First few days of micro

[MNBoldone]

After some time accumulating a couple microscope objectives, bellows, a stackshot, and tubes I thought I was pretty much ready to go. I knew I had to build a good "base", but I underestimated the investment in lighting, finding a good specimen, and specimen cleaning among other factors. I also greatly underestimated the amount of time I would spend with software manipulation (so far it has just been experimentation); I have not even really worried about touch-up yet.

I am just starting to take experimental images knowing that I have many improvements to make before I get usable images (not the least of which is a cleaner specimen).

The attached was taken with a Nikon Plan 10/.30 160/0.17 which was obtained from a forum member. I am confident it is a good objective, so I am troubleshooting starting with that assumption. My other objective is an AmScope Plan 4/0.10 160/0.17. I started my experiments with the AmScope objective, but switched to the Nikon to eliminate objective quality as a significant variable.

I am using a temporary base which is just two boards attached at a 90 degree angle. It s on a carpeted poured concrete floor (carpet is significant only in that it may be the source of the annoying fiber). The body is a Canon 6D. I accidently set it up with 180mm (instead of 160mm) between the end of the objective and the focus plane for this photo series. I am not sure what impact that has on the final image, but I plan to change the length to 160mm. Between the camera and the objective are a couple adaptors, two extension tubes, and a Pentax Autobellows (all are unflocked).

I started out shooting with constant light (2 LED panels). That was my initial plan. I was getting several hot pixels and I also desired to increase image sharpness so I did switch to a flash. The attached photo was taken with a single off camera flash using a very inexpensive soft box and two diffused light panels. I recognize that the light panels may no longer be contributing significant light. I also recognize that I need to increase diffusion of the flash and I should probably work in a second flash.

Exposure and flash were on manual. Exposure time was ~1/200 of a second. I used mirror lock-up.

I experimented with step size until I decreased my steps to 0.008 mm. The attached used 281 slices. I was quite surprised how many steps seemed to be required.

I experimented with Zerene Stacker and Helicon Focus. I tried various combinations and slabs. I really struggled with Halos when using Zerene's DMAP. The attached is one of the better results, it was generated by using the Helicon Focus Method B on the entire stack with a radius of 3 and a smoothing of 2 - although to be fair I am not sure I saved any Zerene stacks of the 281 images for comparison. Most of my Zerene efforts involved experimenting with slabs. I did not attempt touchup.

Oh, the subject is a lichen in the genus Xanthomendoza (Xanthoria). Five years ago I would have likely known the species, I have some relearning to do there as well.

Also, I am someone who is most familiar with photographing in natural light and I do not edit my images. I am a dinosaur from the film age. Flash and image editing software are all a learning process for me.

I am having trouble achieving what I would consider acceptable sharpness in the composite (especially near the background and just below that fiber). Maybe my results as good as I can achieve with the original images.

[NikonUser]

Your image does seem to be less sharp than I would expect from that lens.

I just happened to be photographing a bug head with the same objective, Nikon D7200, 98 frames @ 0.006 mm. 2 diffused flashes, Zerene PMax. Original frames were RAW and processed in Adobe RAW and saved as .tif
Small image is the original stack, large image is a crop of the part I am interested in - no processing after the Zerene stack apart from rotation and cropping.

No idea why your image lacks sharpness

[rjlittlefield]

Welcome to the world of high mag stacking!

For background and disclosure, I'm the fellow who wrote Zerene Stacker. I also answer all of its email requests for support. So, my comments here will be from the background of having helped hundreds of other people in that role, in addition to my postings here at photomacrography.net.

I started my experiments with the AmScope [4X/0.10] objective, but switched to the Nikon [10X/0.30] to eliminate objective quality as a significant variable.

At image center, the sharpness of those two objectives on sensor will be very similar. But because of its smaller NA, the 4X objective will have almost 10 times more DOF and somewhat less tendency to produce artifacts in stacking. Rather than jump into the deep end of the pool with stacks containing hundreds of images, you might find it more efficient to back off to your 4X, get comfortable with working there on smaller stacks, then move up to 10X.

I accidently set it up with 180mm (instead of 160mm) between the end of the objective and the focus plane for this photo series. I am not sure what impact that has on the final image, but I plan to change the length to 160mm. Between the camera and the objective are a couple adaptors, two extension tubes, and a Pentax Autobellows (all are unflocked).

The lack of flocking might produce a little veiling glare. The extra 20mm of extension will give increased magnification. That will slightly soften the image by spreading detail that is already limited by diffraction across more pixels. It will also reduce whatever aberrations you're getting in the corners, by moving those out of frame. Neither effect will be very obvious.

