The sample images show significant vignetting and spherical aberration. The focal plane is curved and clearly not flat over the field of view of the image. There is indeed not obvious chromatic aberration, but I suspect that it is because other issues are significant – there is no mention of the lens's MTF/OTF or any significant tests.
One of the other commenters has highlighted that the lenses in question are part of this telescope kit – https://www.surplusshed.com/pages/item/T1686.html – which is indeed consistent with a reasonably good on-axis or paraxial design but with poor off-axis performance (like most telescopes). I presume these are two apochromatic doublets [1] but it's a bit hard to see.
I wonder how well software could fix that? Could one of the hack firmwares be modified to correct the lens in camera? Seems like software correction would open a lot of possibilities for DIY lenses.
I don't think this is correct. The camera is applying a PSF (point spread function) to the image. If you have a known target image you can find the inverse of the PSF and apply that as a correction (deconvolution).
Spherical aberration is caused by the focus deviating across the image, there being only one point in the centre where the image can truly be in focus. So although the image can be improved in the way you describe, you can’t recover the information that is out of focus.
There's a bit of literature out there claiming the opposite. Certainly if you can take multiple Z images you can correct it that way, but there are methods which work directly to de-blur (aka, in focus) an image.
I know next to nothing about how sensors are manufactured, but we do have the technology to build flexible displays. So what are the major challenges in building "flexible" (for lack of a better word, obviously they would be fixed in place) sensors?
> The only parts of the lens that can’t be 3D printed are the glass optics, and for this Steele used Surplus Shed to acquire achromatic (two-part) lenses. He says that eBay is another option, but to make this particular lens builders will need one with a diameter of about 65mm.
> The only parts of the lens that can’t be 3D printed are the
> glass optics, and for this Steele used Surplus Shed to
> acquire achromatic (two-part) lenses.
The issue is that while you can print 3d lenses (fun read: https://formlabs.com/blog/creating-camera-lenses-with-stereo...), it's just not practical unless you're super into the project (similar for telescope makers who grind their own lenses). The long and short of it is that high quality glass is an industrial process that has hundreds of years of refinement to give you excellent image quality at very low prices.
That said, I have cnc'd a piece of acrylic, polished it, then used it to make a silicone mold and used that to make an epoxy lens.
In photography, a "lens" generally refers to the mechanical component that houses the optical lens.
In 3D printing, "print" generally refers to "most of the device, except those components that one versed in the field would recognized as being composed of materials not suitable for printing".
Though I do agree with you, many people who do 3D printing would not find the title misleading.
Nope, a lens holder is the mechanical component. Most people and people in industry would refer to either the optical element(s) or the complete assembled optic as the lens.
You'll see not a single glass lens by itself, but rather every image will be of an assembly containing optical lens intended for mounting to a camera.
Also, enter any photography store and ask to see the lenses. You'll not be presented with discrete glass, but rather with assemblies containing glass intended for mounting to a camera.
Perhaps in the eyeglass industry the term "lens" refers to the glass itself as a discrete component, but not in photography.
In the imaging industry the lens assembly is normally called an objective. In every field I've worked in, the lens is the glass and nothing more. If photography uses this term, it's wrong, but if it's established... who cares, just know that fact and move on.
It makes sense in photography where we don't really deal with or think about individual lens elements aside from people who repair their own, or when some new kind of element appears. We also call the whole assembly glass sometimes, though some people think it's pretentious. It's more common in blogs/YouTube/etc than in actual practice where focal lengths and f-stops are usually the main language of lenses.
I have a 3d printer and thoroughly enjoy this type of thing but this title is a bit misleading. It is likely you’ll spend more than $15 finding the right lenses plus the cost of filament.
Super cool project, and they did a really great job documenting the process.
That said, a hobbyist level CNC will not be able to make a functional lens. Tolerances are way too high. Maybe it'll be good enough for focusing light and burning holes in paper, but it's not going to be remotely close to photography level.
I really enjoy my 3D printer but the number of real use cases is fairly low in terms of day-to-day utility. This is exactly the kind of thing that makes me excited about the future of the hobby.
