When I've first heard about 3D printing, the first idea that came to my mind was to make figurines of real people. Frankly speaking, who would not like to have a statue, even a small one?

A couple of months later I assembled a reprap printer from a kit I bought from ReprapPro, the company driven by the father of the RepRap project, Adrian Bowyer.

After another couple of months my 3D printing fleet was accounting for three printers, collectively called The Huxies. Obviously Huxy 1 helped replicating herself into Huxy 2 and Huxy 3. So production was more or less assured. We had a lot of fun together, even learned dancing.

However a question was still unanswered: how to create the models? I tried a "design from scratch" approach. I'm an engineer, and I have to admit that my artistic skills are very limited (at least this is what I thought at that time). Since my son promised to help me with some 3D sculpting, I thought I can help him and developed a parametric head generator. It appeared to be an interesting idea, and I have to admit there was some fun coding the scripts for OpenSCAD. Obviously, the output was not a very nice look of a human face, but was far better to start with such a primitive model instead of the default sphere in Sculptris.

After successfully grabbing the entire family's interest, so that the team was enlarged a bit, we launched "Figure The Moments", an Indiegogo campaign, aiming to get some funding to support the 3Dmaker4U initiative that I developed meanwhile in my mind. Well, it was unfortunately not a successful campaign at all and my enthusiasm failed down below ground for a while.

I was evaluating the idea of using 3D scanning, but it appeared pretty expensive and with not so much performance at the time. Suddenly, while I was grieving and think about the way ahead, some articles and news draw my attention. There were about people trying to build full-body 3D scanners based on a Kinect sensor and the release of the new version of a 3D scanning application.

That was what I needed and so I launched the Scanoman project.

Unpacked using js-openctm - Rendered using WebGL

Figurine modelling in OpenSCAD

The main objective was to design and manufacture a full-body 3D scanner that would actually complement the workflow from the real person to the figurine.

Obviously, the idea was not new, but given the lack of detailed information available, it was a real challenge. Therefore, I decided that after I'll finalize the prototype, this will be open source and, if my plans would go well, it might be that I can also produce some more pieces in a way or another and further improve the model.

And it's more or less obvious that Scanoman will... scan a man. Well, a woman or a child, as well.

With these shorter and longer objectives in mind I just kicked it off.

Scanoman at work

The design objectives were simple:

  • full-body 3D scanner,
  • using a Microsoft Kinect sensor,
  • being fully motorized so that the scanning process will be triggered by an one-button-shot,
  • having a small footprint when installed for operations,
  • and being portable so that not only people would have to come to your studio, but you may be also able to go to their premises.

It was also a condition that all the materials should be available from the regular hardware stores, except maybe from some electrical components (e.g. motors, controller) that would have to be sourced from specialized stores (yet known in the DIY world).

Needless to say, as many parts as possible should be 3D printed.

Figurine sample - Daniel

Scanoman - Intro

Scanoman - How to scan

Scanoman - How to mount it

Scanoman - How to pack it

As being envisioned, after a thorough documentation, the information will be published online for those interested in replicating Scanoman, or searching a source of inspiration for their own projects.

Meanwhile, just to complement the information provided in the videos, here is a brief description.

  • The assembly is composed of 2 plates that could be packed together to form a kind of 500 x 400 x 240 mm box, which can be relatively easy (at 15 Kg) carried here and there.
  • One plate is hosting a 40 cm turntable, equipped with a 300-teeth gear (actually implemented through 12 25-teeth circular segment racks) and supported by 8 rollers using 626 ball bearings, turning around a simple 8 mm pivot held by 608 ball bearing. The surface of the plate that the turntable is actually rolling on is covered with a 1 mm steel sheet to better withstand the wear.
  • The turntable is powered by a NEMA 17 48 Ncm stepper motor through a 3-stage reduction gearing mechanism.
  • The other plate is acting like a base for the pole and also accommodates the electronic parts, the wires and the brackets required to secure various parts during transportation.
  • The pole consists of 5 segments made of 25 mm aluminum square profile empty bars. Assembling the pole is done through interior-mounted couplers and secured with M3 screws. There is a 5 mm step rack along the pole, made of multiple segments.
  • The pole plugs into a support on the base plate and could be fixed against the ceiling by means of the threaded rod and a nut-insert button that would press the entire pole assembly between the ceiling and the floor.
  • The sensor is an off-the-shelf Microsoft Kinect for Xbox, placed on a slider on the pole.
  • The slider is equipped with two motors and the associated gearings that can move the sensor up and down along the pole and also tilt it (it was not possible to use the internal motor for the prototype).
  • The controller is a Minitronic from RepRapWorld, perhaps the cheapest standalone 3D printer controller of the moment.
  • To reduce the wiring, there is also a 4-port USB hub that connects both the sensor and the controller to a single USB port on the laptop/PC.
  • All is powered from the mains. There is an extension coupler with multiple sockets on the base plate, a power adapter to supply energy to the motors (19 V). The Kinect is powered by its original power adapter, while the controller is powered through the USB connection. The USB hub, though accepting external power, is powered from the laptop/PC through the USB connection.
  • There are also two spotlights that could be mounted on the top and bottom of the pole, or elsewhere. They are turned on and off automatically through the controller.
  • The scanning process is controlled in terms of positioning the sensor and turning the turntable by an application normally used as a host for 3D printing, called Pronterface (part of the Printrun open source distribution). Some scripts that contain the necessary Gcodes, as well as a Windows mouse-clicking script is providing the one-button-shot for the entire scan.

