Chromatic aberrations filtering

[bkt]

Hello,

I use achromatic objectives, and I'm often disappointed by pictures I get from them, especially at 40x and 60x.

How do you deal with them, and without switching to costly fluor or apochromatic objectives?

I found some software, by image processing experts: https://github.com/teboli/chromatic_abe ... _filtering

I managed to tweak this a bit, but I'm still often in troubles due to the amount of wide spread aberrations on some images.

Following is a rush, plus a result after individual frames processing using this software. Takes some time using an I5... but processing time isn't a major issue.

Any hints or alternate tricks would be welcome.





This was also filtered:





The result on still images, so far, I didn't manage to get much better; not sure this software was designed for such heavy aberrations:

[Topcode]

Well, if you experience lateral chromatic aberration, I have found that using RegiStax can help somewhat, but it seems like you mostly have issues with axial chromatic aberration.

[apochronaut]

Unfortunately, you cannot make a silk purse out of a sow's ear or an apple pie out of shit from an apple eating pig. Chromatic aberration is the result of refraction which is the most obvious to us of the several aberrations and distortions that occur when light moves across a refractive border. Filtering out wavelengths can help but removal of chroma removes detail and ultimately it is the quality of the design in glass that is the arbiter of image quality.
Some modern achromat objective designs have been corrected at more than just green, so going in that direction while staying with less expensive used achromats is one option. Another would be to go with older apos or fluorites.

In general your pictures aren't too bad.

[bkt]

Yes, I noticed RegiStax, designed for astronomy, to compensate waves due to atmospheric changing conditions. I didn't try this, as I'm not convinced it could help on such bad images, green/yellow and more fringes being often very present, depending on contrast and magnification. Seems also inappropriate for videos of moving organisms, were you keep every frames...

And I'm still not sure my CCD plane is correctly positioned on the intermediate image. As explained here, I'll try a variable length tube to investigate this, because it could add blurriness plus spread aberrations: viewtopic.php?p=142484#p142484

I tested also different methods to filter or attenuate colors, with yellow and amber filters, wasn't satisfying or magic either.

Today, I found a spec sheet of apo objectives, corrected for 5 colors. Acro being corrected for one or two colors only. But prices/budget aren't the same, I found no 60x apo ref, and a 50x ref would need oil. This is why I went to search for software fixes first.

This is a picture I took of a calibration slide using my 40x, for some worst contrast conditions. It shows blue/yellow/green/purple fringes I'd like to eliminate, which is almost achievable with the software I mentioned in my initial post, but almost only:

CalibSlide.JPG (67.05 KiB) Viewed 1928 times

This is the difference I get using this software. I may have to tweak/tune it more. Purple/green (one part of the spectrum) mostly gone, blue/yellow still is there:

CalibSlide.processed.jpg (59.71 KiB) Viewed 1927 times

[Topcode]

Well for videos, I have used SerPlayer, but that again, really only for lateral CA. Which it does look like you have some(in addition to the axial/focal CA), but it doesnt scale, only moves, so, its not that useful unless you have a specific target in the frame. But it was used for some of my stuff, for example heres a frame showing a bacteria, before and after CA correction

no correction

ser player correction

[bkt]

SerPlayer... I see not not much info about it using Google. Brought me to SharpCap, a solution embedding it. I'm starting to watch this video. Seems they exclude part of the color spectrum, to rebuild colorful pictures without aberrations, with a "synthetic blue channel", "preventing defocused blue light": https://www.youtube.com/watch?v=Hc7Tjcq5WIg

At some point, I was thinking doing this with a more or less deep amber filter. But I'd prefer to keep the spectrum, cleaning up only artifacts...

[apochronaut]

What microscope is this and how are the pictures being taken? Cell phone through an eyepiece?

[bkt]

What microscope is this and how are the pictures being taken? Cell phone through an eyepiece?

My pictures? A BA310e trino with stock Motic EC lenses. I upgraded it with an achromatic Kohler condenser.

Camera is a BRESSER MikroCamII 9MP 4K 1'', for its global shutter. Mostly running at 30 (4k) to 60 (2k) fps.

Photo tube adapter is a Motic 1x, without any optics inside.

