"Pseudo-axial" lighting using vignetting aperture

[ray_parkhurst]

For quite a few years I've used lighting techniques for coin photography that place the effective light source between the coin and lens. By trial and error I have developed a system that works very well, but I'm wondering how far I can actually push it.

Key is to understand vignetting when an aperture is placed between the lens and subject. As an example, my current system has an aperture of ~14mm. This vignettes slightly larger than the 24mm frame height with a 75mm lens set to ~1.25:1 magnification (19mm diameter subject).

My question is...what is the smallest aperture, and its physical placement, such that it vignettes at the 24mm frame height? The 14mm aperture would vignette either the 19mm subject or 24mm sensor if placed too close to either, so there must be an optimum placement for minimum vignetting that would allow me to use the smallest aperture. Additionally, how is the aperture and position dependent on magnification?

[rjlittlefield]

what is the smallest aperture, and its physical placement, such that it vignettes at the 24mm frame height?

If I understand properly what you're doing, then it seems the same as the baffle problem eventually solved & summarized at viewtopic.php?p=120423#p120423 .

There is no simple answer to the exact question that you've posed. In general the smallest non-vignetting aperture/baffle will be a rectangle with its corners rounded off to match the entrance pupil, morphing in size and shape depending on exactly where it's located. To get the most nearly axial lighting, I think you want the light source as close as possible to the lens, sized and shaped to match that smallest non-vignetting shape.

--Rik

[ray_parkhurst]

what is the smallest aperture, and its physical placement, such that it vignettes at the 24mm frame height?

If I understand properly what you're doing, then it seems the same as the baffle problem eventually solved & summarized at viewtopic.php?p=120423#p120423 .

There is no simple answer to the exact question that you've posed. In general the smallest non-vignetting aperture/baffle will be a rectangle with its corners rounded off to match the entrance pupil, morphing in size and shape depending on exactly where it's located. To get the most nearly axial lighting, I think you want the light source as close as possible to the lens, sized and shaped to match that smallest non-vignetting shape.

--Rik

Well, that thread was definitely interesting, but I don't think it helped answer my question. I think you are probably right that there is no simple answer, so it is probably back to experimentation to determine the solution.

I'm not sure about your conclusion regarding axial lighting and proximity to the lens. Do you know that is the case? If so it would simplify things quite a bit. But for sure I can make the hole in the baffle (I used aperture before, sorry) smaller if I place it away from the lens. Now, if that increases the angle vs subject, then the lighting is more like axial, but if the angle decreases due to the reduced working distance, then it is less axial. I suspect the answer may depend on the focal length and magnification.

[rjlittlefield]

But for sure I can make the hole in the baffle (I used aperture before, sorry) smaller if I place it away from the lens.

How about we just say "lighting area" and "hole in the lighting area"? I worry that the other words carry too much baggage.

Here are some cross-sections that illustrate my thinking. Entrance pupil at top diagrammed as a thin ellipse; subject field at bottom diagrammed as a flat line; optical axis centered as dashes; possible edges of the hole in the lighting area as thin red line. The bold red arrows are illumination rays, drawn from inner edge of lighting area to center of subject for various heights of the lighting area. Top set is with the entrance pupil smaller than the subject, typical of low magnification; bottom set is with the entrance pupil larger than the subject, for example with a microscope objective.

I think it's intuitively clear that the most vertical (least slanted!) bold red arrow happens when the lighting area is close to the lens. The difference is most extreme when the entrance pupil is small and distant; in that case the most vertical lighting and smallest hole both occur when the lighting area is close to the lens. The difference is much smaller when the entrance pupil is large and close; in that case the most vertical lighting still occurs when the lighting area is close to the lens, but the smallest hole occurs when the lighting area is close to the subject.

I'm not sure about your conclusion regarding axial lighting and proximity to the lens. Do you know that is the case?

I'm pretty sure about the angles as I've drawn them. I'm just not sure that those are the angles you care about. From my standpoint, lighting for coin photography seems to have a lot of aspects that are "enough, not too much, season to taste", and I certainly don't understand the nuances.

--Rik

[ray_parkhurst]

I'm pretty sure about the angles as I've drawn them. I'm just not sure that those are the angles you care about. From my standpoint, lighting for coin photography seems to have a lot of aspects that are "enough, not too much, season to taste", and I certainly don't understand the nuances.

--Rik

Thanks for the diagrams, they do help to explain the effect by relating it to the entrance pupil. This info will help with my experiments as I'll need to vary aperture as well. Should be an interesting set of data with distance, aperture, and hole size as variables.

And you're right about lighting for coins, it is usually very nuanced. In this case however the goal of getting as close to axial as possible is more a technical solution rather than aesthetic.

[ray_parkhurst]

OK, first set of experiments...

