Axial Lights II (Ghostbuster?)

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by rjlittlefield »

mjkzz wrote:
Sat Jan 22, 2022 4:13 am
I set up a casual demo just now, consisting of a sheet of window glass, roughly 3 mm thick, a laser pointer with an aluminum foil pinhole, a white target, and a camera with macro lens, shooting a picture of the laser dot with and without the glass in place.
the secondary dot should be 1.414*thickness of glass.
No, not nearly that big. Your geometry omits the effects of refraction, which make a significant difference in the numbers.

Here is a reasonably accurate diagram of what happens when a ray of light hits a piece of glass at 45 degrees.
GlassReflectionAndRefraction.png
This diagram started life in WinLens 3D, tracing the main outgoing ray through a block of glass with RI=1.5182 (Hoya E-C3). I pulled that diagram into PowerPoint and added the other rays by eyeballing. The diagram is not exact, but it should be good enough to show what's really going on, and why the secondary dot is a lot closer than 1.414*thickness of glass. If you'd like to work out the exact angles and distances, apply Snell's law for the angles of refraction.

--Rik

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by rjlittlefield »

ray_parkhurst wrote:
Sat Jan 22, 2022 1:12 pm
rjlittlefield wrote:
Sat Jan 22, 2022 12:28 pm
The issue of aberrations may be in play in other people's reports of ghosts. If you're predisposed to think of reflections and not aberrations, it would be simple enough to describe the image degradation shown above as a ghost, rather than a blurring caused by aberration.
Yes, this is the case with my inputs. I was not sure exactly how to describe the effects you show in your thickness sweep, so in @mjkzz's previous "Axial" post, I stated that the cause was "ghosting/CAs/Distortion", when indeed it is probably other aberrations in play.

One thing I have found though is that the aberrations even on 1.1mm glass are substantial, which is why I was so interested in shooting with 0.7mm and even the 0.17/0.2mm cover glasses.
Excellent -- thanks for the clarification!

--Rik

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

And here, to make it clear to other readers regarding omnidirectional lighting:
This is what I am saying, without omnidirectional light source, just like what you acknowledge if some black paths are not there, you will NOT get a diagram like you drew. So I think it is kinda at least misleading if people do not read carefully. So, it, the diagram in that form, has a lot to do with omnidirectional light sources.

With colimated light sources, you would only get one path in that diagram.
Now, if I interpret your drawing a bit differently, and it might be your original intention -- it is showing different POSSIBLE light paths, NOT light paths that would exist in real world in real time from a single point. Then it totally makes sense. Even with collimated lights, those paths might exist, just not from the same point.

Regarding the distance of the secondary spot, yes, I was aware of the angle change and the 1.414*d is sort of a approximation, but it seems the change in angle is large enough making it, like you said "a lot closer". This is probably why I could not get it to show with 1.1mm glass -- it is too close and weak, they sort of get mixed into one.

I have some stuff ordered and coming soon, to do some more experiment on this. Super interesting.

Macro_Cosmos
Posts: 1511
Joined: Mon Jan 15, 2018 9:23 pm
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by Macro_Cosmos »

No idea, I said "experiment", so it was not for photography at all. I was not trying to mislead anyone, it was obviously applicable to my experiment only.

I would not be allowed to show anything. It is very basic, one can say it was identical but due to the nature of the objective, a hotspot was induced. The hotspot was not present without the beamsplitter. When a relay lens is focused to infinity, light rays are parallel, no hotspot should occur, right? That is why it was quite puzzling, but then, the objective lens was not some kind of ordinary thing designed as an infinity objective. You can think of it more like some kind of lens coupling system, or mounting a finite objective to a relay lens. Altering the distance between the objective and the relay lens would alter its focus. The objective must be placed at a specific distance away from the relay lens, otherwise it will not generate an image. The relay lens images what the objective lens projects. It is called "Binocular Indirect Ophthalmoscopy (BIO)" if you are interested in seeing photos of the retina. In our case, the binoculars was replaced with a detector.

My best guess was that the hotspot was a combination of ceiling lights, and both the beamsplitter and the objective lens itself, which was placed in front of a lens focused to infinity. Using a cube beamsplitter, the hotspots were at least minimalised. With the lights turn off, lots of the glare was gone. I chose to house the optical train in light bafflers because turning off the lights violated safety codes #-o.

Polarisers were an issue because light was scarce, however, the scarcity of light was due to the boost-buck converter being faulty. It ended up being too bright even with polarisers. Installing an iris also eliminated much of the glare, however, one central hotspot remained. My solution was to simply take two images, one with the objective lens tilted, which changed the location of the hotspot. Now, blend the two together. The supervisors did not seem to like this solution.