Use cases may be low, but when you find "the one" it's just nothing short of epic. For me it was my Sim Rig... needed a 3d printer to make the carrier blocks for the actuators, which then triggered a whole rabbit hole of designing & building a (slightly over the top) simulator*
The amount of stuff I've printed with it, km's and km's of filament and the little beast still keeps squirting out molten plastic. Last time I looked at the stats, 91 days print time, 6355m of filament.
One of the best tool purchases in the shed ever. Really stoked with it.
Nice! I have a few 3D printed components for my sim racing setup, but it's only a few brackets and a custom face part for my VR headset. I've always done the 3D printing by uploading the design and having it delivered. That uses to be easily in the $ 10 per part kind of price range but recently there are higher minimum order prices (50 or 75 or so) everywhere.
What kind of printer do you have and what did it cost?
I've had a lot of fun with my printer. Some days it runs almost 24/7 and then I don't use it for a while. Recently, I wanted to make a sliding barn door. Spent some time looking at the products online, then designed the same thing in Fusion 360, and after a few test prints, I had all the parts to make the door.
Similarly, all my shelf brackets are 3d printed now. THe thing that holds up my custom LED light is 3d printed. Some of the art I have around the house is printed...
After 3D printing for years I started teaching myself how to use FreeCAD. It’s not the most newbie-friendly software but being able to design your own parts is incredibly powerful. I design and print parts for the most mundane tasks.
These days I find designing a part relaxing, similar to the feeling of writing code for a personal project.
Apparently Fusion 360 is also free for hobbyists, I’m guessing that it’s a bit easier to learn than FreeCAD.
(I realize that designing your own parts isn’t practical for most folks but I think HN is a good audience for this.)
Came here to say the same, except so far I’ve been using Blender with the CAD Transforms plugin (mainly for the Sketch-like snapping abilities). I watched a few videos on Blender for Precision Modeling (search youtube) and started playing around.
I’ve been meaning to learn FreeCAD or one of the others because the parametric modeling seems useful, but Blender has been good enough for my purposes so far - mostly adapters for tool vacuum ports, and other little things around the house and workshop.
I moved from FreeCAD to Fusion360 (I pay for a pro license) and the difference was astounding. Once I grokked the fusion360 model (which is both easier and more complex than FreeCAD) I was able to make forward progress very quickly. There are a lot of subtle little touches in the Fusion360 UI that make design easy.
Well, someone has to create the models. If You're lucky and there's plenty of good modellers in Your use case / hobby, good for You. Have fun printing.
What the OP however refers to is the "rapid prototyping" that 3D printers unlocked for masses. If You learn even some basic modelling (e.g. few primitives in OpenSCAD for programmers), it's trivial to do test fits and whatever gigs You need.
I haven't realized it myself until some months ago I needed a small funnel for a bottle. After some thinking I realized it's much easier/faster to throw together a difference of two cones and two cylinders in OpenSCAD in about 10min and print it in maybe another 20 than spend the time looking around stores to find some close compromise.
I think it's more like owning a computer without being able to use a text editor or a spreadsheet program. You can open files you get from other people but you can't make changes or generate anything original. Programming in this analogy would be modifying your printer itself.
Obviously there have been some successful computers where users don't modify their files, like mp3 players, but those are much cheaper and more user friendly than general purpose computers.
I suppose, I just don't use it that way. Have never printed something that I downloaded except for testing purposes. I use the printer to make stuff that I need around the house, like a bracket to hold my electric toothbrush charger to the towel bar, etc...
I check for well-designed looking ones first though, if it's not extremely custom (like the razor holder I designed to fit my specific razor over the side of my specific mug-like bathroom holder thing).
I found a nice looking bracket shaped specifically for my Bamix stick blender that someone else had designed, saved me some trial and error and time and filament.
I found someone else's design for a Makita LXT battery fitting that I used as the base to prototype the inverse: I wanted a dummy battery shape that I could mount to the wall in order to hang my vacuum cleaner from it.
I'm glad I can do some basic CAD / am willing to learn, but I've definitely got some use out of it with only a slicer. And then there are the people who only want trinkets anyway.
