There is more to Gunsmithing than having Shematics but it certainly wouldn't hurt to have them.
With todays technology you can Scan and Digitize every part of a Gun. Simulate, modify and distribute it with Software. Why People aren't doing it already?
Other urls found in this thread:
youtube.com
youtube.com
youtube.com
youtube.com
youtube.com
youtube.com
youtube.com
youtube.com
pocketnc.com
wired.com
youtube.com
twitter.com
I'm a drafter and that upsets me.
why?
It certainly would be great for discontinued guns like H&K MP5 and such.
bump
Hi.
What you get from a 3D scanner is not the kind of model that you can actually work off of. You get a "point cloud". Millions of individual points that MAY OR MAY NOT form a solid mesh, based on how well the scanner and the software perform. For making a 3D model of the clay sculpture you made for 3D printing? Sure thing. For making very precise components that interact with other precise components to perform a mechanical task? Not so much.
Even if you DID get 100% accurate scans in a sketch-based parametric modeling format (what we actually use for Engineering), then you have to realize that it tells you fucking nothing about the geometric tolerancing, the surface finish specifications, heat treating, etc.
Can you make one-offs function this way? Sure, but you'll have to do a lot of hand-fitting. It's not viable for large scale production without thousands of hours of work.
MP5's are actually going to be flooding the market over the next couple years in the US. There were like 3 companies announcing them at SHOT.
>Why People aren't doing it already?
Because consumer 3d scanners are still shit and not capable of plug and play. Especially hand scanners.
Also because museum curators are welfare niggers who don't actually care about actually preserving or showcasing history. And because private collectors collect for themselves.
We should be doing so, but there hasn't yet been a wealthy enough autistic willing to do so.
"point cloud"
This is just how data is interpreted. Even if it's currently that way, things will improve. There is also different 3D-Scanning Methods, the cheapest is using cameras. Lasers have been used for some time, too.
So, with improvement of the Software that interpretes the scan-data and maybe a combination of hardware, getting "solid", correct and workable models.
Even after all that, currently Drafters should be able to make something workable from "point cloud" and convert them to CAD files. (Don'
t lie, I know they can)
>It's not viable for large scale production without thousands of hours of work.
Oh, you mean reverse engineering takes more than just dimensions? It's more than a one step process?
I'LL BE DAMNED
...some people on Jow Forums don't live in the US...
>then you have to realize that it tells you fucking nothing about the geometric tolerancing
If you have an accurate measurement of components, then does it really matter?
>the surface finish specifications
Of all the things to care about, I believe this to be the least important thing that the least amount of people care about and/or think to be important.
>heat treating
Non-destructive ultrasonic hardness testing, accurate measurements of mass, emag absorbance, and other non-destructive measurements can tell you the exact depth of hardening and the composition of the material.
>Can you make one-offs function this way? Sure, but you'll have to do a lot of hand-fitting.
You're assuming that scans remain shit. We have the wavelength technology to improve resolution. All that we need is better 3d tracking for portable models. (We already have absurdly precise 3d magnetic tracking, but it is expensive and very very sensitive)
It kills drafting jobs.
>Oh, you mean reverse engineering takes more than just dimensions? It's more than a one step process?
>
>I'LL BE DAMNED
AI (neural networks) and standardized consumer manufacturing will let us brute force reverse engineering for simple mechanical systems like firearms. The technology is already more than developed enough, we just need some bored faggot with a spare 50 million or so.
Design patents for one, computational errors for two
I'm not saying that scanners are worthless, I'm saying that you basically have to do all the work reverse engineering it anyway. It's the same as tracing a drawing. It doesn't show you the technique they used to draw it or anything like that.
A lot of people think that scanners and 3D printers mean we're in Star Trek now. It might as well be magic to them.
It really does. Unless you plan on cutting absolutely every dimension, including the less important ones, to the exact same dimension that your specific example has. What if the dimension is at the very lower end of what is acceptable and when you manufacture it, it dips beyond that and it's off in the direction away from MMC? Just scrap the part every time that happens?
>Of all the things to care about, I believe this to be the least important thing that the least amount of people care about and/or think to be important.
