Powers of 10 (Cambridge Silicon Radio BC4-EXT)

[Andy Resnick]

Here's a series of macro images zooming in on a Bluetooth chip, beginning at 1:1 reproduction and continuing. This was a learning experience, for sure... so much to keep track of! Because of the limitations on the number of images per post, I am breaking this sequence up. I'm also not putting scale bars on every image...

The first sequence consists of 1X (55mm Micro Nikkor), 3X (Zoom Luminar), 10X (63mm Luminar), 30X (25mm Luminar), and 100X (Nikon 20/0.75 fluar objective).

For the 1:1, I include a penny for scale. This image, by far, was the most difficult to get a satisfactory result. I shot at f/8 to give me sufficient depth of field, and the background is the lens cap.

Then, at 3X:

10X:

30X: here I needed to switch to epi-illumination because there wasn't enough space to pipe in light.

To get higher magnifications, I now switch from using Luminar lenses to microscope objectives. Still a macro setup!

So, this chip seems to have an informal (project?) name: "Coyote". I wonder why? Well, looking around with the 20x Nikon:

Next post: a suite of 40x objectives.

[Andy Resnick]

Continuing on, I tried 3 different 40X objectives to see how they work in an extreme macro setup.

First is the Zeiss 40x/0.85 POL: A dry no-coverslip objective:

Next is the 40x/0.95 Olympus SPlan APO: a dry yes-coverslip objective:

Next is the Zeiss 40x/0.9 Neo-fluar multi-immersion, here using oil:

I'm not terribly surprised by these results. The numerical aperture of these lenses are somewhat low for the extreme magnification I am using (170x-ish), and the uncompensated lateral chromatic aberration is apparent. I was hoping the Neo-fluar would perform better, because in transillumination (especially DIC) it's freaking awesome. To be sure, all of these lenses are... let's call them 'old'. How does a modern lens compare?

Here's the result of using a 40/1.3 Leica HCX PL APO:

Nice! Using infinity-corrected lenses is no problem, since the bellows length is close to 1 meter. It's apparent that ol' Wile E. consists of 2 separate layers. Looking around the chip perimeter, I also find this:

If you don't recognize this, it refers to a meme-type thing dating from ancient times. CSR uses this particular objet d'art again in the BC5, the subject of a future sequence.

Next: maximum magnification!

[Andy Resnick]

At this point, I have magnified the object abut 170x. How much further can I go? Clearly, there is some limit set by the wavelength of visible light, and I don't plan to start imaging in XUV.

The highest NA lens I have is a Leica 100/1.47 HCX PL APO, which gives me a magnification of 400X with the current setup. Since Wile E. consists of 2 layers, I'll simply montage the two images:

But this isn't the highest possible magnification I can achieve. By removing the 0.8X reducing lens and racking the Ultraphot mirrors to maximum bellows length, I can increase the magnification to 620X, but this is clearly 'empty magnification':

Gotta love the vanishingly small depth of field.... I'm also pushing the limit of what the (manual) z-stage mechanism can provide.

And then I can zoom in digitally. Again, what you see here is 'empty magnification'. And a dirty sensor.....

The bar represents 2 wavelengths of visible light! The transitions from light-dark are slightly less than 0.25 microns wide. Using a Richardson test slide, I can verify the ability of this macro setup to resolve 0.25 micron wide bars spaced 0.5 microns apart: Note, this image was taken at the 'minimum' magnification setting, so the size of the features here won't match the scale bar.

Phew! Like I mentioned at the top, this was a real learning experience for me. Onwards and upwards (downwards?) to see other denizens of the silicon zoo!

[rjlittlefield]

Because of the limitations on the number of images per post, I am breaking this sequence up.

The limitation is actually on number of images per day, because its intent is to have people posting their best stuff thoughtfully.

We formalized the rule long ago as a defense against having to look at 20 pictures of the same fly at the same size on the same leaf, in slightly different random poses.

The limitation is also administrative only, and we're happy to make exceptions by prior approval where there's a good reason, such as a powers-of-N series.

The bar represents 2 wavelengths of visible light!