Those two projects made me realize that 3d printing can be a full time fun hobby for exploring all those things it would be too hard to machine in a garage.
I've been enjoying using OpenSCAD whenever I can (instead of Fusion 360). The language is a bit janky but it's great for hacking something up in a purely procedural manner, and using Git.
OpenSCAD is fine for what it does and I do use it a lot for very simple stuff. ImplicitCAD is another interesting project in that space that has fillet unions.
FreeCAD would be great if the Interface wouldn't be so janky. I give it a try every now and then, but never really get over the re-learning aspect.
So, I am still stuck with Fusion360 running in a VM on my otherwise FLOSS-only system :-/
pretty misleading title. the bokeh is kinda outta whack, that grass pic is pretty terrible...but a couple of the other example shots are appealing in my opinion. it's too bad the whole premise of the article is kind of dishonest.
Now that might be a bit unusual choice of lens, but news photographers around the world rely on their zoom lenses that won’t open up more than f2.8.
And for a prime lens in this long focal length, f2.5 is still quite okay, Canon’s EF135mm opens up until 2 that’s only 2/3rd of a stop faster.
Really fast lenses get attention because they are the most expensive, and shallow depth-of-field has been in fashion… I think in part because it gave photographers a way to distinguish themselves from smartphone wielding amateurs. I predict this is on the way out now that portrait mode fakes this look! But a super fast lens is in no way necessary for taking great pictures.
I’m sure there’s plenty of valid criticisms for this project but this one does not convince me :) Personally I wonder more from an ecological point of view, i.e. why use all these resources when there’s plenty of unused lenses on the second hand market. But then again the Open Source aspect is commendable.
I second this. For this focal distance, vintage lens tend offer around f/3.5 - f/4. Compared to that, this is pretty impressive for DIY and just two lens elements. Might be more lightweight too, which makes it suitable for street photography.
Everything is a compromise. This optical design manages to be f/2.5 most probably because it does nothing to correct the plenty of aberration and defects it produces. Also f-numbers are deceptive regarding brightness and we should see how much light actually comes out from the other side, i.e. calculating its t-stops.
Also, a manual 165mm lens on full frame? I'm sure it could be great for interesting portraits, but street photography?
Once you stop caring about image quality, the f-stop of a lens is only fundamentally constrained by the physical diameter and diopter of the lens. I suspect the f/2.5 figure is only being held back by the size of the lens barrel and aperture mechanism.
If it were actually well-corrected and apochromatic, then a 165mm f/2.5 lens would be fantastic for astrophotography. There's a reason the manual-focus Rokinon 135mm f/2 lens is so popular for that purpose.
And there is a reason why theanual focus Nikon 200 f/2 is, used, at over 2k Euro. Good professional lenses are expensive for a reason. Ehich is the reason why I stick to used ones for DSLRs, those are finally getting cheaper and easily available on the used market now that a lot of people switch to mirrorless and dedicated mirrorless lenses are covering more and more focal length and zoom ranges.
Too long focal length. For street photography in a budget a Nikkor 50 mm f/1.8D is used in excelent condition under 100 bucks. Fast lens, tack sharp, great in low light and small so you do not stand out among the people usibn their phones.
It's only hard to make a fast lens when you also have a narrow lens barrel or a strict internal quality standards to satisfy. In absence of those, an aperture disc is just a disc that can be drilled, and 50[mm]/3.5 = 12.5mm or 0.5" is a nice round number around the context of a 135 format camera.
Legend has it that some of the disposable cameras had curved film planes to, uh, take advantage of the abundant spherical aberration and astigmatism. I'm sure this lens need something like that too.
No. The "depth of field" for a lens parallel to the sensor is a space between two planes, so you'd expect it to be clear all the way across the image but blur closer and further from the sensor (up and down the image). This blurs out to all the sides, it's due to spherical abberation from inappropriate glass (more suited to telescopes where only the center of the image circle matters).
Amongst other things, the f number affects the depth of field (how much of the image is in focus.