You'd be surprised how important it can be on many components.
>Non-destructive ultrasonic hardness testing...
None of which you get from a simple 3D scan.
>We have the wavelength technology to improve resolution.
3D models that we actually use for drafting are the theoretically ideal dimensions. That far surpasses any resolution achievable.
>design patents
patents last for 14 years... other nation's laws are irrelevant.
Showing a picture of a trademark is not an infringement of trademark law, neither is showing a scan of a trademark.
>computational errors for two
Did someone say basic statistical analysis is stupid easy to code for if you are anything other than a javamonkey?
I doubt any drafter will lose his job because of this, but if we asume it does I'll wonder how often they drew things machines that later got build and replaced jobs due to automation?
Cars kill horseshoe jobs. Luddite bitch.
Honestly, society will collapse before then. Self-driving cars alone will kill off more jobs than were lost in the Great Depression.
>That far surpasses any resolution achievable.
even if you stick to mundane stuff like near visible IR or UV, you can achieve theoretical resolutions of 0.1 to 0.2 thou.
The tolerances on most things are far greater than a fraction of a thou.
>None of which you get from a simple 3D scan.
True, but heat treats are a pretty loose ordeal with a large amount of variance compared to something like dimensions.
>You'd be surprised how important it can be on many components.
But couldn't you just err on the side of caution? It's not like a rougher part is desirable in 99.999% of cases, right?
>Just scrap the part every time that happens?
For one of a kind gun purist reproductions, yes.
For other more widely produced post-interchangeable parts guns, then obviously you are going to use more than a single example's scan to go off of.
>What you get from a 3D scanner is not the kind of model that you can actually work off of. You get a "point cloud". Millions of individual points that MAY OR MAY NOT form a solid mesh, based on how well the scanner and the software perform.
This, the surface model you get from 3D scanning is pretty useless for anything but hand-fitted 3D prints.
>collapse
More so a valid reason to preserve and distribute designs of current technology. Who knows how much knowledge will be lost otherwise?
>...some people on Jow Forums don't live in the US...
lord I hope so
>No understanding of GD&T
Stopped reading.
>firearms
>simple
No. They're not aerospace-level, but they're pretty close (both in stress, strain, pressure, and heat issues).
It would still be loads helpful to someone building a gun or replacement parts compared to having nothing or only rough sheet-metal-gun-drawings.
>3D models that we actually use for drafting are the theoretically ideal dimensions. That far surpasses any resolution achievable.
Nobody give a shit about the ideal dimensions being ideal because, yeah, they're ideal. That's what tolerances are for.
Brute forcing aerospace would be way easier than brute forcing something mechanical.
Well, even industial professional gunsmith actually test and QA their guns, so unless some dumb nut let it blow up in his face without testing, it pretty safe to say that people can manage by trial and error in that case. (which will then improve the design if results are shared)
>Brute forcing aerospace would be way easier than brute forcing something mechanical.
That's a great recipe for disaster. Also, aerospace encompasses most mechanical aspects, not the other way around.
Which is what thinks neural networks combined with multiphysics simulation will do, but I remain skeptical.
There are a few cases of individuals succeding in small projects like that, but usually it's for stuff that doesn't have the complexity, high-transient and high gradient (temperature or stress) conditions that firearms have to deal with. Not to mention vibration/fatigue.
user, I'm talking about providing a 3d scan to a self-learning neural network and letting it design production methods based on a standardized set of resources using simulated manufacturing.
Things like autocad can automatically generate tool paths for simple shit, so it's not that absurd.
>Brute forcing aerospace
What did he mean by this?
>I want something that looks like this to fly
>design something that looks like that and have it be able to fly
youtube.com
I thought more in terms of firing that thing a few times from a save distance to see/note whenever it breaks but your post sound good too.
Giving an AI direct real world non-scrubbed input is dangerous.
>cs nerds and their prajeet army reinvents a fukin roo to use in the next LOTR animated sequence
Nice. Very impressive. Now how do we use this technology to do anything useful like make a gun?