I installed an adjustable iris on the front of the lens, a 75ARD1 set to ~1.2:1 mag (20mm field height).

No vignetting occurs even at the smallest iris opening (~2mm) when the lens is set to f4. The iris does act as an optical aperture, affecting the exposure time and depth of field, but does not vignette. I see slight darkening in the corners at f5.6, but actual vignetting does not occur until f11 with the 2mm "aperture".

Adjusting the iris hole to ~10mm at f5.6, there is no exposure time increase or DOF change within the 20mm field height, so that is indeed smaller than I have been able to achieve with more extension from front of the lens. At f8 I can go down to 6mm. Smaller than these "apertures" and I see increased exposure times.

I was expecting actual vignetting, but it is not a requirement. Indeed if I can bring the "aperture" down to the 6-10mm range I can likely achieve an excellent "pseudo-axial" illumination, especially with the added angle offered by mounting the lighting close to the lens.

More experiments to do, then I'll build an illuminator that takes advantage of the new information and report back.

[kaleun96]

Interested about this topic as well as I also use a "pseudo-axial" lighting setup and vignetting is something I've experimented with but only informally (i.e. 3D printing different parts and seeing what starts to vignette).

Since coin diameters vary (at least for the coins in my collection), and thus the magnification and distance between coin and lens changes, I eventually made a setup which lets me swap in different parts to either reduce or increase the "hole in the lighting area". The pink shaded parts are the reflectors which bounce light coming from the left side (Godox TT350) down towards the coin and then back up through the adapter to the lens. The reflectors and lens adapter are swapped out depending on the diameter of the coin, accounting for the different working distance and also to better distribute the light over the coin.

The blue dashed lines just show the reflected rays which get closest to the centre of the coin (and also their reflection off the coin), as you can see there's still theoretically a "blind spot" in the middle of the coin but in practice the imperfections in the reflector bodies etc seem to fix that. The reflected rays coming from the other extreme of the reflector bodies aren't depicted but just narrowly miss the coin's edge (19mm diameter "coin" illustrated) if I remember correctly.

As Rik mentions, placing it closest to the lens is best. I have optimised the adapter to have it as close to the lens as possible as that lets me have the smallest angle of reflection off the coin's surface given the other constraints, which is about 10 degrees. I've not yet been able to achieve anything less than 10 degrees while still getting good coverage across the entire coin and without vignetting due to having to reduce the size of the "hole in the lighting area" to let the reflectors encroach further over the coin from above.

[ray_parkhurst]

Nice looking system!

I ran an experimental sweep using a small ringlight. I had the ringlight mounted for use in imaging phono styli, but had not tested it with coins until today. It is a "30mm" LED ring, with actual dimensions of the illuminating phospor of 26mm OD, 16mm ID, with some flocking on inside reducing the "hole" to 14mm.

Sony A7Rm4
75mm Apo Rodagon D M1:1
Magnification ~1.2:1
Subject is a Lincoln Cent with Red/Brown toning

The overall WD from lens filter threads to coin is ~100mm. This is divided by the ringlight into Coin-Ringlight (CR) and Ringlight-Lens (RL) distances. I shot both the coin and the reflection of the ringlight by placing a mirror instead of the coin. This gives a nice visualization of the areas of "pseudo-axial" direct reflection. The dark areas are also illuminated by the ringlight, and produce diffuse reflection back to the lens.

With this fixed diameter "hole", it's clear that the "most axial" position is close to the lens as expected. At f4 the 14mm hole defocuses significantly, producing a fairly uniform direct reflection illumination across the coin. There is no vignetting at f4, and only the far corners are vignetted at f16, so the hole can be reduced further.

10mm RL / 90mm CR f4 Mirror

10mm RL / 90mm CR f4 Coin

10mm RL / 90mm CR f16 Mirror

10mm RL / 90MM CR f16 Coin

35mm RL / 65mm CR f4 Mirror

35mm RL / 65mm CR f4 Coin

35mm RL / 65mm CR f16 Mirror

35mm RL / 65MM CR f16 Coin

60mm RL / 40mm CR f4 Mirror

60mm RL / 40mm CR f4 Coin

60mm RL / 40mm CR f16 Mirror

60mm RL / 40MM CR f16 Coin

[kaleun96]

Great tests! The mirror idea is a good thought to test for pseudo-axial coverage, I hadn't considered that before, even when using a mirror to try and achieve perfect background separation. It's also very interesting to see just how small you can make the "hole" without vignetting, as well as how the lens aperture affects the focus of the reflection. If you're up for it, I'd be interested to see what the minimum "hole" diameter without vignetting is given the distance from the lens for a fixed aperture (e.g. f4). In my testing by trial and error (and as your results show), a matter of a few mm doesn't seem to make a huge difference but it is something I can still better optimise.

