Color aberrations

[PaulFurman]

SEM = Scanning Electron Microscope

[bklein]

Just concentrate on the white smurge on the leftmost of the head shot. What the heck causes that? Nothing in the other shot to cause it.
I don't get it.

[rjlittlefield]

Just concentrate on the white smurge on the leftmost of the head shot. What the heck causes that? Nothing in the other shot to cause it.
I don't get it.

I can't tell what you're talking about. Which image, what coordinates?

--Rik

[rjlittlefield]

I still have this fly set up, and it occurred to me just now (duh...) to do a direct visual inspection -- naked eye and with 10X loupe. Sure enough, a lot of those bristles act like tiny cylindrical mirrors, becoming very bright when the light, the bristle, and the eye are in just the right positions. Move the eye, and the bright bristles go dark while some others get bright instead. On direct visual, a small fraction of the bright reflections are intensely colored.

--Rik

[Lou Jost]

Bumping up this old thread since it was recently referenced by Rik. To me this looks a bit like the speckle one sees when working with laser illumination. You can see something like this on your thumbnail in daylight on a very clear day, or even from a streetlight at night. Light is interfering with its reflection in random ways due to nanometer-level roughness of the surface. It would be really interesting to see if the rainbow would disappear by decohering the light, for example by passing it through a fast-rotating ground glass placed directly in front of the light source (so that it does not affect the apparent size of the light source), and using an exposure that is relatively long. Maybe just vibrating the light source and making a long exposure would eliminate the rainbow?

[rjlittlefield]

It would be really interesting to see if the rainbow would disappear by decohering the light, for example by passing it through a fast-rotating ground glass placed directly in front of the light source (so that it does not affect the apparent size of the light source), and using an exposure that is relatively long. Maybe just vibrating the light source and making a long exposure would eliminate the rainbow?

Well, the illumination discussed here originates as random photons emitted from a hot radiator. I don't know how to make anything less coherent than that. So, I would be very surprised if further decohering the illumination were effective at removing the colors.

To me this looks a bit like the speckle one sees when working with laser illumination. You can see something like this on your thumbnail in daylight on a very clear day, or even from a streetlight at night. Light is interfering with its reflection in random ways due to nanometer-level roughness of the surface.

On the other hand, I agree completely with all of this.

Yes, the colors are speckle patterns that result from interference due to slight roughness of the surface, and yes, they still appear even with nominally incoherent light.

For me, the key to understanding this effect is to realize that even "incoherent" light is in fact largely coherent over short time intervals, corresponding to only a few cycles at the average illumination frequency.

In setups where the incoming light spans only a narrow range of angles, this also means that the illumination is largely coherent over every small spatial extent of the surface, as seen by the observer. Focus your attention on any small area, and what you see is that one wavelength (say red) interferes constructively, while another wavelength (say green) interferes destructively. If you were to observe over very short time intervals, you would see the light reflected from that area flickering in both wavelength and intensity, but when you average over longer time intervals you simply see that the area looks colored -- red in this example. At the same time, a nearby area with different texture will have different interferences and will appear to be a different color.

To understand how a diffuser removes the color, note that the interference depends on the angle of incoming illumination. From one angle interference may reinforce red, while a different angle may reinforce green and a third angle reinforces blue. When you provide light at all those angles, and add together the individually interfered results, the interference effects average out and the colors go away.

Some years ago I learned that there's a bit of standard terminology for describing this situation. That term is "partially coherent". For that discussion, see viewtopic.php?p=126689#p126689 and the surrounding (long) thread.

--Rik

[Lou Jost]

Rik, yes, I agree with all that. But my point is that you may be able to eliminate the interference pattern even without diffusion. At least with lasers, the method I described works well. I may be completely wrong though.

[rjlittlefield]

Rik, yes, I agree with all that. But my point is that you may be able to eliminate the interference pattern even without diffusion. At least with lasers, the method I described works well. I may be completely wrong though.

OK, I set up a simple test. It uses a fly, illuminated by LED flashlight shining through a 3mm hole located 200 mm from the subject and positioned to create some nice colors. I shot the fly using the LED by itself for 2.5 seconds. Then I shot a second time with the LED shining through a single layer of Scotch Magic Mending Tape mounted on a glass slide, which I waved around by hand during a 5 second exposure.

Here are the test conditions:


And here are the results, presented as a two-frame animation, one frame per second:



Applying that same moving-tape treatment to laser speckle kills the speckle, but here it has essentially no effect.

Personally I find this pretty convincing. Do you agree?

--Rik

[CrispyBee]

I see those balls all the time with beetles, butterflies and reptiles - it's refraction, like a prism, breaking the white light into multiple colours but as already mentioned due to the microstructure (or nanostructure?) being uneven it's not a regular pattern but varies throughout the image.