Exposure and flash were on manual. Exposure time was ~1/200 of a second. I used mirror lock-up.

Mirror lockup and flash exposure, excellent. That combination pretty much guarantees that motion blur will not be a problem.

I experimented with step size until I decreased my steps to 0.008 mm. The attached used 281 slices. I was quite surprised how many steps seemed to be required.

DOF shrinks in proportion to NA squared, so higher mag objectives always have shallow DOF. The nominal DOF at NA 0.30, calculated by standard wave optics formulas, is about 0.006 mm. With some subjects you can use a little larger step size; other subjects require an even smaller step size for best details. Your 0.008 mm is probably around optimal for this lichen, in terms of getting a result that will look good while pixel peeping, in the minimum number of frames.

I really struggled with Halos when using Zerene's DMAP.

This is a common problem for new users, who often have trouble figuring out how to set the contrast threshold slider. In most cases, the best way to set the slider is to recognize halos in the first-pass result, then adjust the slider so as to cover those halos with the black mask. Then in the final result the halos will be gone. There's a tutorial, "How To Use DMap", linked on the main tutorials index page at https://zerenesystems.com/cms/stacker/d ... rialsindex . It doesn't have an example that is exactly like your lichen, but with luck the fake flowers will be close enough to get you going.

By the way, one very common mistake made by new users is to push the contrast threshold slider far to the left, so that the preview image has little or no black in it. That is a recipe for maximizing halos. Pushing the slider far left works OK when a subject has simple 3D structure that is covered with fine detail, like the face of a coin. But with complex structure like this lichen, it's important to learn about using the slider as intended.

Most of my Zerene efforts involved experimenting with slabs. I did not attempt touchup.

Slabbing is definitely a useful technique with deep stacks, like your 281 frames. One of its advantages is that you can generate the slabs using PMax, which has no important controls to set, then when you combine the slabs with DMap, the DMap step goes relatively quickly because it has relatively few images to work with. This makes it a lot faster and easier to experiment with different settings of the contrast slider and so on. A second advantage of slabbing is that usually retouching can be done entirely from the intermediate slabs, with no need to go clear back to original source. Again, this makes the retouching process go faster and easier.

All that said, I think of slabbing as being an advanced technique. I would never advise somebody to learn slabbing until they have the basics of stacking and retouching well in control.

I am having trouble achieving what I would consider acceptable sharpness in the composite (especially near the background and just below that fiber).

I expect that your troubles are due to a combination of several different factors.

One factor, and usually the most important because it's unavoidable, is diffraction blurring. The effective aperture of a microscope objective can be calculated as Feff = magnification/(2*NA), so at 4X and NA 0.10 you're running at effective f/20, while 10X NA 0.30 is f/16.7 . Both of those are well into diffraction territory with any modern DSLR.

As a result of the diffraction blurring, images shot through microscope objectives both allow and require strong sharpening in post-processing. There is some discussion of this at viewtopic.php?p=208762#p208762 and in the replies by me later in the same thread. Note that my "Photoshop USM, 200% at 0.7 pixels" was for operating on an image that had already been somewhat resized for web presentation. When working on images at full camera resolution, I often filter with settings like USM 100% at 1.5 pixels. Such strong sharpening would convert most "ordinary" images to a trashy mess of artifacts. But when applied to say an f/20 image that has been heavily softened by diffraction, strong sharpening can reveal detail that would previously have gone unnoticed, while also not introducing over-sharpening artifacts.

Another factor is the nature of the subject. Many of the lichens that I'm familiar with just do not have much detail of the proper size to appear sharp at 10X. The overall structure is too big, the cellular structures are too small, and there's not much in the middle. Often the sharpest details in my lichen photos consist of debris on the surface, like grains of mineral dust.

So, if you're interested in developing photographic technique, I'd suggest checking around your windows and light fixtures for some dead insects. Those things almost always have lots of well defined detail at many size scales. NikonUser's fly is a good example; the forum galleries contain hundreds of others.

I hope this helps.

--Rik

[MNBoldone]

Thank you both very much.

Nikonuser, thank you for the comments, the comparison images, and providing the step interval you were using with the same objective. I am also shooting RAW and converting to TIF in Canon software. While comparing your images to mine it quickly occurred to me that I likely had subject movement - not vibration during the taking of the photos, but a gradual drifting of the subject across the stage due to vibration. My subject is glued to a very thin piece of wood which is sitting on a wooden "stage". To reduce subject movement, today I glued the subject to my stage. I think that helped significantly.