A low number (f/1.6 is fairly low. f22 is fairly high) means that only the object you focused on will be sharp and anything nearer or further away will become blurry. This is great for portaits for example as it draws attention to the person.
(f is the aperture and also affects the amount of light. You can shoot with a lower shutter speed if you have a wide aperture/low f number)
The "mm" number is the amount you're zoomed in. Wide is a low number, telephoto from a distance is a high number.
But blur can come from other things. Cheap lenses, unsteady hands or moving subjects. Depth of field is just one type of blur.
3d printing is amazing but it is not extremely precise. At the end of the day you are extruding plastic from a stepper motor controlled head not much different than extruding glue from a glue gun (I know about other mechanisms but this was done with an extrusion printer) . You have many factors at play such as the acceleration rate of the heads, the extrusion rate and how it ramps off at the end, how true your axis are, mechanical play... it is endless how much you can tweak.
And at the end of all of that you are not going to make anything that has micron or even thousandths level precision. For something like a lens, machined metal and at a minimum injection moulding is needed. It would be as if I said I could produce a gauge block to measure 10mm for a fraction of the cost. But for that cost savings my precision will be unusable when my block is not flat and is 9.956 mm which in 3d printing would be very close but in machinist land is off by a mile. Lens casings and aliignment need similar precision.
when I was a kid I remember buying lenses from Edmund Scientific and making telescopes, microscopes and similar things by cementing lenses into cardboard tubes. The results weren't great but they were workable.
nobody makes perfect lenses as part of the initial manufacturing method. Lenses are always smoothed/polished. In the 3d printer case (hobbyist) you would 3d print a solid object oversized, then sand/polish it down.
The automated microscopes I build are mostly 3d printed parts. The precision doesn't matter much- I always have to add additional hardware to adjust the precise positioning (typically screws going through 3d printed threads).
But you are right there are definitely other methods which work faster and cheaper and ten thou is a typical target for a good machinist.
Very very cool. If the author or inventor come across this comment, it would have been great to see some photographs of faces. I try to only focus (!) on photographing people, these days.
The inventor, Steele, describes a cardboard jig for calibrating the parametric design. It sounds almost as important as the final lens casing itself.
I believe there are lots of websites where you can upload your CAD files and get sent the finished product. You can choose colours, finish options, materials... Certainly very good to get started.
I noticed that too. If you measure the MTF, it'll probably drop off quickly. Two lenses doesn't seem enough to correct aberrations.
Still, pretty impressive for a DIY project. The lack of clarity might be taken positively as an interesting lens characteristics; might even attract some camera enthusiasts to use this.
Yup that's the only reason I opened the article to read. While obviously the most important thing in a lens is the optical element(s) it's a bit disappointment.
I think the title should read "print a housing for $15 glass to make a 164mm lens" or something like that to indicate that it's not the actual lens elements that they're printing, but of course it wouldn't be clickbait and many people including me wouldn't have clicked on it.
Prusa did a whole article about 3D printing optics some time ago[0]. I've also had good luck coating transparent filaments with epoxy resin to get clear parts.
Was also a video[1] over on the Breaking Taps channel about casting clear epoxy resin to make your own lenses. Results were not perfect but good enough for some uses. Of course, that requires some actual lens elements to make the molds from, but it's something.
That's definitely an idea worth trying, though it was mentioned elsewhere on this post that you can 3D print and polish your masters for casting rather than relying on having the lens already.
> I've also had good luck coating transparent filaments with epoxy resin to get clear parts.
That might work if you only care about "clear", but unless you exactly match the optical properties of the resin and (printed) filament, it will be pretty useless as an optical device.
True, also the evenness of the surface is a little off without grinding which also limits the usefulness for optics. I just used it a few times when I needed an odd cylinder size that needed to be see through.
I already expected that to be the case, so I had realistic hopes. The tech for 3d printing good lenses is still within the realm of futuristic companies/researchers:
https://www.luxexcel.com/the-luxexcel-technology/
It's surprisingly easy to 3D print the shape of a lens and polish by hand. I made a working telescope this way with a 3D-printed body and two cast epoxy lenses. As an added bonus, you can easily make aspherical lenses this way.