Out of curiosity, what ringlight are you using? Is it just a standard 30mm LED COB or similar? I've used the Laowa ringlight in the past with a lot of success, it's plenty bright but gets very hot and doesn't have any in-built brightness control.

If it's of any interest, I finally got around to posting a second blog post on my custom flash adapter shown above. It just details some of the problems I was having with my original adapter (though see part 1 for more on that) and some tests of various prototypes I tried. It doesn't touch on the vignetting problem of this topic and is very non-scientific but it shows some actual pictures of the prototypes rather than a cross-sectional drawing. I still have to write part 3, which sees me end up on the final design that I attached in my previous message above.

[ray_parkhurst]

If you're up for it, I'd be interested to see what the minimum "hole" diameter without vignetting is given the distance from the lens for a fixed aperture (e.g. f4).

In my tests a couple days ago, I placed a variable iris on the filter threads, and was able to bring the iris hole down to 2mm without vignetting. 2mm was the smallest it could go. In that test I didn't have a diffuser or ringlight so could not do any axial lighting tests, just vignetting. I didn't vary the distance since the iris was not easy to work with.

Out of curiosity, what ringlight are you using? Is it just a standard 30mm LED COB or similar?

Yes, just a cheap 30mm from Aliexpress. I think it was this seller:

https://www.aliexpress.us/item/3256805702729664.html

They also have a "20mm" with 17mm OD and 7mm ID.

[rjlittlefield]

...was able to bring the iris hole down to 2mm without vignetting. 2mm was the smallest it could go

Just being sure the issues are clear...

Vignetting happens when two apertures fight with each other, with each one blocking some rays that would otherwise get through the other and help to form the image.

If the front aperture is very large, then it blocks nothing except what the rear aperture will anyway, so you get no vignetting.

If the front aperture is very small and the rear is large, then the rear blocks nothing that got through the front and again you get no vignetting.

However, in that second case the front aperture also determines the working f-number of the optics, so if you make it too small you also get hit by diffraction. That 2 mm hole, in front of a 100 mm lens extended by another 100 mm to give 1:1, would give you something like effective f/100.

--Rik

[ray_parkhurst]

...was able to bring the iris hole down to 2mm without vignetting. 2mm was the smallest it could go

Just being sure the issues are clear...

Vignetting happens when two apertures fight with each other, with each one blocking some rays that would otherwise get through the other and help to form the image.

If the front aperture is very large, then it blocks nothing except what the rear aperture will anyway, so you get no vignetting.

If the front aperture is very small and the rear is large, then the rear blocks nothing that got through the front and again you get no vignetting.

However, in that second case the front aperture also determines the working f-number of the optics, so if you make it too small you also get hit by diffraction. That 2 mm hole, in front of a 100 mm lens extended by another 100 mm to give 1:1, would give you something like effective f/100.

--Rik

Yes, the competing apertures discussion makes sense. Based on the effects to the image I had concluded the 2mm hole acts as the limiting aperture when the lens is at f4. But by f11 the 2mm hole showed vignetting, so I assumed the f11 lens aperture was limiting at ~EA22. This is much larger than the EA100 you describe. Why is this the case? For sure the 2mm hole was some distance in front of the lens, mounting slightly in front of the front of the filter threads. The answer is probably not important to the application since 2mm is hopefully much smaller than needed to "close the hole" with sufficiently large lens aperture plus close mounting to the front of the lens.

Edited to add: as I said it is probably not important to the application, but just to understand better...when the hole is placed close to the subject, it is purely a hole, with almost zero effect on optical properties of the image except for the vignetting. When it is close to the lens, it can act as a limiting aperture and won't show vignetting at all given large enough lens aperture. So I'd conclude that the distance from the front of the lens will be a significant factor in determining the effective aperture of the hole, correct? I don't know how to calculate this.

[rjlittlefield]

Based on the effects to the image I had concluded the 2mm hole acts as the limiting aperture when the lens is at f4. But by f11 the 2mm hole showed vignetting, so I assumed the f11 lens aperture was limiting at ~EA22. This is much larger than the EA100 you describe. Why is this the case?

It's because EA100 describes the width of each ray cone, while your f11 lens aperture describes the distance off-axis where peripheral ray cones start to get blocked. The cone for center of image would not start to get blocked until a much narrower lens setting, at which point most of the periphery would have gone black.

the distance from the front of the lens will be a significant factor in determining the effective aperture of the hole, correct? I don't know how to calculate this.

There's a trick that makes calculating this simple. First, you need to know that the effective f-number on the rear (image side) is always equal to magnification times the effective f-number on the front (object side). Then, consider that the effective f-number on the object side is just equal to the distance from hole to object, divided by the hole diameter. Put those together to get

• Feff_sensor = magnification * distance_from_hole_to_object / diameter_of_hole

For example, magnification 1 with a 2 mm hole located 200 mm from the object, gives Feff_sensor = 1*200/2 = f/100.