This is a stock image, from the Motic blog, I used it for software test purpose:

Amoeba 100x EC plan_1.jpg (99.18 KiB) Viewed 1874 times

This is the same picture, processed, for less chroma aberations, an attempt, and for comparisons I had made:

Amoeba 100x EC plan_1.processed.jpg (105.95 KiB) Viewed 1874 times

[apochronaut]

Those micrometer images . Are they a center crop or if not can you provide a dead center micrometer image?

[bkt]

Those micrometer images . Are they a center crop or if not can you provide a dead center micrometer image?

The micrometer image was a crop. The 4k png file is too large to be stored here, downsized to 2k, seems it gets reduced here...

Maybe, get a bigger image via X: https://pbs.twimg.com/media/GH3NLmVX0AA ... =4096x4096

40xMotic0002.jpg (65.2 KiB) Viewed 1874 times

[Topcode]

SerPlayer... I see not not much info about it using Google. Brought me to SharpCap, a solution embedding it. I'm starting to watch this video. Seems they exclude part of the color spectrum, to rebuild colorful pictures without aberrations, with a "synthetic blue channel", "preventing defocused blue light": https://www.youtube.com/watch?v=Hc7Tjcq5WIg

At some point, I was thinking doing this with a more or less deep amber filter. But I'd prefer to keep the spectrum, cleaning up only artifacts...

This is the ser player im talking about https://github.com/cgarry/ser-player
synthetic blue channels are another method I have used in the past, but was I was talking about just shifts around the color channels to reduce some forms of chromatic aberration

[bkt]

but was I was talking about just shifts around the color channels to reduce some forms of chromatic aberration

Their shift around solution seems clear to me. They drop all the blue channel, and rebuild an alternate using the red and green channels.

In my setup, I feel I get two spectrum of aberrations, one only being being handled by the software I found. One in the blue area, one in the green area. This being very present depending of the slides, magnifications and contrast. And this is what I'm trying to fix, instead of buying apos, a budget... Not sure this is achievable, at least, not by me

This is a video of Motic Europe, representative of what I'm trying to correct by software, depending of slides/conditions, blueish/green explodes:

[apochronaut]

Since you are posting images from a camera mounted on a trinocular, is the same level and spread of ca, evident through the eyepieces?

[bkt]

Since you are posting images from a camera mounted on a trinocular, is the same level and spread of ca, evident through the eyepieces?

Our eyes and brains are magic. They are able to accommodate and to compensate a lot, lack of depth of field, and high dynamics also...

In the eyepieces, the aberrations look similar. The CCD is only accurately catching images like they are provided by the tube lens.

[apochronaut]

Then there is something optically wrong in your system. The objectives, which are planachromats will have some residual ca yes but it should be symmetrical. Achromats are naturally undercorrected, so should show slight yellow banding to the outside of objects and slight blue banding on the inside edge. Your image captures are assymetrical. There is excessive blue and yellow banding along the north-south axis of the micrometer running roughly from 6:30 to 12:30 and excessive green and purple colour banding along the other axis and the blue/yellow banding is reversed. An undercorrected achromat should have blue on the inside borders and yellow on the outside borders. With an achromat as well, the on axis section should be just about ca free with slightly increasing ca as you go out towards the periphery. The ca pattern is one of a badly overcorrected apochromat, although it would as well be symmetrical and capable of being compensated in a compensating eyepiece.
That's why I asked about the eyepiece presentation. If you are getting the same skewed ca through the eyepieces, and it is universal across the objectives, then I would be looking into a mis-aligned or delaminated telan lens before I get into software.

[bkt]

This is the version I used and tweaked a bit, a version of 2023. A PDF and some code is downloadable from here: http://www.ipol.im/pub/art/2023/443/

According to what I understood, Ser Player or SharpCap do drop the blue channel and rebuild it from red and green channels.

That other software/solution does things differently, and this is maybe my trouble (apos being corrected for 5 colors, achros for one two only). This solution works, but partially, my assumption being that it doesn't deal with all the spectrums in which I get aberrations:

Abstract

This article presents an implementation of the chromatic aberration correction technique of Chang et al. [Correction of Axial and Lateral Chromatic Aberration with False Color Filtering, IEEE Transactions on Image Processing, 2013]. This method decomposes aberration correction into a cascade of two 1D filters. The first one locally sharpens the red and blue edges such that they have similar profiles to that of the green channel serving as guiding image throughout restoration. The second one shifts the red and blue corrected edges to the location of the green ones to remove the color fringes. These two successive estimates are ultimately merged into a final prediction, free of most chromatic aberrations.