You can see some of that effect occasionally even when the specular highlights are in focus, but when out of focus the light really starts to get separated:



Those "bokeh-balls" often seem to have a harsh structure to them that is similar to (or rather an enlarged representation of) the actual surface texture, where lines/ridges show as streaks and dimples appear as a sort of onion-ring... but that might just be me.

You might think that these highlights are a result of bumps, so the bokeh balls should reflect that:



But as you can see here they're actually a result of ridges and lines catching and refracting the light:

[rjlittlefield]

it's refraction, like a prism, breaking the white light into multiple colours but as already mentioned due to the microstructure (or nanostructure?) being uneven it's not a regular pattern but varies throughout the image.

Why do you say it's refraction, instead of interference?

This effect occurs even with totally opaque metallic surfaces, as discussed in the thread that I referenced (viewtopic.php?p=126689#p126689).

I agree with the characterization of breaking the white light into multiple colors, but the mechanisms of refraction and interference are entirely different.

--Rik

[CrispyBee]

it's refraction, like a prism, breaking the white light into multiple colours but as already mentioned due to the microstructure (or nanostructure?) being uneven it's not a regular pattern but varies throughout the image.

Why do you say it's refraction, instead of interference?

This effect occurs even with totally opaque metallic surfaces, as discussed in the thread that I referenced (viewtopic.php?p=126689#p126689).

I agree with the characterization of breaking the white light into multiple colors, but the mechanisms of refraction and interference are entirely different.

--Rik

with beetles in particular (sub-surface) refraction from the many layers of chitin play a big role:
https://www.researchgate.net/publicatio ... ab_beetles

Chitin is really surprisingly translucent; For this shot I've not used any rear illumination or reflectors and had everything very well diffused - and still light managed to get though and get reflected back again at a different point:



On a microscopic level that wouldn't cause interference but on a macro level it certainly could - similar to a thin film interference (which is also caused by sub-surface refraction and reflexion).

[Lou Jost]

Rik, yes, I agree with all that. But my point is that you may be able to eliminate the interference pattern even without diffusion. At least with lasers, the method I described works well. I may be completely wrong though.

OK, I set up a simple test. It uses a fly, illuminated by LED flashlight shining through a 3mm hole located 200 mm from the subject and positioned to create some nice colors. I shot the fly using the LED by itself for 2.5 seconds. Then I shot a second time with the LED shining through a single layer of Scotch Magic Mending Tape mounted on a glass slide, which I waved around by hand during a 5 second exposure.

Here are the test conditions:


And here are the results, presented as a two-frame animation, one frame per second:



Applying that same moving-tape treatment to laser speckle kills the speckle, but here it has essentially no effect.

Personally I find this pretty convincing. Do you agree?

--Rik

Yes, I am convinced. The tape trick worked for the laser and not for this. Case closed. But I still wonder why. Thanks for doing the test!

[rjlittlefield]

Earlier, I mentioned

This effect occurs even with totally opaque metallic surfaces

Following is an illustration of that statement. All that I've done here is to remove the fly and replace it with a piece of aluminum that has been freshly sanded with 1500 grit paper.

In addition to the two treatments shown with the recent fly, I have added a third treatment in which the layer of translucent tape is simply moved close to the subject so it serves as a diffuser, greatly increasing the angular spread of the illumination as seen by the subject. Although the moving tape located at aperture has almost no effect, the introduction of diffusion causes a striking disappearance of color blotches and bogus extended DOF. These effects are discussed at length in viewtopic.php?t=19582 .



In the case of this aluminum, there is certainly no refraction by the subject. The spectral colors are strictly an interference effect.

I believe that most other instances are interference effects also. That's why I twitched at the characterization that

it's refraction, like a prism,

I asked:

Why do you say it's refraction, instead of interference?

and got the reply that

with beetles in particular (sub-surface) refraction from the many layers of chitin play a big role:
https://www.researchgate.net/publicatio ... ab_beetles

I like that reference, but I take away a message that is different from what I read in your own brief description.

In the paper, which is titled "Simulation of light scattering from exoskeletons of scarab beetles", refraction is mentioned only as a cause of scattering due to spatially varying changes in refractive index. There are wavelength dependencies, but they relate to spatial variation in the structure. I see no mention of refractive index varying by wavelength, which is the mechanism that a prism uses to separate colors.

So, as I read the paper all of its ways to break white light into colors are properly described as forms of interference, not as "refraction, like a prism".

For a photographic practitioner this distinction probably makes no difference. The key aspect is just that "color separation happens, and adding diffusion makes it happen a lot less".

It's just that if I'm going to offer people an explanation, I like it to be as correct as possible in context.

--Rik

[CrispyBee]

I believe that most other instances are interference effects also. That's why I twitched at the characterization that

it's refraction, like a prism,

I asked:

Why do you say it's refraction, instead of interference?

I'm so sorry, I meant to say that the cause of the interference is not just surface scatter but there's also refraction at work. With metals it's of course only surface scatter that results in the interference.