Also, here are some of the challenges I faced and thoughts of improvement.

  • The design and construction of the turntable is really simple, cheap and effective. However the actual area of the turntable itself is relatively small. Remember, "scan a man" really means "one man". There is definitely a need to enlarge the turntable plate. But this pose certain issues in respect to portability and also questions if the current rolling system would support more persons (currently will work with no problems for two lovers tightly holding into each one arms - subject to turntable size constraints).
  • The pole supporting the sensor slider was made of simple square aluminum empty bars and will provide, at full extension, a reach of up to 2.7 m. This is enough for a regular apartment room. Moreover, anchoring the pole and fixing the pole segments with screws will allow for scanning even in a park (assuming you can source 220 V somehow). However, the rigidity of the pole suffers if the top is not fixed to the ceiling. While this would actually not necessarily affect the scan quality, it might be not so reliable during time.
  • The motors and the controller are "normal" components for a 3D printer. The motors are NEMA 17 size, which are equipped with adequate gearing, and able to carry the load on all axis (well, X is the turntable, Y is sensor tilt and Z is vertically along the pole). The controller is a bit too much for the purpose. For the number of the required control lines , an Arduino Uno would suffice. Coupled with some additional stuff (3 stepsticks, 2 endstop lines and one relay driver for the spotlights), such an installation may come cheaper and more flexible when controlled with CNC specific host software.
  • There are certainly some over-dimensioned parts, such as the plates (in thickness), the power adapter or the depth of the pack. Working to optimizing these aspects will definitely be included in further development.
  • It would have been great to have time lapse still picture capturing, but the USB 2.0 connection does not have enough bandwidth to accommodate both the sensor and a webcam. An additional USB connection would be required.

Scanoman packed

Scanoman pole plate

Scanoman turntable plate

Scanoman parts

And here we come to the core application that actually allows Scanoman to do its job.

After testing some other software available on the market (e.g. Skanect by Occipital, Scenect by Faro) I stabilized my option to ReconstructMe by Profactor.

This was by far the most adequate application for my purpose:

It is actually made for this.
While you can definitely scan some objects as well, all is playing around people 3D scanning.
It's evolving fast.
Remember my history? When I started, ReMe was more of a development platform then a product. Currently, it's very user-friendly software backed also by a solid and performant SDK. And the guys at Profactor seem to get very excited by new developments.
It's fashionable
Well, what you can ask more from a 3D scanning application than to produce a ready to 3D print 3D selfie, with one click of the mouse?
It does the job very well
Actually this is the main point. Even with a single turn full-body scan I can get a good mesh that I can easily refine in a 3D modelling application. With 3-turn full-body scan you get a level of detail that would suffice for most reasonably scaled figurines (i.e. 1:10).
It's almost free
You can use it for free for non-commercial purposes, with full capabilities, though you have to wait somehow longer more when processing the captured data and to work a bit to get rid of the watermark saved with the model.

 

Well, of course, there are things that could make the life (of people like me) even easier:

  • Provide a one click process for full-body 3D scanning (presumably more difficult than for 3D selfie, but with a bit of "standardizing" the processes and postures, might work)
  • Provide a mean to take time lapse still pictures during the scanning process. The rationale is to provide more visually accurate picture reference for further processing of the model (e.g. 3D sculpting or painting). And this will make an additional webcam unnecessary (though the resolution of the Kinect camera is very low).
  • Provide independent simplification and smoothing options. For example, as of the current version (2.1.348), 3D selfie comes really smooth, but the large number of vertices/faces in the mesh is not required. So to say, after smoothing, a simplification would be nice to have). The same apply for full-body scans, where the level of detail might not need to be too high.

ReconstructMe logo

ReconstructMe screenshot

ReconstructMe scan

You may wonder, after so many words, what is actually the result of putting Scanoman at work. Here are a few figurine samples.

Figurine samples

We are now in final stage of preparing to launch a beta-test of the initial idea, and see if it can make sense from a business perspective as well.

This project is part of a larger initiative we called 3Dmaker4U. You may find out more from our website.

Please note that the website is Romanian language only, so you may want to use a translation service available for your browser.

3Dmaker4U website link Website

We've also created many channels were we can promote our ideas and interact with other people that are interested in 3D technologies. You may want to use one of these channels to contact us and find more information or discuss various topics.

3D Hubs link 3D Hubs
Our main point of presence in case you want to 3D print something, even if the design is available as just a sketch.
Facebook link Facebook
Our main social networking channel.
YouTube liink YouTube
Our channel to provide informational and entertainment video content.
LinkedIn link LinkedIn
Our discussion group about advances in 3D technologies and their impact on our lives.
Thingiverse link Thingiverse
Our 3D models for 3D printing repository on Thingiverse.
Twitter link Twitter
Well, we tweet from time to time :)

3Dmaker4U is a registered mark of Daniel Constantin.