Is this image shot with oil immersion? If so, are we talking about the wavelength in air, or its much shorter wavelength in oil?

Onwards and upwards (downwards?) to see other denizens of the silicon zoo!

I'd be very interested to see some of the deep 3D structures like at viewtopic.php?f=27&t=45872&p=288161#p288161 . With well behaved illumination and very shallow DOF, I'll bet you could make some really nice stereos. (I don't see either "zerene" or "stereo" in any of your postings, so I'll mention to see https://zerenesystems.com/cms/stacker/docs/syntheticstereo if you're not familiar with the technique.)

--Rik

[Andy Resnick]

The limitation is actually on number of images per day, because its intent is to have people posting their best stuff thoughtfully.

We formalized the rule long ago as a defense against having to look at 20 pictures of the same fly at the same size on the same leaf, in slightly different random poses.

The limitation is also administrative only, and we're happy to make exceptions by prior approval where there's a good reason, such as a powers-of-N series.

Thanks for the clarification- I understand.

Is this image shot with oil immersion? If so, are we talking about the wavelength in air, or its much shorter wavelength in oil?

I was referring to visible wavelengths in air, it is true that the lens used for that image is an immersion objective. 555 nm or 365 nm, take your pick

I'd be very interested to see some of the deep 3D structures like at viewtopic.php?f=27&t=45872&p=288161#p288161 . With well behaved illumination and very shallow DOF, I'll bet you could make some really nice stereos. (I don't see either "zerene" or "stereo" in any of your postings, so I'll mention to see https://zerenesystems.com/cms/stacker/docs/syntheticstereo if you're not familiar with the technique.)

I have not tried looking at an RFID device- that looks intriguing! Regarding stereo/stacking, I only have very limited experience with stacking (CombineZP, Picolay), I struggled for a while to get decent results before giving up. This is possibly the best one I was ever able to do, I think 10 images were used:

If you think improvement is possible, I'm always interested in trying. My biggest peeve is the sensor dust.... because in image space my system NA is effectively zero, every single mote of dust on the sensor is imaged perfectly. If you look closely at the image above, you can see trails created by the dust motes due to the stacking. Are there any apps I can try that can remove those? At some point, I can take another set of images and give it another try.

Stereo is an interesting idea, not sure how I would implement it... know any tutorials?

[rjlittlefield]

Regarding stereo/stacking, I only have very limited experience with stacking (CombineZP, Picolay), I struggled for a while to get decent results before giving up. This is possibly the best one I was ever able to do, I think 10 images were used:

That looks very promising.

I am intrigued by the lighting. The fact that all surfaces are bright suggests axial illumination, but the dark outlines that appear left and below some features look like shadows caused by directional illumination from the upper right. Can your system control the axial illumination so as to get some directionality?

My biggest peeve is the sensor dust.... because in image space my system NA is effectively zero, every single mote of dust on the sensor is imaged perfectly. If you look closely at the image above, you can see trails created by the dust motes due to the stacking. Are there any apps I can try that can remove those?

If your focus stepping mechanism is well behaved so there is no significant sideways movement from one frame to the next, then the simplest approach is to turn off computational alignment in the stacking software. That way spots in the input produce spots in the output, rather then turning into trails.

In your example here, I notice that the dust trails are short, dense, and almost straight, and also that the subject is darker on one side than the other. That combination of aspects suggests that the lateral shifts that turned spots into trails was actually introduced by the computational alignment process, which can easily be led astray by how features change appearance as they go into and out of focus. So, my best guess is that just turning off computational alignment would solve your trails problem.

In Zerene Stacker, you can turn off computational alignment by just un-checking all the options at Options > Preferences > Alignment. I've put that into the recipe that appears below.

If you do need to allow computational alignment, then stacking software commonly lets you provide some sort of dust mask that tells the software where to replace dust spots with pixel values from their surroundings. That works OK, but generating the mask in the first place is extra work. Except for working with film scanners that have IR channels, I don't know any software that is very good at automatically identifying and getting rid of dust spots.

Stereo is an interesting idea, not sure how I would implement it... know any tutorials?