Now suppose you drop the magnification to 0.5, still using a 100 mm lens. That gives you longer working distance, nominally 300 mm in front and 150 mm behind. It also gives you more choice of where to put the hole. If you leave it as a 2 mm hole at 200 mm from the object, your Feff_subject remains f/100 (=200/2), while the reduced magnification widens your Feff_sensor to be f/50. If you move the same hole to be at 300 mm from subject, then Feff_subject narrows to be f/150, with corresponding Feff_sensor = f/75. Or if you widen the hole to be 3 mm while moving it to be 300 mm from subject, then Feff_subject is back to f/100 with Feff_sensor f/50. For vignetting, the setup with hole at 200 mm from object and thus 100 mm from lens would be much worse.

--Rik

[ray_parkhurst]

Rik...thank you for the explanation. I need to study it a bit before it fully makes sense to me, and then see how it affects the application.

Unfortunately, practicality limits the range of hole dimensions, distance from the lens, etc. I ended up building a small illuminator that threads onto the front of the M40.5 filter mount on the 75ARD1. The construction started with a M39-RMS adapter. This provides a good base / heatsink for mounting the 30mm ringlight, which is JB-Welded to the adapter. The adapter is in turn JB-Welded to a M40.5 UV filter with filter removed. This gives it proper mounting to the lens. I inserted a small piece of cardboard cone through the adapter and ringlight. The cone extends ~7mm in front of the ringlight. The hole in the cone is 12mm, a bit smaller than what I had used before. A donut / ring of diffusion material is glued to the front of the cone. I made the diffusion ring 32mm diameter to block any direct light hitting the subject. Overall distance from filter ring to diffuser is 18mm. I could make it a little shorter at expense of the hole diameter.

Here are the results:

Donut Diffuser reflection in mirror:

Coin Image:

My stage is not quite flat vs the sensor, so the diffuser is offset a bit, but it's not a huge problem. The coin image shows the excellent contrast between the field and edges of the features that axial lighting is prized for, giving a nice impression of 3D feature height. This coin is not highly reflective so is a good subject for axial lighting. More reflective coins will show a much brighter field, and the resulting shorter exposure time will darken the edges of the features, often to black. A coin like this one, and similarly most ancient coins, are the best subjects for this type of lighting.

Edited to add: I went ahead and shot the same coin using the backlighting system I built for ancient coins. In the process of doing so, I realized a simpler approach to the edge illumination problem due to backlighting...shoot the coin on a mirror! This eliminates the calibration and color matching issues completely, and also eliminates any side-lighting issues since the axial lighting does not have any significant angle vs the mirror. I will be looking into this further!

Edit #2: I am replacing the 32mm diameter circular diffuser with a 50mm square one. This should push the direct reflection well outside the radius of a smaller coin, though I will need to test it on a larger coin to ensure a fully bright field. Also realized this system "should" be self-calibrating. The coin surface will always reflect less than the mirror, so it should be easy to ensure a fully-bright field while simultaneously minimizing blown highlights on the coin surface.

[kaleun96]

Edited to add: I went ahead and shot the same coin using the backlighting system I built for ancient coins. In the process of doing so, I realized a simpler approach to the edge illumination problem due to backlighting...shoot the coin on a mirror! This eliminates the calibration and color matching issues completely, and also eliminates any side-lighting issues since the axial lighting does not have any significant angle vs the mirror. I will be looking into this further!

The mirror approach definitely works well but there are a few limitations, or at least some issues I ran into when trying the method. I briefly touched on them in this thread:
"My current method, using a custom 3D printed "ring light" that holds two Godox TT350 flashes, is to either use a white bit of plastic or a mirror as the background. The mirror is great and results in a perfectly white background as it reflects the white plastic from my "ring light" (which is attached to the lens) back to the lens and separates the coin from the background perfectly. There are a few limitations with this, however: I can't tilt the coin; the mirror tends to produce a slight "halo" around the edge ,which appears as a purply pixelated area on the very edges, only a few pixels wide, in the focus stack; and I think the mirror is perhaps diffusing the image slightly from stray reflections (or I need better flocking)."

The haloing issue I've mentioned before (I think in your thread about backlighting) and Rik had previously shared some tips to help deal with it. Recently I tried Helicon Focus to see how much faster it stacks with the GPU acceleration and the default settings for "Method B" largely fixed the haloing issue. It's probably a case of me needing to further tweak the settings in Zerene but just wanted to mention in case you run into it too. You can see an example of the yellow-green haloing in the examples in this post.

But the main reason I moved away from the mirror is that my tilt setup would tilt the entire coin platform rather than just the coin, and I had the mirror mounted to the platform so tilting the coin would tilt the mirror and render it useless for background separation in my case.