A new 3D Scanner for the Hackspace

[Ulisses]

I've been thinking that a resource that could be very usefull for the  hackspace would be a 3D Scanner. I suppose it didn't really need to be able to scan very big objects; it's probably more important that it achieves a good level of detail.

I'd like to have your feedback about which 3D Scanners you think would be more appropriate for the hackspace.

If there are many people that find it usefull it would be worth starting a pledge for it.

[Peter "Sci" Turpin]

If you're after a high level of detail and accuracy, I think a
contact-probe is still the best unless you're spending thousands on
precision laser scanners.
Then again, one of the Leap Motion devices and a stepper-driven
turntable would probably be excellent, but the device isn't out yet.
(expected feb 2013)

[Martin Dittus]

Don't people simply use the Kinekt these days? Afaik we have one or two at the space.

m.

[Ulisses Pinto]

Kinect is not that accurate.

"Kinect by nature does not see/scan small objects, recommendation was “Minimum is roughly size of a shoe”.

I got this infomation from the URL:

http://www.jannehansen.com/2012/08/15/kscan3d-kinect-budget-3d-scanning-at-home/

[Olfin]

It depends upon what you want to scan: sizes of objects, and expected detail.

The Kinect is fundamentally unsuitable for use as a scanner: it is a very, very noisy sensor, and it has a very low resolution. (time-of-flight is not good for precision work)

I'm guessing that the most likely use is something in the 1cm to 200cm a side range, for use in a 3D printer.

The best option with our budget at this point is to make a scanner. Laser line scanners or structured light are a good way to go, as they are inexpensive, and can be adjusted to scan objects of different sizes, after calibration.

There are a good number of builds out there, and it's a fairly well-documented thing.

In general, you need a good camera, turntable w/ stepper, and either a laser line or a decent small projector that can be controlled precisely.

Contact probes are very accurate, but they're problematic to use, not to mention expensive.

This is the Hack Space! Let's build one!

[Billy]

Chris the hipster started building one.

I've still got some of the parts that he gave me for the reprap i'm
working on.

I'm not sure where the laser-scanner he was using went though.

[Jim]

I wonder if one of these would be any use: http://www.kickstarter.com/projects/1108292587/99-lidar-project

On Monday, 8 October 2012 19:24:00 UTC+1, Ulisses wrote:

[Benjamin Blundell]

Ok so, the Kinect isn't time of flight for starters, its based around
structured light. It is noisy but the resolution is not as bad as
you'd think. Again, some great work from Microsoft research on the
macro level stuff (desk and room) but not for small objects.

Laser line scanners of the home brew variety and the one hipster built
are flawed. The software is utter crap (I know because I've had the
debug it).

Structured light scanners are worse than the kinect and a bitch to setup.

Dual camera rigs and time of flight... yeah, they dont work so well either.

Touch probes, not so sure about. One thing that does work however is
my scanner! :D

http://blog.section9.co.uk/2012/08/17/Organ-Reconstruction.html

however, again, the resolution is questionable but with a better
laser, you essentially have a touch point sensor. A higher resolution
and only 4 cameras or maybe even 3 would work a treat. This is easy to
build and the maths is not so bad.

Ultimately though, the nicest solution would be the monocular SLAM approach!

http://www.robots.ox.ac.uk/~gk/


Again though... why do we need a scanner? Its been done and we have no
3D printers or things anyways.

[Benjamin Blundell]

two kinects and the bumblebee addon is not such a bad thing btw.

http://www.notcot.com/archives/2011/04/be-your-own-souvenir-by-blabla.php

http://www.cs.unc.edu/~fuchs/kinect_VR_2012.pdf

Again, still macro level though. Yes there is the leap but I'll
believe that when I actually have one on my desk.

B

[Peter "Sci" Turpin]

Actually, this post gave me an idea. There is another option; a CAT scanner.
I already have that x-ray machine head and would need help building it
up anyway. I'd be surprised if the resultant data couldn't be turned
into useful model files. And it'd show internal voids/structure.

Would require an x-ray lumenescent panel, mirror, decent resolution
low-light camera with PC control and a stepper-driven turntable. And of
course a lead-sheilded enclosure with safety interlock and a PC to do
the image conversion.

[Russ Garrett]

On 9 October 2012 14:27, Peter "Sci" Turpin <s...@sci-fi-fox.com> wrote:
> Would require an x-ray lumenescent panel, mirror, decent resolution
> low-light camera with PC control and a stepper-driven turntable. And of
> course a lead-sheilded enclosure with safety interlock and a PC to do the
> image conversion.

It's worth talking to Mike Harrison because he just got a line-scan
baggage X-Ray machine which he's intending to strip for parts. The
sensors in it seem useful and pretty easy to interface. As always, he
has a great teardown video:

http://www.youtube.com/watch?v=KPyE29ABmoA

I think you'd have some difficulty cheaply making it safe, though.

--
Russ Garrett
ru...@garrett.co.uk

[Peter "Sci" Turpin]

Yes, it would require a good few kilos of lead sheet.

[Ulisses Pinto]

Building a new 3D scanner seems in fact more in line with the Hackspace philosophy ;)

There are however some points I'm worried about:

- Are there several people that built the scanner that reported it works properly and has a good level of precision?

- The documentation to build the scanner is detailed enough, so that we don't get stuck in the middle of the build, not knowing how to proceed, or building it incorrectly.

- Is there is software developed to work with the scanner? Just having the machine without software is really not that usefull...

In case someone knows a project that answers all these points, can you send a link to it?

[Benjamin Blundell]

tbh, the project I was on was a lot of fun and I'd like to see more
work on the reconstruction from movement with a single camera. I think
that would be a useful application. In medical surgery, such cameras
are used a lot and reconstruction on 3D is quite useful. Hence the
motivation, and money, for the work we did. I think scanners are one
of these things that really need an application to be worth the time.
X-Rays are, I suppose, fun but reducing the amount we use is probably
a good thing.

Ultrasound, on the otherhand.... ;)


On 10 October 2012 10:13, Ben Blundell <o...@section9.co.uk> wrote:
> tbh, the project I was on was a lot of fun and I'd like to see more
> work on the reconstruction from movement with a single camera. I think
> that would be a useful application. In medical surgery, such cameras
> are used a lot and reconstruction on 3D is quite useful. Hence the
> motivation, and money, for the work we did. I think scanners are one
> of these things that really need an application to be worth the time.
> X-Rays are, I suppose, fun but reducing the amount we use is probably
> a good thing.

> --
> S E C T i O N 9
>
> |- Benjamin Blundell
> |- www.section9.co.uk
> |- www.twitter.com/secti0n9
> |- Unit 30, Cremer Business Centre, Hoxton, London