Rik, I read much of your material on this forum and in the Zerene documentation on and off over the past year - especially this week as I started shooting. I have found it very valuable. Thank you for your detailed reply to my post. My trouble shooting started with a post of yours on troubleshooting soft images and I relied heavily on one of your posts and information on the Zerene website when I determined a starting step size. I am amazed by your grasp of the technical aspects of photography. I also really like your Zerene stackshot interface.

I will likely go back to experimenting with the 4x and I may wait a bit before I do too much with slabs. I did like the balance I was seeing creating slabs with PMAX and stacking those with DMAP, so I think I will find myself using that approach sometime in the future. I think that when I slid the DMAP slider far to the right was also when I had the most extreme loss of detail in the smoother parts of the lichen (I could be remembering that wrong); I think that is why i wasn't aggressive enough to remove the halos.

I have several preserved spiders in vials which didn't make it to specialists this year due to COVID. I may have to play with them (or insects) for a while as you suggest. Thank you for that suggestion as well. I thought my subject was probably not the best for a first attempt. I will certainly return to lichens though. High magnification photography of lichens was the principle reason I pursued this style of photography. I really appreciate knowing that they may not be as sharp as the insect images I see, just adjusting my expectations may save me considerable frustration.

I retried photographing that lichen. This time I glued the subject to my "stage" to eliminate movement, I added a bit more diffusion for the flash, and I changed my step to 6 microns. I used Zerene to generate PMAX and DMAP and I tried doing some "quick and dirty" retouching to see what a combined result might look like. I also took some information (primarily a void) from an input image. It looks different, but I am not sure it is an improvement They may not be tack sharp, but I think I should eventually be able to generate lichen images which could be useful as identification guides.

Thanks again to you both.


Brad

[NikonUser]

Zerene does an incredible job to correct for this. I have had specimens in water that moved considerably between frames.

By checking "Align each frame against first frame separately" in Preferences, all frames line up.

[rjlittlefield]

Brad,

Thank you for the further information.

The possibility of softening due to subject movement had not occurred to me. In normal circumstances, all the stacking programs will automatically correct for simple shifts across the frame. However, if the subject does not sit in a stable position, or if it is sliding on a slightly rough surface, then it might rock a little bit from one frame to another. If that happens you could end up with quite a variety of artifacts.

If subject motion is considered to be a possible problem, then in Zerene Stacker a good way to check for that is to use the process described at https://zerenesystems.com/.../faqlist#how_can_i_detect_movement_in_my_stack . Very briefly, you process the stack once, then place a checkmark on "Show as aligned", then scroll through the input images as if you were viewing a filmstrip. For a properly shot stack, when "Show as aligned" is checked you will see nothing changing from frame to frame except for focus. If you do see the subject moving around even with "Show as aligned", then there was some sort of movement that could not be compensated for.

I think that when I slid the DMAP slider far to the right was also when I had the most extreme loss of detail in the smoother parts of the lichen

That makes sense. When the slider is too far left you get the worst halos; when it is too far right, you lose detail. Often there is a "Goldilocks" position somewhere in the middle, that makes the result image come out clean. But sometimes there is no single position that will work that way across the entire frame. In that case a good attack is often to run DMap two or more times, using different slider positions that are optimized for different areas of the images, then use retouching to combine the best parts of the multiple DMap outputs. This method is illustrated in the "Advanced Retouching" tutorial.

I retried photographing that lichen.

At the scale shown here, the image looks completely reasonable to me. If you post an "actual pixels" crop of some area that you care about, we could tell better about the sharpness.

--Rik

[rjlittlefield]

By checking "Align each frame against first frame separately" in Preferences, all frames line up.

To clarify...

For normal focus stacking, you'll get equal or better results by leaving that option at its default setting, which is not checked. The option is provided for high precision measurement tasks, where all images have the same focus and the goal is to measure lateral shifts of the image.

With the option not checked, the alignment process proceeds as image 1 against image 2, then image 2 against image 3, 3 against 4, 4 against 5, and so on, comparing only images that are adjacent in the stack, and calculating total alignment for each image by accumulating all the frame-to-frame alignments. Because this method compares only images that have similar focus, it is not vulnerable to large errors in deep stacks. However, it is vulnerable to accumulating small errors.

For high precision measurement tasks, focus is not an issue, and in that case accumulation of errors can be avoided by selection the option to "Align each frame against first frame separately", so image 1 against 2, then 1 against 3, 1 against 4, and soon on. As it is explained in the tooltip, "If checked, align each frame against first frame separately, instead of against previous frame. This will prevent accumulating errors when using alignment to optically measure small lateral movements."

--Rik

[MNBoldone]

Thank you for the offer to look at an "Actual pixel" crop. You are being very generous with your time.