[bkt]

Then there is something optically wrong in your system. The objectives, which are planachromats will have some residual ca yes but it should be symmetrical. Achromats are naturally undercorrected, so should show slight yellow banding to the outside of objects and slight blue banding on the inside edge. Your image captures are assymetrical. There is excessive blue and yellow banding along the north-south axis of the micrometer running roughly from 6:30 to 12:30 and excessive green and purple colour banding along the other axis and the blue/yellow banding is reversed. An undercorrected achromat should have blue on the inside borders and yellow on the outside borders. With an achromat as well, the on axis section should be just about ca free with slightly increasing ca as you go out towards the periphery. The ca pattern is one of a badly overcorrected apochromat, although it would as well be symmetrical and capable of being compensated in a compensating eyepiece.
That's why I asked about the eyepiece presentation. If you are getting the same skewed ca through the eyepieces, and it is universal across the objectives, then I would be looking into a mis-aligned or delaminated telan lens before I get into software.

Thanks, I'll think about all your answer during the next days. And before ordering any apo objective to check how this could improve an image... And I'll compare again images in eyepieces to those of the camera port.

I know some microscopes need a compensating piece in the photo port. This shouldn't be needed, as the stock 1x adapter comes empty?

"mis-aligned or delaminated telan lens", I guess you mean the tube lens. But several other images/videos I saw from the same model show the same artifacts. I suspect those acro objectives were optimized for contrast, not for CA.

Your image captures are assymetrical. There is excessive blue and yellow banding along the north-south axis of the micrometer running roughly from 6:30 to 12:30 and excessive green and purple colour banding along the other axis and the blue/yellow banding is reversed.

I'll also check if I could get something different by tuning slightly the photo port length with the variable adapter I found.

[Topcode]

but was I was talking about just shifts around the color channels to reduce some forms of chromatic aberration

Their shift around solution seems clear to me. They drop all the blue channel, and rebuild an alternate using the red and green channels.

Yes, that's what your video describes. What I am describing, is changing the alignment of the color channels, shifting them around, in order to correct for certain types of CA, in a way that does not drop any channels. Heres what your calibration slide looks like after changing the alignment via the method used in registax and ser player.

versus how it looks like corrected through the method the software you found here https://github.com/teboli/chromatic_abe ... _filtering

versus how it looks like corrected with a synthetic blue channel

The differences should be pretty clear in how they work

[apochronaut]

"mis-aligned or delaminated telan lens", I guess you mean the tube lens. But several other images/videos I saw from the same model show the same artifacts. I suspect those acro objectives were optimized for contrast, not for ca."




No. Then those videos are showing images from a defective microscope unless there are elements in the sample that are highly refractive. You seem to be missing the point about the lack of symmetry in the images you posted though and they are coming from a flat uniform sample with no specific unusual gradients across the field. You cannot have reversed yellow and blue ca banding along the y axis and green and purple ca banding along the x axis of a micrometer scale and have everything alright and explainable or that software will fix it. Microscopes don't need software to image well, they just need excellent engineering and precision manufacturing.
All aberrations or distortions should be absolutely symmetrical. You aren't using anamorphic lenses.
It has nothing to do with contrast enhanced optics as opposed to ca corrected. Assymetry is assymetry.

[MichaelG.]

Apochronaut’s comments are wise, and well-expressed.

To the best of my knowledge, software fixes can only work by discarding information and/or making guesses.

The one relatively safe approach is when pure lateral CA is known to be present, and the channels can then be individually re-sized to overlay correctly.

But, of course, any chromatic asymmetry in the original image will defeat this ^^^

MichaelG.

[Hobbyst46]

In my experience, such unsymmetrical CA might originate from a misaligned condenser.

[apochronaut]

In my experience, such unsymmetrical CA might originate from a misaligned condenser.