For this application, the easy way to do stereo is to shoot a single stack with shallow focus steps, then use the "synthetic stereo" capability of Zerene Stacker to generate off-axis views to make the stereo pair from the single stack.

If you're new to Zerene Stacker, I recommend to start by reading the Getting Started section of How To Use Zerene Stacker, at https://zerenesystems.com/cms/stacker/docs/howtouseit. Then take a look at the page on Synthetic Stereo & 3-D Rocking at https://zerenesystems.com/cms/stacker/d ... eticstereo .

A step-by-step recipe that goes straight to stereo looks like this:

• Launch Zerene Stacker
• Drag source images to Input Files.
• Optional: go to Options > Preferences > Alignment and remove all the checkmarks, to avoid dust trails.
• Go to Options > Preferences > Stereo/Rocking and make settings as follows...
• Checkmark "Generate stereo pair or rocking sequence"
• Number of output images: 2
• Click "Show calculator" to bring up the stereo calculator.
• Enter values for sensor width and magnification. (Calculator computes stack width.)
• Enter value for focus step size, for example 0.001 mm for 1 micron.
• Click the P button to fill in number of frames. (Calculator calculates stack depth.)
• Set angle off axis to be 6 degrees. (Calculator computes a number for % shift in X.)
• Back in the the Preferences panel, fill in minus and plus that number, in the two Shift X Limits fields.
• Click OK to close the Preferences panel.
• In the main menu, Stack > Align & Stack All (PMax). (Some time will pass while two output images are generated.)
• Tools > Stereo > Start Preview to display the output images as a stereo pair. While in this mode, press-and-drag will pan the image, and mousewheel will zoom in/out.
• Click the Swap Sides button as needed to make the pair be either parallel or crossed-eye, whichever way you want the final result.
• Tools > Stereo > Stop Stereo View (or just click Stop Stereo View in the toolbar).
• Tools > Stereo > Make Stereo Pair(s). This assembles the separate left- and right-eye views into a single output image.
• File > Save Output Image(s), select format, Save Image, navigate to desired location, Save.

For cropping and other adjustments, I recommend using StereoPhoto Maker, https://stereo.jpn.org/eng/stphmkr/ .

I hope this helps!

--Rik

[Scarodactyl]

I am intrigued by the lighting. The fact that all surfaces are bright suggests axial illumination, but the dark outlines that appear left and below some features look like shadows caused by directional illumination from the upper right. Can your system control the axial illumination so as to get some directionality?

It looks to be epi DIC which gives this effect (slope is rendered into color).

[Andy Resnick]

I am intrigued by the lighting. The fact that all surfaces are bright suggests axial illumination, but the dark outlines that appear left and below some features look like shadows caused by directional illumination from the upper right. Can your system control the axial illumination so as to get some directionality?

I was using epi-illumination for this, nothing fancy (epi-brightfield?). I suspect the surface textures and 'shadows' arise because the aperture stop is not perfectly centered with respect to the objective lens, leading to slightly oblique epi-illumination conditions. The effect is definitely dramatic.

In your example here, I notice that the dust trails are short, dense, and almost straight, and also that the subject is darker on one side than the other. That combination of aspects suggests that the lateral shifts that turned spots into trails was actually introduced by the computational alignment process, which can easily be led astray by how features change appearance as they go into and out of focus. So, my best guess is that just turning off computational alignment would solve your trails problem.

Thanks for the idea- I can give it a try.

Except for working with film scanners that have IR channels, I don't know any software that is very good at automatically identifying and getting rid of dust spots.

Booo!

For this application, the easy way to do stereo is to shoot a single stack with shallow focus steps, then use the "synthetic stereo" capability of Zerene Stacker to generate off-axis views to make the stereo pair from the single stack.

[...]

I hope this helps!

--Rik

It does- many thanks!

[Andy Resnick]

I am intrigued by the lighting. The fact that all surfaces are bright suggests axial illumination, but the dark outlines that appear left and below some features look like shadows caused by directional illumination from the upper right. Can your system control the axial illumination so as to get some directionality?