When I wrote "like a prism" I didn't mean it's the exact same thing, I was merely (mis)quoting articles such as this one:

https://phys.org/news/2011-04-beetle-bl ... allic.html

This is similar to the way in which a prism breaks white light into the colors of the rainbow by refraction, but in the case of these beetles, different wavelengths, or colors of light are reflected back more strongly by different layers of chitin. This creates the initial palette of colors that enable the beetles to produce their distinctive hues.

Here it also likened to a prism and sub surface refraction+reflection is again the base principle vs pure surface scattering at least from what I understand - I am no physicist. I just try to make sense of this stuff so you have to allow for some leeway when it comes to being hyper-precise.

But for refractive indices there actually is data:

https://opg.optica.org/oe/fulltext.cfm? ... &id=224298

and here's a nice read for surface scattering combined with thin-film-interference in butterfly scales:

https://opg.optica.org/oe/fulltext.cfm? ... 7&id=63420

I doubt it's really possible to nail the precise refractive index of specific beetles, even within the same species there can be a huger variation in the resulting coloration or "visible" coloration (I recently found a rose chafer with the deepest shade of red/dark copper I've ever seen).

[rjlittlefield]

I am convinced. The tape trick worked for the laser and not for this. Case closed. But I still wonder why.

For those who are not familiar with the tape trick, let me illustrate.

The following pictures show the setup. The objective is looking at a piece of aluminum, the same one illuminated by white light in the previous post. For this post, it's illuminated by a laser pointer (barely visible at upper left) that is shining on a fixed diffuser, so that the subject is illuminated by diffused but still coherent light. In one situation everything stays in one place while the exposure is made, and in another situation there's a moving piece of magic mending tape in the path of the laser beam just before it hits the diffuser.



Now let's look at what the camera sees. With everything fixed in place, the laser light produces an extremely strong speckle pattern. The general character of the speckle pattern varies almost not at all regardless of how far out of focus the real subject is. However, the details of the speckle pattern are extremely sensitive to the combined alignment of the laser, the diffuser, and the subject.

In the following animation there are three frames. The first two frames are shot in "identical" situations, except that between exposures I slightly shifted my weight from one foot to the other. That slight shift of weight flexed the floor, which flexed the table, which flexed the optical breadboard, which altered the alignment of the laser, diffuser, and subject. The net change in alignment was tiny, but you can see how drastically the speckle pattern changed. The third frame is with the moving piece of tape in the path of the laser beam. (Fixing the tape in place produces an image with the same character as no tape, but with all the details different. I forgot to shoot an example of that.)



So, the reason why moving the tape kills the speckle pattern is because during the exposure a gazillion different and essentially independent speckle patterns get averaged together. The result of averaging the gazillion independent patterns is almost no speckle.

That part is simple enough, at least after a good physics course.

But then I agree it's much harder to understand why a similar moving tape trick does not work with white light shining through a narrow aperture.

I think maybe I can explain that, at a high level, if I'm permitted a certain amount of handwaving.

A good place to start is to ask why, with the laser, are the gazillion speckle patterns essentially independent of each other?

The answer to that question is because the laser has so much coherence that for its speckle pattern "everything affects everything". Perturb any part of the setup by a few wavelengths and you get a totally different pattern. In particular, if you disturb the illumination wavefront by even a small amount, you get a new speckle pattern for which the intensity at each pixel position is statistically independent of the previous one. Averaging over a gazillion patterns, the statistical independence gives the same average value everywhere, and hey presto the speckle pattern is gone.

Now consider the situation with narrow angle white light. In that case the coherence is only local and transient. It's transient because the EM field oscillations are more-or-less stable in temporal neighborhoods of only a few cycles, and it's local because the temporal variations prevents different parts of the setup from consistently interfering with each other if they're more than a few wavelengths apart. In this situation, visible interference patterns are much less likely to form, but when they do, the pattern is determined primarily by local structure in the subject and not by the exact shape of the illumination wavefront. In this case, when you average over a gazillion different shapes of illumination wavefront, you're not averaging over independent interference patterns so the pattern doesn't go away.

Another way of forming a useful intuition is to restrict your attention to planar wavefronts that come in from slightly different angles at different times. With the laser, the large scale coherence means that reflections from millimeters apart can interfere with each other, and that in turn implies that very slight changes in illumination angle make huge phase changes to contributing components from different parts of the structure. But with the white light where the coherence is only local, it is only reflections from a few microns apart that can interfere with each other, and that in turn implies that slight changes in illumination angle make only small phase changes in the components that cause the visible speckle. To get large phase changes in the contributing components, you would need to make large changes in the illumination angle, and those changes are prevented by the setup.

On the other hand, averaging over a gazillion large differences in angle will give statistical independence that makes the speckle pattern average out, and that's why adding a diffuser kills the speckle with white light.

This is the first time I've tried wordsmithing this explanation, so I'm not sure if it makes sense to anybody else.

Does it?

--Rik