This is true but he has an achromatic condenser and referenced köhler. Usually, coming from the condenser one would expect a subtle enhanced spread of ca unevenly into a region of the field, kind of like those classic radar sweeps but static.
What I am seeing here is a discreet quite uniform colour shift from blue and yellow borders on the micrometer bars of one lateral axis to green and magenta borders on the perpendicular axis. The green/magenta is loaded towards about 4:00 and the blue/yellow to about 1:00 ; as though an optic or element is tipped from the perpendicular.
Normally as well ca caused by design chromatism, the kind you would encounter for instance using an achromat with an imperfectly correcting Ramsden or Huyghens eyepiece is radial. The blue banding is always along the inside border, so shifts from the left side of objects on the right side of the field to the right side of objects on the left side of the field with the field stop displaying varying degrees of blue to violet banding.
This pattern is reversed when an apochromat is used with a poorly compensating eyepiece. The right side has yellow banding on the inside or left side of objects and the left side has yellow on the right side of objects with the field stop displaying varying degrees of yellow to orange banding.

In the unaltered image that was posted, the banding colour pattern is uniform across the crop and does not invert at the longitudinal axis, displaying a more or less typical achromat colour contrast along one axis and more what I have seen in advanced( more than green corrected ) achromats or fluorite objectives along the perpendicular axis. That's why I requested an image of the center of the field but apparently it is part of the crop. You can't really tell however because there is no inversion of the chroma once you cross the longitudinal axis.
It seems a bit too severe to be the result of an off center condenser. Perhaps one that is severely tipped but that would show also as having defocused field diaphragm leaves on one hemisphere. Köhler was referenced as being of importance but a condenser tipped that severely is hardly Köhler. I'm not sure a severely mis-aligned condenser can alter the wavelengths of chromatic aberration. This seems like something refracting in the optical pathway, as well as the lack of hemispheric inversion of the ca.

[bkt]

You have written a lot, I'll now have for some days to study and understand what you added

You seem to be missing the point about the lack of symmetry in the images you posted though and they are coming from a flat uniform sample with no specific unusual gradients across the field. You cannot have reversed yellow and blue ca banding along the y axis and green and purple ca banding along the x axis of a micrometer scale and have everything alright and explainable or that software will fix it. Microscopes don't need software to image well, they just need excellent engineering and precision manufacturing.
All aberrations or distortions should be absolutely symmetrical. You aren't using anamorphic lenses.
It has nothing to do with contrast enhanced optics as opposed to ca corrected. Assymetry is assymetry.

In my experience, such unsymmetrical CA might originate from a misaligned condenser.

This is true but he has an achromatic condenser and referenced köhler. Usually, coming from the condenser one would expect a subtle enhanced spread of ca unevenly into a region of the field, kind of like those classic radar sweeps but static.

This is what happens when I adjust my achro Kohler condenser, which is a used part, and without a sample. According to Olympus, I shall set it for blue at the outside, and red to the center... something I cannot achieve, it remains unequal, and I'm using it that way. Which shall explain the asymmetry in my pictures (not of all pictures).

I checked, it looks the same in the eyepieces.

https://www.olympus-lifescience.com/en/ ... aphy/abbe/ and even if they are properly adjusted using Köhler illumination, an undesirable color effect will be present in photomicrographs made using these condensers

I'm not sure everything is precisely aligned in there. Neither if I'll be able to tweak this hardware.... I quickly moved parts (condenser and trino head) around without noticeable differences, will try again later. But if I rotate the trino head,the image/artifacts rotate also; the same happens for 3 different objectives (one is an alternate brand); so it seems to be an illumination issue.

UnequalIllumination.jpg (33.58 KiB) Viewed 1679 times

Using the other condenser, illumination seems equal. So I know what I'll have to do, I need to fix the used condenser (which is in an almost new condition)...

OtherCondenser.jpg (35.78 KiB) Viewed 1679 times

What I am describing, is changing the alignment of the color channels, shifting them around, in order to correct for certain types of CA, in a way that does not drop any channels. Heres what your calibration slide looks like after changing the alignment via the method used in registax and ser player.

Ok, I misunderstood you, I'll check again what you describe and Ser Player features. I'll download and test this also. Because I'm still interested in what softwares could achieve. With both condensers, I get a lot of aberrations depending of the slide/magnification.

versus how it looks like corrected with a synthetic blue channel

The differences should be pretty clear in how they work

In what you now show, blue and some yellow is gone. I see remaining green and magenta. Processing your picture with the above software could be a solution; I'll have to check and see with some samples, using the two software solutions.