We did not use a beamsplitter cube because of its size. We needed a beamsplitter that was around 60mmx60mm, a beamsplitter cube of that size would be forbiddingly expensive.

Well, none of this really matters in photomacrography. I do have an arsenal of beamsplitters, but I would likely not have any time to produce a sound comparison/experiment until longer holidays. What is there to compare anyway, image quality maybe? It is hard since my cube based one is a DIY setup and plate based one is pre-made into the microscope I use. My DIY job is likely going to underdeliver. I have yet to acquire the parts necessary for Köhler illumination.

I have used both plate and cube beamsplitters with infinity systems and I have not seen hotspot/glare/ghosting issues within the objective's accepted projection circle on a full-frame camera. Outside of it, sure, degradation is huge, more exaggerated, but still none of these issues unless at low magnifications (<4x), such as 2x, which makes what you have shown here very useful.

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

Thanks for the image. I think in this case, if you shine a laser at the mirror, you would see 3 visible dots coming out of it: one is due to first surface as expected, one is from the 2nd surface, also as expected. The third one would be the one reflected at point 4 in your other diagram where, because of it is a mirror instead of glass, no refraction is occuring, and because of high reflectivity of silver coating, the dot coming out of this part is still significant as it only goes through reflection by glass (vs silver plate) ONCE, the intensity is reduced due to earlier reflections, but only slightly. It is also evident that the wire seems wider/thicker than doubling, more like tripling because of the third reflection.

This is super interesting because if I can show it in an experiment, this would be much sounding, concrete proof of "secondary ghosting" How exciting is that!!! :D

Image

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

Well, it is three very visible dots, there are more but weak enough . . . sort of point this out :D

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by rjlittlefield »

mjkzz wrote:
Sun Jan 23, 2022 8:10 pm
I think in this case, if you shine a laser at the mirror, you would see 3 visible dots coming out of it
In principle there is an unlimited sequence of dots, with exponentially decaying amplitude, due to a repeating sequence of reflections off the front and rear surfaces. In a quick setup just now, I counted 5 dots visible by eye in subdued room lighting. The spacing is not large, so I strongly suggest to use the pinhole aperture trick.

If you live in an area where double-pane windows are used, then this sort of ghosting is commonly seen with small decorative lights around windows.

It happens that I still have some Christmas lights up, so it is easy for me to shoot an example.

Here is the view from outside my house, looking in at the lights and my computer. I have taped a piece of black paper behind the lights to make their reflections more visible:
XmasLightsOverview.jpg

Here is a closer crop of just the lights and their reflections:
XmasLightsCloser.jpg

And another crop, closer yet, where I have annotated the major groups of reflections:
XmasLightsClosest.jpg

The labeling here needs some explaining. A double pane window consists of two panes (two sheets) of glass, separated by a distance much larger than each pane's individual thickness. There are reflections from each surface of each pane, but because the panes are widely separated, the reflections organize into groups, each group representing a cycle between reflecting from some surface of the outside pane and then reflecting from some surface of the inside pane. So, the "direct" cluster actually contains the primary unreflected ray, plus closely spaced double reflections within each pane; the "2-bounce" cluster contains rays that reflected from one surface of the outer pane and one surface of the inner pane and then got out; the "4-bounce" cluster contains rays that reflected from outer/inner/outer/inner, and so on. In the second picture, reflections up to 8-bounce are clearly visible, with 10-bounce faintly visible if you go looking for them.

It is an amusing and educational exercise to try figuring out the structure of the reflectors, given what you can see of the reflections. Sometimes the process is not simple, particularly when the panes are not quite parallel.

--Rik

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

In principle there is an unlimited sequence of dots, with exponentially decaying amplitude, due to a repeating sequence of reflections off the front and rear surfaces. In a quick setup just now, I counted 5 dots visible by eye in subdued room lighting. The spacing is not large, so I strongly suggest to use the pinhole aperture trick.
Yes, I rectified it with another post, indicating more "bounces", some later ones are just weak in certain configuration.
It is an amusing and educational exercise to try figuring out the structure of the reflectors, given what you can see of the reflections. Sometimes the process is not simple, particularly when the panes are not quite parallel.
So true!!! It is really intriguing. Sometimes, something we are looking for is right in front of us all along.

