[dorkbotpdx-blabber] 3d scanning with Kinect or otherwise?
Mykle Hansen
has anybody on the list played with 3D scanning? i remember
this used to be really expensive and require lasers and
stuff, but lately it seems everybody's hacking on the
Microsoft Kinect and using it to capture basic 3D data that
can then be massaged in software.
what i'd like to do is scan some of my wife's small hand-made
figurines, and try printing them on my friend's new MakerBot.
it's more a proof-of-concept than anything really important,
so i'm looking for an easy, low-cost solution.
cheers,
-mykle-
_______________________________________________
dorkbotpdx-blabber mailing list
dorkbotpd...@dorkbot.org
http://music.columbia.edu/mailman/listinfo/dorkbotpdx-blabber
Brian Collins
I've worked with a couple different approaches to scanning, including the kinect and using photogrammetry methods. For what you're doing, I might (gently) steer you in the pg direction.
The kinect is very cool, but it's low resolution and suffers from what I call the "swelling" problem. For example, take a paper coffee cup, with a fairly thin lip, when you're trying to scan it, that lip swells, or fattens up, due to either the movement of the cup, or the kinect (since you can move either of these things to achieve a 3d scan). For quick scanning of large scale things (e.g. a room, desk, etc.) though, it's great.
The photogrammetry approach (more specifically, structure from motion) is interesting because it has fairly modest requirements and has the potential to generate very nice results, with just a decent digital camera and a PC. You could rig a simple rotating table, put your figurine on that, and take say 30-40 (as a start) photos, evenly spaced as you rotate the table. Then you feed the images through a feature detection algorithm which finds "points of intererest", and then from that you attempt to reconstruct the positions of the camera (even if the camera is held still, the duality of movement means you could always consider the camera to be moving while the model remains still).
Once you have 3d pose data, you can feed that, along with the features and original images, to software that will generate a dense mesh for you, including color! It's quite amazing to me that this is possible, and I'd definitely recommend anyone interested in this to check it out further, because it's pretty accessible to most techies (but expect to put some time into it). Here are some links to get you started:
Sift, a feature detection algorithm (there are others, but this one is good)
http://www.cs.ubc.ca/~lowe/keypoints/
Bundler, which can wrangle the whole process for you
http://phototour.cs.washington.edu/bundler/
Pmvs2, which generates the final 3d meshes
http://grail.cs.washington.edu/software/pmvs/
This is only the most high level description of one way to do this stuff, but this post is already getting kind of long, so I'll stop here. Good luck with your project,
Brian
Anselm Hook
at http://pointclouds.org/ ...
--
@anselm 415 215 4856 http://twitter.com/anselm
ioan ghip
---------- Forwarded message ----------
From: <matr...@aol.com>
Date: Tue, Mar 27, 2012 at 1:13 PM
Subject: Re: Social Media
To: ioan...@gmail.com
I appreciate the details, I have another question I want to ultimately
do what this company www.reallifeconnect.com is doing with
their rfid reader and software. I'm going to have somebody else build
the software end so the rfid reader can scan wristbands,
send the info to twitter & facebook, allow event attendees to register
on a user interface for wristbands and collect they event
data.How much would you charge to create something like this?
-----Original Message-----
From: Ioan Ghip <ioan...@gmail.com>
To: matrxent <matr...@aol.com>
Sent: Wed, Mar 21, 2012 1:13 pm
Subject: Re: Social Media
Hi,I suppose the best way is to make first a configurable reader that
can connect to the internet. When you have this tested and working, then
you can duplicate it to make as many readers as you need around the event.
Then when you give away the rfid wristbands, you collect from the person
their facebook , twitter, email, etc (whatever they want to register
with) and keep all those info on a central server. When a person
activate the rfid at a remote reader location, the remote location would
send a request to the main server, for example something like this:
http://server_address/?rfid=123456788996&remote_station=2
Then you know that the person that wares the rfid tag 123456788996 just
scanned it to the reader station 2. You find the person in the database
and send to their email/twitter etc, the info that are presented at the
expo stand no 2.
To make the reader, you would need a serial rfid reader, a wireless
router that supports openwrt, an arduino board (or compatible).
The arduino board receives the info from the rfid reader, sends it to
the router over a wired serial connection. The router runs a script that
waits for info on the serial port and when is received makes the request
to the central server.
I think that should be pretty simple and a fun project.
-ioan
-----Original Message-----
On Wed, Mar 21, 2012 at 12:59 PM, <matr...@aol.com> wrote:
I seen your site with the tweeting cat door opener and wanted to know
if you can breakdown how
to make to setup a similar system that can be used for events. I want to
use rfid wristbands
for event attendees and then setup readers throughout they event that
can send information
via twiter, fb etc. once they scan the wristbands.
Hans Lindauer
good/easy way to go:
-Hans
Jim Larson
need in step-by-step form. I met the guy who wrote this when he stopped
at Brain Silo on his road trip. Pretty interesting guy doing some nice
stuff. (He needs help with titling, IMHO,)
On 3/30/2012 9:01 AM, Mykle Hansen wrote:
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 2012.0.1913 / Virus Database: 2114/4905 - Release Date: 03/30/12
Jim Larson
http://www.instructables.com/id/How-to-Teleport-Free-3D-Scaning-and-Cheap-3D-Prin/step3/2D-3D-transformation/