I was using epi-illumination for this, nothing fancy (epi-brightfield?). I suspect the surface textures and 'shadows' arise because the aperture stop is not perfectly centered with respect to the objective lens, leading to slightly oblique epi-illumination conditions. The effect is definitely dramatic.

In your example here, I notice that the dust trails are short, dense, and almost straight, and also that the subject is darker on one side than the other. That combination of aspects suggests that the lateral shifts that turned spots into trails was actually introduced by the computational alignment process, which can easily be led astray by how features change appearance as they go into and out of focus. So, my best guess is that just turning off computational alignment would solve your trails problem.

Thanks for the idea- I can give it a try.

Except for working with film scanners that have IR channels, I don't know any software that is very good at automatically identifying and getting rid of dust spots.

Booo!

For this application, the easy way to do stereo is to shoot a single stack with shallow focus steps, then use the "synthetic stereo" capability of Zerene Stacker to generate off-axis views to make the stereo pair from the single stack.

[...]

I hope this helps!

--Rik

It does- many thanks!

[Andy Resnick]

I am intrigued by the lighting. The fact that all surfaces are bright suggests axial illumination, but the dark outlines that appear left and below some features look like shadows caused by directional illumination from the upper right. Can your system control the axial illumination so as to get some directionality?

It looks to be epi DIC which gives this effect (slope is rendered into color).

Nope- epi-brightfield (maybe slightly oblique epi-brightfield). AFAIK, there are a few different reasons for the colors: both metallic and dielectric thin film effects.

[rjlittlefield]

Except for working with film scanners that have IR channels, I don't know any software that is very good at automatically identifying and getting rid of dust spots.

Booo!

On review, my statement looks too pessimistic.

In special cases where the optics are stable and you can shoot a blank test frame, it is not difficult to construct a workable dust mask by using general tools for levels adjustment and selection.

See viewtopic.php?t=41343 , where I write

For the second example, I found that it was tedious and unnecessary to make the mask by hand from a source frame of the stack. Instead, I simply made a very long time exposure of a black scene to make the hot pixels easily distinguishable, then used levels-adjustment in Photoshop to isolate hot pixels, selected those, expanded the selection by a couple of pixels to be sure of covering up the “crosses” that are often produced by JPEG compression, and filled the selection to make the mask. Unlike dust, hot pixels don’t move around or accumulate rapidly, so this mask will be usable for a long time in the future with that same camera.

The same approach can be used for dust, modulo the issue that the dust can move around or accumulate.

--Rik

[CrispyBee]

If you have a such a blank image with the dust spots clearly visible you can create a dust mask and a vignetting removal in CaptureOne.
It's a pretty neat tool that (imo) was mostly meant for digital medium format backs that reduce tiling and avoid having to clean them all the time (they are very likely to get a bit dusty and cleaning them in the field is not recommended).
But it works really well for any camera that is supported in CaptureOne and does a much better job at removing vignetting than the manufacturer lens profiles.

[rjlittlefield]

I am intrigued by the lighting. The fact that all surfaces are bright suggests axial illumination, but the dark outlines that appear left and below some features look like shadows caused by directional illumination from the upper right. Can your system control the axial illumination so as to get some directionality?

I was using epi-illumination for this, nothing fancy (epi-brightfield?). I suspect the surface textures and 'shadows' arise because the aperture stop is not perfectly centered with respect to the objective lens, leading to slightly oblique epi-illumination conditions. The effect is definitely dramatic.

Thanks for the info. I have sometimes considered building a custom epi-illuminator for just this purpose, but it seemed like a lot of trouble for uncertain result. This is definitely a good result!

Is your illuminator commercial or custom? If the latter, can you share the design?

--Rik

[Scarodactyl]

Huh, I would not have expected that hillshade effect from epi brightfield. I wonder if somethjng is a but off-center.

[Lou Jost]

Regarding dust removal, astrophotography software always has algorithms to remove any irregularities. Astrophotographers take what are called "Flat frames" (ten or twenty) with a white diffuser over the lens. The programs use these frames (averaged together) to adjust all the real frames (the "light frames"). These are very well developed standard procedures in astrophotography. The same process also automatically corrects soft vignetting.