1. someone get's paid to provide a gun for scanning
2. someone get's paid to draft and convert data to something usable
3. for 1 and 2 someone needs to pay!
the muscle/skeleton probaby could be (mis)used to simulate metal hardness, stress and points of failure. so that an AI could be trained to choose the right material by itself.
Start with turning the 3d scan into a 3d model. It could be done very easily algorithmically, but fuck it, let's use a nerual network because because.
Then comes the task of material selection. We could just use what they used, or use whatever we feel like and then adjust feeds and speeds for different materials, but fuck it, let's use a neural network. Have it randomly select materials and run simulations for stress and fatigue, have it spit out a list of suitable materials for a given part and desired lifetime and SD.
Then comes the task of designing a means of manufacture. Have it randomly generate tool paths, tooling, feeds, and speeds. It then compares it to the learning reference 3d scan and/or 3d model and/or desired materials. I
I fucking got bored. The point is simulated iterative design given only the input of a 3d scan, aka brute forcing it.
Kinda like this.
youtube.com
Except with guns.
The very same thing has been possible for a 100 years by using a set of calipers, using computers doesn't magically make it more feasible
actually machine parts are designed an "tested" in software. so even without AI assitance it makes a load of sense to have a digital model of a gun and simulate it's parts.
making a actual real model of something you measured yourself with cost a lot of time, money and material. Digital modeling will allow you to adjust designs on the fly and let you move it's parts 24/7 and millions of times. if you then build a real model/prototype based on that, it quite a bit more likely it will work as intended.
bümp
if it gets easier to make homemade guns, it would be more difficult to ban them.
only if it has methods of output unconstrained by the user.
there's only so much an AI can do with visual dimensions without environmental context or concepts beyond the immediately material.
Jokes on you, I have a fucking Micro-CT with 5 cubic micron voxel size
yeah but user, what's the largest size object you can scan? And do so have it on a spindle to make proper 3d scans, or is it snapshots?
Draughting is going the way of the dodo, and it's more or less a good thing.
I could probably get away with something around the size of a Kriss Vector, but the idea is to break it up into parts to scan.
Everything sits on a turntable to scan
I've been printing medical devices with it, just thresholding to the device's density, converting to an STL and printing
>AI is going to figure out metallurgy and heat treatment from 3d scans
You're going to be waiting a while on that one, my dude.
I'm suprised nobody has mentioned this yet
ghostgunner net/collections/featured-products/products/ghost-gunner
It's fucking $1,600 but it's a step in the right direction. I remember seeing somewhere that the original one was only $250 or something, I'd definitely be prepared to drop that much money on an automated mill but $1,600 is just stupid expensive considering you can buy a manual one for 5x cheaper.
Reverse engineering from a point data cloud is real pain in the ass.
You get better results just by measuring the real parts and making schematics from scratch.
user, the answer will literally always be 1040, 1080, 4140, or 4180. Or 6160 alum, or brass (yeah there are technically alloys, but fuck off).
There are a highly finite number of high production materials.
what if you integrate this into 3d scaning? I imagine a sensor "wheel" getting dragged over the surface of an object in addition to a photo/ laser scan. Then maybe add even a sonic type and the data should be a lot more complete and useful from the get go.
While someone got to interpret this data in software to make sense before that, I guess laser (spectrumeter) and sonic (vibrometer) could even do a materal and density analysis. Then, if nothing else you'd be at least able to tell whenever it plasic or metal form the data and at least pre-filter and more acurately specify what you want to scan.
hyothetically and ignoring that this would be many times more expensive, adding rontgen to the mix built-in of scanners, not only would the scan-resolution improve a lot, you'd probably would even be able to do a wear level analysis (micro cracks) of the scan-object.
>rontgen
Am I mistaken or isn't that non-non-destructive?
I don't really know. I heard it's used to check aircraft turbine blades of structual integrity so I asumed it's usable. Wouldn't make much sense to check whenever there are cracks and you'd destroy it in the process of doing so. I could be wrong, though.
the files already exist dingus
A lot of people mock the capabilities of it for the price, but the main advantage it has is that all of them are, by default, set up the same way and so GCode is a lot more interchangeable between units than your typical CNC mill.