The attached are unretouched DMAP with the Estimation Radius set to 10 and the smoothing set to 5, the step was 0.006mm. I think the DMAP slider Contrast Threshold used was around 40 (57 if the software remembers the last value used). This was a re-stackng of the same images used yesterday, but using lower numbers for all settings (Estimation Radius, smoothing, and Contrast). I think this image is an improvement over yesterdays DMAP effort.

One crop is what to my eye appears to be one of the sharper portions of the image and the other is among the softer areas. I think I need to start my trouble shooting by looking at the best areas. I assume they would indicate if my original images are of sufficient quality.

One other thing, I believe that my camera is set to "Faithful" (color saturation) and I do not have my camera do any sharpening - which is probably atypical.

Brad

[rjlittlefield]

The attached are unretouched DMAP

Probably you've noticed the bold "swirly" halos in the first image. Those are typical of a 3D structure that has detailed foreground obscuring detailed background, with a sharp edge or even an overhang separating the two regions in depth. Different stacking methods produce different artifacts in this situation, but there is no method that will be completely free of artifact because the camera never saw what we would like to appear in the final image. Focused background that appears near the foreground structure will always be seen either out of focus, or contaminated by foreground, or both. The problem is diagrammed at viewtopic.php?p=135042#p135042 and is discussed in some detail earlier in that thread.

In Zerene Stacker, those areas are usually handled best by retouching from a PMax output into a DMap master. PMax outputs are always noisy compared to DMap, so you should be prepared for some extra "grain" in the retouched regions. Shooting your stacks at base ISO will minimize this issue.

The soft areas in the second crop strike me as probably being loss-of-detail halos caused by a slider setting that was too high. You might be best off to set the slider to a low value and handle the resulting "swirly" halos by retouching from PMax. Flashing between PMax and DMap by using the "s" key in retouch mode is a powerful way of seeing where DMap has given halos rather than useful detail.

I think I need to start my trouble shooting by looking at the best areas. I assume they would indicate if my original images are of sufficient quality.

Yes, that's the right approach. Safe places to check are in areas that are far away from any possible contamination by out-of-focus structures that are farther forward.

Your images look fine to me. I loaded the actual pixel crops into Photoshop and did a little exploration of strong sharpening to compensate for diffraction blurring. I'm pretty happy with 70% at 1.4 pixels. But this sort of sharpening is still more art than science so your preferences might be different. Play with the concept and see what you like.

--Rik

[MNBoldone]

Thanks again. I think I have considerable experimenting to do. I really appreciate the comment about the slider probably being too high, but the base images probably being fine. It really helps me know where to exert my efforts.

[rjlittlefield]

Thanks again. I think I have considerable experimenting to do.

You're very welcome. Keep us informed, please.

...very generous with your time.

I look on exchanges like this as being a win-win learning experience. One of the problems I have is that it's no longer possible for me to see focus stacking through the eyes of a beginner, unless I get to use someone else's eyes. The ones I have know too much. Thank you for your patience in struggling through the maze of difficulties in getting good results at high mag.

--Rik

[MNBoldone]

I have continued to play around with software settings (primarily on stacking software, not so much on sharpening). I have photographed a few additional subjects, but have not yet made it to any invertebrates.

One of my biggest challenges was getting the shutter to trigger consistently when using Zerene Stacker with Stackshot. Somewhere along the line I must have changed my workflow so the Stackshot could not trigger the shutter. I think the conflict was the live view through the Canon utility software. The last couple days stackshot has always been able to trip the shutter when I have turned off my live view through Canon Utilities.

Anyhow, even though I am still learning, I have been shooting most of my subjects at 4x, then shooting a portion of the same subject with the 10x objective - just to start to test a workflow I suppose.

I thought that an image taken with the 10x objective of what I believe is an Arthonia fungus was insightful. The structure of this subject is considerably different than the earlier Xanthomendoza lichen. The attached are all from a stack of 115 images utilizing DMAP with an estimation radius of 6 and a smoothing radius of 3 and a step of 7 microns. For the image below, a Contrast Threshold of 5.8 was used. I tried several higher number for all these settings, but for this complex subject the lower settings seemed to yield the best image (to my untrained eye).

One other thing which is apparent to me in these images is the fall-off in image quality away from the center of the view. The three images below are: a reduced image of the entire view, a 100x crop showing the reasonably sharp center of the image, and a much softer 100x crop taken from near the left edge of the image. The softness of the outer portion of the image seems to be due to optical limitations. All the images are unsharpened.

Anyhow, just sharing some observations as I continue to experiment.

If anyone has tips for cleaning lichens and lichen-like fungi, please feel free to share. I did not try the ultrasonic bath to clean this specimen.




Brad