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

In your tree light example, I think the (incident) angle is large enough, maybe almost parallel to the surface of window. I googled the formula for reflectance (reflectivity), and it seems the larger the incident angle, the larger the reflectivity, this makes sense because as we tilt a piece of glass, the more it is tilted, the more it looks like a mirror. I did not graph the relationship between reflectance vs incident angle, but I think the local maxima is when the incident angle is close to 90 (ie, parallel to glass surface), but this is just a guess and seemingly conforms to our daily life experience. If it is not the local maxima, do not pick on me :D

Therefore, in your tree light example, the reflectivity is very large, causing even the 6th bounces is bright enough. Again, thanks for amazing example!!!

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

Furthermore, the extreme reflectivity will not happen due to 1. cone of field of view, those rays will never hit the sensor 2. in case cone of field of view is really wide, the tube I placed around the subject will limit the extreme angled light rays from reaching the sensor. So overall reflectivity will be well controlled, making secondary rays much weaker.

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

Finally found some graphs plotting incident angle and reflectance (reflectivity), it looks like it is part of some physics class, PDF is here. Note, the bottom graph is when light travels from denser medium to air where the incident angle is actually the refracted angle from air to denser medium. Using typical 1.52 for glass (blue graph), at 40 degrees, that angle is 25.01, at 45, that angle is 27.7, and at 50 degrees, that angle is 30.26. Therefore change of incident angle from air to denser medium results in smaller change in that angle.

Since the secondary rays in Rik's diagram happens within denser medium, in this case, glass. We should use the bottom graph, and for incident angle from air to glass, after translating to incident angles in glass, we can see reflectance (reflectivity) might vary from 8% to 12% (blue graph). Even at 12%, I think there is a 6 stops of difference in intensity when incident angle (from air to glass) varies from 40 degrees to 50 degrees.
Incident_Reflectance.png

rjlittlefield
Site Admin
Posts: 23564
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by rjlittlefield »

I have not worked through your numbers in detail, but the description seems generally correct. As the angle becomes less glancing, the reflectivity drops off. When I'm looking at the scene in real time, this change of reflectivity causes the higher-order groups to appear and disappear as I change my viewing position to be more or less glancing.

With the Christmas lights, at 45 degrees, the 2-bounce reflections (first group) are still large enough to see even in subdued daylight against a light background. All the others are gone. Here is the image, again looking at lights through the glass so that we cannot see any one-bounce reflections.
XmasLightsDaytime.jpg

Of course this image shows reflections from very bright sources. It is the high brightness, multiplied by the low % reflected, that makes the reflections visible here.

In the application that started this discussion, using angled plain glass to get axial lighting, I agree that even the two-bounce reflections will be too dim to matter.

--Rik

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

I think in this example here, the field of view is very wide. If you use a macro lens, say, 1:1, I am not sure if you can capture those bounces, probably you just capture the lights -- the distance between the start of bounces and the light is quite far, looks more than the length of the lights.

And if you place a tube around the led lights on same side of light, which should have certain distance from the glass, I think those bounces will disappear or number of bounces will be reduce and weak, this is what I was saying that when field of view is wide, that tube should help limit the incident angle to the glass, or at least have the reflectivity under control.

Image

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

Another difference between these images vs my setup is this: in my setup, the LED light was barn-doored, having about 5cmx5cm area, at about one meter away, it would appear to be very directional and parallel. A facet on the subject will reflect this light and is also directional, ie, the original graph would have a few narrowed grouped lines (depending how "parallel" the light source is).

On the other hand, the LED lights seem to be diffused and are emitting lights, so a dot on the LED "radiates" a sphere. I think this will complicate things -- it is a LINE where the sphere of light intercepts windows glass plane rather than a POINT. Not sure how this plays out, but interesting.

mjkzz
Posts: 1681
Joined: Wed Jul 01, 2015 3:38 pm
Location: California/Shenzhen
Contact:

Re: Axial Lights II (Ghostbuster?)

Post by mjkzz »

Actually, looking at graph, I think the place where maximum secondary rays would be where reflectivity in glass is 50%, looking it up, it would be 40 degrees, translate that into incident angle in air to glass, it is about 77 degrees.

So, I can't wait for stuff to come, found a 10mm thick acrylic board, refraction index for acrylics is about 1.49, very close to 1.52 of glass, so I did the following experiments. Laser outputs a line, instead of a dot, have to tape it up a bit as this is a cheap laser.

45 degrees (those rays on either side of acrylic are at 90 degree)
45.JPG
60 degrees (to the normal), if you pixel peeping, you can see some faint secondary rays already.
60.JPG
70 degrees (did not measure, eyeballing), the secondary rays are visible, there is a faint third one,
75.JPG
I also tried to vary the angle a bit at 45 degrees, no observable secondary rays that forms ghosting. For the larger incident angles, those rays will not be seen by sensor, so no ghosting, either.

Post Reply Previous topicNext topic