This.
I personally think the future is in something like a widely adopted washing machine sized standardized 5axis. It would unleash a consumer manufacturing revolution.
heh heh for only $100,000/piece tho
Prices are coming down and capabilities for the price point are going up.
youtube.com
This was 3 years ago. Today, you can get the V2 for 4900 USD.
I "fly", I crash, I "fly" again.
Aerospace Engineering by trial and error, every tag is flugtag.
Procedurally generated geometries DO have a place in engineering (aerospace included), but it's going to be as a tool for an engineer/drafter, not let loose on its own (except for maybe academic purposes). Things like generating honeycomb structures and the like.
wired.com
I simulate flight, I crash, I simulate flight again.
Or the Luger.
You could have a 3d printed one without the kike price because MUH MACHINING even though it costed 600 back in the day in today’s money.
>kike price
it's mostly an issue of production scale and regulatory burden.
Besides, 3d scanning/preservation and consumer manufacturing are completely separate things.
They go hand in hand, but they are not directly related.
How long until I can type in plain text "A japanese trials rotary magazine pedersen rifle with krag-rifle loading gate chambered in 556 with a clip guide and that also has an underbarrel 25acp barrel with tandem magazines (2 mag wells with discrete magazines) that is fired from a thumb trigger and the safety for which is a garand safety lever"
And an AI program will design, and design a means of production that I can then stick onto a miniaturize (nano-miniaturized) solid core memory sphere and walk into my garage to upload to my standardized consumer manufacturing center and have the rifle produced and assembled all within 72 hours?
I fucked around on Solidworks making a model of the P08 based on some old shop drawings. The thing is a fucking nightmare, DESU. Even just basic shit like the ejector design is fucking unbelievably overcomplicated.
The data you get from a 3D scanner gives you a surface mesh, not a parametrically defined body. The latter is way better if you actually want to do anything with the model.
now, you're just trolling. what's the problem in converting the mesh into an digital object and from there into a model? I can understand it's a problem if the scan also got another object in the picture and you'll have to clean up shit, but one you've got a got scan with real world measures (as in here does it start and there does it stop) it's basically childsplay to work with it.
you can have a new production one for 5k
>lugerman
>I have no idea what I'm talking about so I'm just going to say you're trolling
Trying to create parametric solids from a 3D meshes is similar to trying to generate vector graphics from raster graphics. It can be done but it takes effort and know-how.
This: Let's say you scanned a cube. In a parametric model, every angle is going to be exactly 90 degrees, every line perfectly straight, every surface perfectly flat. It is a hypothetical geometry that is impossible to achieve, because even atoms have a radius to them. It's a mathematical representation.
What you get from the scan will have curved/angled/rough surfaces and none of the angles are going to be exactly 90 degrees. The deflection of the mill as it cut the profile of the cube alone will be enough to create what is essentially a draft angle. Even if you're unable to perceive it with your eyes or even your finger tips, it is invariably a collection of imperfections.
When you machine the first surface on a part, that becomes your first datum. It's what you measure everything else off of. If it's at an angle or curved or has a rough texture, measuring it twice will invariably give you a different result. The more imperfect, the worse the accuracy.
When you haven't studied or worked in the field, it's hard to really grasp how anally retentive you have to be. What you're asking about simply isn't feasible at the moment. Until the AI is as advanced as the human brain that would remake it from scratch based off the input it gets, then you have to do all the work anyway.
Scanning a Sig P210 and trying to make it off the mesh may as well give you a model of a grease gun.
Additionally, editing meshes sucks donkey dick, as does trying to create engineering drawings from them.
Fist of, making a solid out of a mesh is literally done by the click of one icon. If it can be 3d-printed or imported in game-engines it already perfectly fine for simulation and reproduction.
What you seem to want is vector models that can be used in professional software? This is also no problem as described a few posts back, because the scan data from the draging wheel would have actual measurements useable to create 2D vector models from the get go.
Effort is as much as placeing the object in the scanner. Know-how is still needed but high-school level drafting basics are probably enough.
