I think it's a nice start. As materials and tech get more precise, I see a good future of people making their own lenses, at least for general hobby purposes.
People (hobbyist astronomers) make their own lenses (well, mirrors) starting from glass blanks and tediously grinding and polishing them to precise shapes. However, it's quite challenging and time consuming and you'd only do it if you absolutely needed a mirror of a specific geometry. And it's quite hard to reach the levels of accuracy that a routine mirror or lens making operation can.
When I was a young kid, my father ground a 6in mirror with a 48in focal length. I remember him walking around a table at night for weeks. The first mirror didn't work out so he had to grind a second.
The resulting telescope was, and still is, a remarkable device, although we had to have the mirror re-silvered a few years ago.
The first time I saw Saturn in all its glory was through that telescope, looking the size of a softball, and I remember thinking, Wow, it really exists!
Sure, but these days you can buy most any lens or mirror you need rather than grinding it yourself, and then spend more time viewing stars and planets and less time breathing glass particulates.
I think in most cases making your own lenses is just a labor of love, not a way to save money or achieve better results. This is merely because lens and mirror making was industrialized and has impressive economies of scale.
This was back in the early 60s and although he could probably have purchased a ground mirror, I think getting one with a 48in focal length may have been beyond his budget.
This is super interesting, I wander if 3d printing lenses could be a good way to explore light field photography and projection?
You could potentially design and print some quite complicated lenses. I suppose the one difficulty would be the epoxy pooling in small crevices when dipping. But then if you take that into account in the design it could be compensated for. You would need to find a way to predict the flow of the epoxy over the printed substrate.
Quite true, alternatively you could hold the part inside a rotating drum, a bit like rotational moulding. But then it will be harder to predict the thickness.
I suspect you are right about spraying, much more likely to get a uniform thickness. Maybe a combination of the two spray a constantly rotating (in 2 axis) part.
I think this article was mostly a summary of the engineer’s personal blog post: http://amosdudley.com/weblog/SLO-Camera - linked from the post with example photos in another comment.
One problem that remains for home 3D printed lenses is chromatic aberration.
Refractive index(angular multiplier for lights entering at angles) is wavelength and material dependent, which means magnifications/focal distances vary for each colors, causing rainbow-blurred images at image planes. btw this is obviously how prisms work.
The variance is largely monotonic and the effect is well studied; the coefficient of this aberration is called Abbe number or Vd, defined as Vd = (nD-1)/(nF-nC) where nD, nF, nC is indices at 656.3, 589.3, 486.1nm respectively. Materials that shows Vd > 50 is called "crown" glasses and materials with Vd < 50 is called "flint" glasses. There are lens materials with Vd of exactly 50, such as Fluorite(CaF2) crystals as well as other engineered materials such as borosillicate "Extra-low Dispersion(ED)" glasses. These lenses show abnormally low aberrations, at cost.
Wikipedia[3] explains the legendary Carl-Zeiss Tessar design by Paul Rudolph as follows:
> A Tessar comprises four elements in three groups, one positive crown glass element at the front, one negative flint glass element at the center and a negative plano-concave flint glass element cemented with a positive convex crown glass element at the rear.
Without relying purely on such exotic materials, lenses are designed with pair and triple elements(doublets and triplets) to counter aberration. A convex crown glass may be used to focus light, then convex flint glass lens are added so as to cause focal point for both red and blue lights to coincide(achromat, or non-chromatic, doublet). Or an another convex lens can be added to force RGB lights to converge at a same point(apochromat triplet, APO). Focus errors dependent on wavelengths follow an exponential curve to the power of lens count. An achromatic doublet consists of Schott BK7 and Schott F2 is often used in textbook exercise questions.
There are other types of aberration such as "the five Seidel aberrations", and some might be able to be solved by clever use of same materials, but as far as I understand, chromatic aberration cannot be solved by computational force alone. Maybe it's something dumb and obvious to materials scientists as adding certain baby powder with specific composition into the resin, like how obvious a UART console pads are to PCB designers, but for now this rules out high quality lens manufacturing by direct stereolithography, as there is no hobbyist known way of controlling and modifying Abbe number for UV resin materials, save for using metamaterial surfaces. If someone could dump that here it'll probably be a decade worth of progress, for both hobbyist photography lens creation AND likely for entire optics industries as well!
You can buy a wide range of UV resin materials (with a range of refractive index) as a hobbyist.
I can think of any number of alternative ways to deal with this, for example you can project a series of colored images and measure the response, then use that to back-correct images. We're in an era of fast computers and powerful software, which can correct for things that are normally corrected by expensive techniques.
Refractive index isn't a problem, but the problem is the index gradient over wavelengths(Abbe number). VR goggles and some digital cameras does software correction but that is unheard of, in, e.g. film photography.
They seem to be a bit confused with optics. What they minimise by reducing the aperture size is the spherical aberration, if they used a spherical lens as pictured.
Coma is present on parabolic mirrors, which do not exhibit spherical aberration or any shape related aberration when the rays are parallel to the boresight of the camera, but only when at an angle.
