithm detected some higher contrast in background and tried to slew the depth map so as to pass through those. Suppressing that sort of behavior is pretty much what the external mask is designed to allow. If you had a mask that was white everywhere, except for being black in spots where background was erroneously selected, then you could force the map to either stay attached to foreground or be interpolated close to the foreground surface.

I wonder, could you construct the mask automatically, or mostly so, by identifying areas where the original map is significantly darker than its smoothed version?

--Rik

iconoclastica
Posts: 486
Joined: Sat Jun 25, 2016 12:34 pm
Location: Wageningen, Gelderland

Re: Anemia

Post by iconoclastica »

rjlittlefield wrote:
Wed Oct 05, 2022 1:54 pm
I wonder, could you construct the mask automatically, or mostly so, by identifying areas where the original map is significantly darker than its smoothed version?
If this is what you mean:
depthmap.jpg
then the result is this:
2022-10-06-10.55.46_04DMap.jpg
(I remember having done something similar before, when I was using the depth maps to create 3d models)
--- felix filicis ---

rjlittlefield
Site Admin
Posts: 23563
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Anemia

Post by rjlittlefield »

Yes, like that.

I can see by carefully comparing the output images that this has eliminated some defects.

It may be that you need to expand the black blobs, or harden their edges so the mask is entirely black/white, in order to get full effect. Also I now see a few defects that are caused because the map is too far forward, so there may be opportunity for improvement by doing the same thing to areas of the map that are too bright compared to their surroundings.

All of these machinations are of course just rules of thumb, not likely to hit all and only those places where the depth map is wrong. Human intelligence could do much better by working with the original imagery, interpreted in light of a priori knowledge about the real structure of the spores. But that would be what I described earlier as "something equivalent to manual retouching".

It would be great if that process were automated as an AI task. I suspect that would be pretty simple compared with say self-driving cars, but that approach is still too hard for me to make any progress on it.

--Rik

Post Reply Previous topicNext topic