It seems now blurry (in a crop), this is what I'd like to check using a variable length photo port tube:

MGKgUV0.processed.jpg (63.45 KiB) Viewed 1679 times

[apochronaut]

Then test the imaging with the abbe condenser, if it produces a reasonable Köhler adjustment and post some photos.

[bkt]

To the best of my knowledge, software fixes can only work by discarding information and/or making guesses.

I do agree with this. Part of the images being so rebuild, build from neighbor pixels, or alternate colors. By the way, other parts of images get also processed, producing color alterations. But I'm not willing to swap my achro objectives, and I'm trying to get out more from them, software being an option.

Then test the imaging with the abbe condenser, if it produces a reasonable Köhler adjustment and post some photos.

I get high level of aberrations using either condenser... especially at 40 and 60x. I started with the abbe only, had hit the same issue.

I checked the files timestamps. That is a stack which was made using the abbe. At that period, I started to search for any software fix.

[apochronaut]

There should be no aberrations at the longitudinal axis and any ca should be symmetrically radial with hemispheric inversion of the colour sequence. That is normal.
Remove the condenser and post a cropped photo of your micrometer with the highest magnification objective that a condenserless illumination beam can fill on your stand. Probably 10X.

[bkt]

There should be no aberrations at the longitudinal axis and any ca should be symmetrically radial with hemispheric inversion of the colour sequence. That is normal.
Remove the condenser and post a cropped photo of your micrometer with the highest magnification objective that a condenserless illumination beam can fill on your stand. Probably 10X.

That is 10x, 40x and 60x... auto exposure. 4k files. Without a condenser. Slide not being that clean.

[bkt]

Adding the 4x. Aberrations are there, but not so annoying to me.

[bkt]

That is a crop from the 40x. This is typically what I want to fix also by software.

CropIn40x.JPG (73.69 KiB) Viewed 1629 times

[apochronaut]

I expanded the 10X 40X and 60X to the maximum on my screen and your chromatic aberrations are all over the place and as before completely unsymmetrical as well. . Firstly, the 10X shows an extremely yellow ca in the center but as you transit to the periphery the blue band is internal of the graduations on one side and external on the other side of center. Same with the perpendicular axis. Those aberrations should be consistently blue inside and yellow outside.
With the 40X, the vertical scale has green on the inside and pale blue on the outside and south of center it is yellow on the inside and green on the outside Going laterally, it is green on the inside and yellow on the outside changing to red at the periphery.
The 60X has yellow inside and blue outside on one vertical flank, changing to blue inside and yellow outside in the other hemisphere. At least the east west graduations are somewhat consistent across hemispheres but they are green inside and red to orange outside depending on the side.The whole is overlain with a kind of blue haze.

It might not be that unusual for different objectives in a family to have unique chroma characteristics but usually and especially since the post war glass revolution they should be pretty consistently corrected. It used to be that manufacturers would build a system around a certain objective, often the 40X and then get the others as close as possible. Ideally then, each objective would need a different correcting eyepiece but that didn't really happen on a commercial scale and it got more extreme with apochromats. AO introduced a common correction system using the same eyepiece for achromats and apochromats around 1960 and since then all objectives in systems have progressively become more harmonized in terms of their ca corrections and eyepiece compatability quite simplified. This means that the hues and distribution of ca should be uniform within an objective family. This is not the case with your Motic objectives, nor is there a rational optical symmetry to the ca in any of them. As before with the condenser installed, the ca shifts polarity from hemisphere to hemisphere, wavelengths between objectives and even within different areas of the field of an individual objective. It isn't the condenser causing this random distribution of chromatic aberration but a tipped condenser wouldn't help.
With achromats the ca should be some version of blue on the inside of objects and some version of yellow to red on the outside, virtually non-existent center stage, and radially symmetrical. None of your objectives conform to that norm. In general too, there is far too much of it for modern objectives and the current glass availability. This is the kind of distortion and aberration one would expect of a 100.00 student scope.
I think perhaps, address the optics first and maybe the software won't seem such a necessity.