WOW! I hope this technology will help us resurrecting DVD/BR|BD-R drives! If we could make lenses to replace old blurred ones... That'd be a life saver!
PS: The page so heavily depends on JS that it literally gives me a blank page unless I disable JS in browser :D :(
30 comments
[ 3.1 ms ] story [ 81.9 ms ] threadThe resulting telescope was, and still is, a remarkable device, although we had to have the mirror re-silvered a few years ago.
The first time I saw Saturn in all its glory was through that telescope, looking the size of a softball, and I remember thinking, Wow, it really exists!
I think in most cases making your own lenses is just a labor of love, not a way to save money or achieve better results. This is merely because lens and mirror making was industrialized and has impressive economies of scale.
You could potentially design and print some quite complicated lenses. I suppose the one difficulty would be the epoxy pooling in small crevices when dipping. But then if you take that into account in the design it could be compensated for. You would need to find a way to predict the flow of the epoxy over the printed substrate.
But perhaps, there is a way of holding the concave part upside down and spraying epoxy on it that might work. No idea.
I suspect you are right about spraying, much more likely to get a uniform thickness. Maybe a combination of the two spray a constantly rotating (in 2 axis) part.
Wow!
I bet you could print those mustache-shaped cross-section lens that correct chromatic aberration? (I can't find the paper at the mo'.) :(
https://www.researchgate.net/figure/Spherical-aberration-cor...
Found it: "General formula for bi-aspheric singlet lens design free of spherical aberration" https://opg.optica.org/ao/abstract.cfm?uri=ao-57-31-9341
https://petapixel.com/2019/07/05/goodbye-aberration-physicis...
https://phys.org/news/2019-08-physicists-year-old-optical-pr...
https://news.ycombinator.com/item?id=20369960 and https://news.ycombinator.com/item?id=20703140
> The lens was clear as a magnifying glass, but as a focusing lens it wasn’t accurate enough.
Refractive index(angular multiplier for lights entering at angles) is wavelength and material dependent, which means magnifications/focal distances vary for each colors, causing rainbow-blurred images at image planes. btw this is obviously how prisms work.
The variance is largely monotonic and the effect is well studied; the coefficient of this aberration is called Abbe number or Vd, defined as Vd = (nD-1)/(nF-nC) where nD, nF, nC is indices at 656.3, 589.3, 486.1nm respectively. Materials that shows Vd > 50 is called "crown" glasses and materials with Vd < 50 is called "flint" glasses. There are lens materials with Vd of exactly 50, such as Fluorite(CaF2) crystals as well as other engineered materials such as borosillicate "Extra-low Dispersion(ED)" glasses. These lenses show abnormally low aberrations, at cost.
Wikipedia[3] explains the legendary Carl-Zeiss Tessar design by Paul Rudolph as follows:
> A Tessar comprises four elements in three groups, one positive crown glass element at the front, one negative flint glass element at the center and a negative plano-concave flint glass element cemented with a positive convex crown glass element at the rear.
Without relying purely on such exotic materials, lenses are designed with pair and triple elements(doublets and triplets) to counter aberration. A convex crown glass may be used to focus light, then convex flint glass lens are added so as to cause focal point for both red and blue lights to coincide(achromat, or non-chromatic, doublet). Or an another convex lens can be added to force RGB lights to converge at a same point(apochromat triplet, APO). Focus errors dependent on wavelengths follow an exponential curve to the power of lens count. An achromatic doublet consists of Schott BK7 and Schott F2 is often used in textbook exercise questions.
There are other types of aberration such as "the five Seidel aberrations", and some might be able to be solved by clever use of same materials, but as far as I understand, chromatic aberration cannot be solved by computational force alone. Maybe it's something dumb and obvious to materials scientists as adding certain baby powder with specific composition into the resin, like how obvious a UART console pads are to PCB designers, but for now this rules out high quality lens manufacturing by direct stereolithography, as there is no hobbyist known way of controlling and modifying Abbe number for UV resin materials, save for using metamaterial surfaces. If someone could dump that here it'll probably be a decade worth of progress, for both hobbyist photography lens creation AND likely for entire optics industries as well!
1: https://en.wikipedia.org/wiki/Abbe_number 2: https://en.wikipedia.org/wiki/File:Comparison_chromatic_focu... 3: https://en.wikipedia.org/wiki/Tessar
I can think of any number of alternative ways to deal with this, for example you can project a series of colored images and measure the response, then use that to back-correct images. We're in an era of fast computers and powerful software, which can correct for things that are normally corrected by expensive techniques.
http://amosdudley.com/weblog/SLO-Camera
Coma is present on parabolic mirrors, which do not exhibit spherical aberration or any shape related aberration when the rays are parallel to the boresight of the camera, but only when at an angle.
They really changed the game with their SLS. We bought it last year and it performs much better then an EOS device that was 5x more expensive…
Hoping to see similar effects with this line…
PS: The page so heavily depends on JS that it literally gives me a blank page unless I disable JS in browser :D :(