If you coat the surface with silver you can also make mirrors, but the polishing requirements for this are a lot more difficult. My attempts at a 3D-printed telescope mirror have very bad image quality compared to the excellent results from the lenses.
you may be able to 3d print a mirror telescope-style "lens". But you'd probably need to somehow "polish" down the plastic and coat it. Maybe a bit of acetone and then depositing silver on it?
I made several attempts at acetone vapor smoothing on mirrors printed in ABS and ASA and the surfaces always came out very well polished at the micrometer scale but unacceptably warped, wrinkled, or patterned at the millimeter scale. The best finish I've gotten on a 3D-printed mirror was from sanding and polishing with a cotton ball soaked in cerium oxide slurry while the mirror blank spins on a pottery wheel.
Chemically depositing silver with the old-fashioned immersion setup is also very tricky as the reaction is quite temperamental. I recommend the two-part spray process if you want to try silvering a telescope mirror.
It's still ultimately a silver nitrate reduction reaction, but it's extremely consistent. I tried reproducing the classic silver nitrate/glucose/sodium hydroxide/ammonia silvering baths astronomers used before the modern switch to aluminum vacuum coatings, and my success rate was only about one in five attempts. It's much more difficult to execute than the typical silver nitrate demonstrations where someone silvers the inside of a glass flask because you need the silver to produce a perfectly even layer on the outside surface of the glass for a telescope. The spray reaction has worked for me every time.
Dumb question. I've been wanting to get into IR optics for a while. Does every camera lens come with IR and IR filter, and would this one somehow not? It seems IR optics are insanely expensive for some reason.
there are lenses optimized for near ir 700-1000nm. above 1000nm you'll need InGaAs sensors that are so expensive that the lens will be a secondary problem
What wavelength range? Silicon detectors cut off at about 1000 nm, glass optics start cutting off around 2500 nm. Then you get into exotic materials that are expensive.
If you want to work with 5000 nm - 10,000 nm in that price range, you can use an affordable thermal camera with a microbolometer sensor. Surplus silicon and germanium lenses can sometimes be found cheap on eBay, and there are always a lot of very low priced zinc selenide lenses for sale (intended for the CO2 laser cutter market).
Actually it is the other way around. The sample pictures show you can build a cheap and bad lens easily. But to achieve the optical quality of professional lenses is an entirely different thing. That makes you quite appreciative of them and shows how well spent the money for them is.
You can buy a ton of ok lenses in the $100-300 range, as there are several vendors in that range from Asia. Some of them can be really a steal. But overall you get the quality you pay for. Look what the lenses of the big camera companies offer, both optically and from features like AF and IS, and you see why 3D printing won't just turn the market upside down.
As an enthusiast, I find the tinkering possibilities opened up by 3D printing highly interesting though.
I don't see why the ability for a hobbyist with hobbyist equipment to assemble a very particular lens with a ton of caveats should have any impact at all on the prices of a full consumer product with a vast amount of benefits and features.
Which is not to say that lenses are not overpriced, but the fact that you can theoretically 3d print your own... anything (gun? Car parts? Doorknobs?) doesn't imply anything can or should happen to the price of a much more functionally complete product.
Indeed, most of those things you listed don't make sense to 3D print because they're mass produced which is cheaper with injection molding. I'm sure you can also buy a cheap low quality mass produced camera lens too. 3D printing's advantage is in unique things.
Just thinking of getting the focal plane of a DIY lens pan on the sensor plane requires more precision manufacturing then is available for the DIY crowd. A relescooe, sure. And even then you would look at sub-pat optical results. A camera lense? Fun project for sure, but nowhere near eveb the cheapest lebses on the market. Let alone good condition vintage lenses from the prime manufacturers and their third party alternatives.
You can glue a seat on top of your lawnmower, it won't make the prices of cars go down, for the same reason this won't make pro photography equipment cheaper.
We had better lenses than this in the early 1900s already
I do some optics design, albeit not photographic lenses. But a f/2.5 lens is not considered to be an easy design problem.