3D scanners

Anyone got experience? What’s the cheapest way to get reasonable results? I don’t need great precision. I want to scan something, then 3d print a piece to fit around the original object.

I’ve been into 3d printing for a couple of years now & got a 3d scanner a few weeks ago. It’s very cool obviously, but creating a decent complete 3d scan actually takes quite a lot of time and effort, and without paying thousands for reverse-engineering software (which I haven’t done lol) to generate a CAD file from the output you’re somewhat limited in what you can (easily) do with it.

What is the object? What I’ve been doing up until now is just measuring with calipers & drawing in CAD, that still might be the easiest way.

The cheapest way to 3d “scan” is photogrammetry but you’ll need a LOT of patience to get good results. There is free software available though. I got as far as getting a decent partial 3d image which proved to myself that it worked, but didn’t want to spend any more time on it. You can also try using the FaceID camera on an iPhone, I couldn’t get decent results this way but I only tried for about 5 minutes!

A friend just bought a new iPad and got a pretty decent 3d model from the built-in scanner. Might be worth a try if you know someone with one to borrow

If you can measure it by hand I’d say photogrammetry, with a bit of a caveat depending on the object/surface/access. Got a picture?

Software wise, if Agisoft still do free trials for metashape I’d start there, otherwise I hear ok things about WebODM. Not tried it myself though.

The thing is, making a 3d shape isn’t actually that hard, it’s then turning that shape into something you can work with and design around. Very dependent on what other software you’re happy using- what would you be doing the ‘design’ bit in?

BB cluster of a bike frame.

It is a geometric shape so it would be possible to measure it up. But measurements are always potentially inaccurate.

Which bike? And what do you want to fit to it?

Long story.. I am bolting a new BB onto my Patriot behind the old one to add reach. There’s a thread about it somewhere where I get mocked by about 80% of respondents :)

My only experience has been helping with a job where some sculptures (of pollen spores) were scanned and then we tried to scale and shell the scans, slice a hole in the bottom and model in some mounting points before 3d printing as lamp shades.

Was easily the most frustrating cad/additive manufacture process I’ve been involved in only made marginally easier in that it was ‘just’ a bit of art so any errors were not very critical. But my surface modelling skills are weak in comparison to my solid modeling and what I was doing was trying to blend from one (the scan) to the other (the mounting points).

Hopefully things have improved (with a better operator too) but if it was me (with my skill set) I’d be trying to solid model the critical touch points on the existing frame and going from there.

TBH your best bet is to just measure and model it.

You could do a decent job with photogrammetry but it really requires expert knowledge to get the precision you’d need to make parts mate well. That said, if your design clamps up/has some adjustment in it might be ok.

If you go a visual route, you want to matt out the finish somehow, make sure there are good features to capture. You can do this by dusting flour on it quite effectively.

Laser methods probably better for this application but unless you know someone with the kit it’s not gonna be cost effective.

Also, what are you printing in that will be strong enough to take BB loads?!

If you’re doing titanium or something I can see it, but if you’ve got that money what are you doing bodying a patriot…

Anyways, if you have time and can learn, photogrammetry could do it cost free. Read up, practice with e.g. MVE for free then do final thing with free trial of metashape pro. You’ll need access to a decent PC, tho I think you’d get away with 16/32gb ram for something this small.

If you want an easy way to try it, I have a talk I have a few years back where I provided scripts to go from folder of images to model with MVE.

https://iainwallace.co.uk/2018/04/01/bristech-talk-april-2018/

I don’t cover scaling there though.

Also, what are you printing in that will be strong enough to take BB loads?!

If you’re doing titanium or something I can see it, but if you’ve got that money what are you doing bodying a patriot

This!

Even so, it seems unlikely you’re going to be making some sort of flush fit appendage. Photogrammetry done right would aid with the measuring, but I’d still be measuring things! Heck, I’m used to a scanner worth more than my car and for quite a few things it’s still easier to break out the callipers, drill blanks and rad gauges.

@IA- if you’ve not seen them, there’s been quite a bit of work with open source photogrammetry around getting more user friendly front ends. I haven’t had as much of a play as I’d like but things like Colmap, ODM and something else who’s name escapes me just now, even if quite often they’re built off things like MVE.

Also, Epic games bought Reality Capture so I’m really hoping they’re going to push something out to the hobbyist end of the market

Also, Epic games bought Reality Capture so I’m really hoping they’re going to push something out to the hobbyist end of the market

Maybe, but this is really a play to support Nanite in UE5 and production workflows for content. That said, UE has always been cheap/free for hobbyist use to maybe RC will become more accessible. Metashape still the one to beat IMO though.

Yeah, I’m aware of some of the guis etc, tho less relevant for me. I’m into 3D professionally not as a hobby, so I’m either writing my own stuff or using pro stuff.

I’m into 3D professionally not as a hobby, so I’m either writing my own stuff or using pro stuff.

cool, what do you do? Any reverse engineering software you use/recommend?

Well at the moment nothing, between roles – I was CTO of a subsea robotics technology company doing real-time 3D from vision (rather than post processed like photogrammetry, but there’s a lot in common and sometimes you want to do both).

Reverse engineering not really my thing – more of a downstream use case for the data so no recommendations there.

What’s the printed part actually going to do? Basically, if we understand the part and the goals we can probably advise on how to do it… But bottom line is, I can’t visualise a part you’re going to make which is both useful for the job, and which is also going to benefit from 3d scanning.

Or put it another way, you could scan the existing parts and use that to get a really close fitting printed part to fit to it. But the question then is, why do you want that.

The thread is not about my bike bodge, it’s about 3d scanning. I have a composite design with which I am happy, but the good folk of STW have confirmed it’s probably not worth trying the scanning.

All will be revealed when it’s done.

It might be- Northwind hit what I was thinking of with the ‘close fitting’ bit. I’ve thought about it to try and make a little custom frame box that holds on by being a perfect fit to the tube junction and a springy bit wedged in somewhere. Like those ibis frame bags.

It’s just not something to think of as a magic fix for designing something- I get asked if someone can do ‘x’ once they have a 3d scan, where sometimes x is along the lines of asking why, once they’ve taken a picture of a cake, they still have to worry about ingredients and recipes to make a new one. You don’t appear to be thinking along those lines, but some people have definitely been watching too much iron man- wave a red light at it and BAM! You have a CAD model, blueprints, the lot

molgrips
Full Member

The thread is not about my bike bodge, it’s about 3d scanning.

Eh not so much, the thread was about 3d scanning for your bike bodge- most of your posts mention it after all. And the application’s always relevant to the recommendations.

There will need to be a 3d model of the frame, the question is whether or not a scan is worth it. But measuring isn’t going to be too difficult since the frame is made of geometric components in the first place.

I have a composite design with which I am happy,

Ah, is it a scan to print a mould to form composites? That’s a nice idea.

Though maybe there’s something you could use to just take a mould of the structure directly then use that? Or just wrap the structure directly and bond to it.