Also, now I'm done taking to you. There is always someone implying "it's to difficult, don't do it!!!! when it's actually not".
Dragging wheels have been used in the past to take measures and probably still in use somewhere right now, and while there is currently probably no comercial 3D-Scanner that got that addition, it's so damn easy to build even I could do it so it's not an argument.
>Fist of, making a solid out of a mesh is literally done by the click of one icon. If it can be 3d-printed or imported in game-engines it already perfectly fine for simulation and reproduction.
>What you seem to want is vector models that can be used in professional software? This is also no problem as described a few posts back, because the scan data from the draging wheel would have actual measurements useable to create 2D vector models from the get go.
>Effort is as much as placeing the object in the scanner. Know-how is still needed but high-school level drafting basics are probably enough.
>Also, now I'm done taking to you. There is always someone implying "it's to difficult, don't do it!!!! when it's actually not".
>Dragging wheels have been used in the past to take measures and probably still in use somewhere right now, and while there is currently probably no comercial 3D-Scanner that got that addition, it's so damn easy to build even I could do it so it's not an argument.
just going to save this idiocy for posterity
>idiocy
nigger
>Filling in a mesh to make a dumb solid is the same as making a parametric model
Dude, I give up. Multiple people have tried multiple times to explain to you why these are not the same thing and do not have the same capabilities. We do not live in the 80's. Only old farts too stupid and/or lazy to learn parametric modeling use AutoCAD for anything mechanical. What little "drafting" you may have learned in high school is ridiculously low-level and outdated, yet you insist on thinking you have even the slightest amount of knowledge and experience necessary to make a judgement call on the viability of it.
You want to see what this kind of Mickey Mouse shit turns into?
youtube.com
The telling part is how he references "game-engines", which is a pretty clear sign he's coming at this from a computer graphics perspective and not an engineering or manufacturing perspective, and as such he should probably be entirely ignored.
shills
glow in the dark CIA niggers
Yes. I am a massive shill for proper Engineering and Manufacturing practices, especially for things that could blow up in someone's face, killing them.
Recently, there has been success in using misusing existing WiFi to get a layout of a house and get the location of people living there.
The resolution is still super bad but using it on small scale with 3d-scanners (maybe even DIY Radio) or right away something like low power magnet fields would certainly improve the scan data.
>it's another "people grossly overestimating the capabilities of emerging technology" episode
It costs that much because the inventor uses the profits to fund his legal defence from being sued by the government.
Bro, I activly push this tech. Why? It makes better drafters. By this I mean you scan the shit, study it in 3d, and improve it. You know your material tolerances and reactions and can design it to perfection.
You see a threat. I see a fucking glorious tool.
How many goobers will be able to actually understand what they are doing beyond scanning and trying to reproduce?
To all the American drafting anons in this thread, do you know if it's easy/possible to get a working visa with a drafting qualification? I'm currently studying drafting, wondering if I can use it to escape.
If yes, please convert to metric in the next couple years, thanks.
[Currently there is near DDoS like spike in CAD/CAM Software and Tutorial downloads on known Torrent bays]
As someone who's worked with 3D printing and scanning, as well as being a traditional 3D modeller, for the love of god OP, talk to 3D scanning is utterly fucking atrocious. Point cloud stuff like the Kinect is fun as hell, and has its uses, if configured correctly, an array of them can make short work of otherwise daunting jobs in fantastic fucking detail, but for any object smaller than a foot, especially one where fine, centimetre level details matter, let alone fractions of a millimetre, scanning just won't do the job. If it's a flat part, a 2D flatbed scanner will help a lot, but get someone who knows how to draft and give them any and every gun part you want duplicated or documented, and let them do the job perfectly
>centimetre
Centimetres are for interior decorators.
Real men use millimetres!
I know, it's that 3d scanners are hardly accurate to the nearest centimetre if you're lucky, if you want millimetre precision you're fucked.
>for the love of god OP...
Too late, mate. it's going down. People will copy and make funs at home like crazy now. It will happen around the world and many smart people will make it easier and more automated fast and faster. It will be glorious! The Golden Age of homemade weapons.