3D Scanning and Printing

Design and 3D print an object (small, few cm) that could not be made subtractively.

3D scan an object (and optionally print it)

Printing theory

3D printing comes down to a combination of:

  • Nozzle heat
  • Print speed
  • Flow speed (speed of material)
  • Bed temperature.

    Bed temperature effects of how well the material sticks to the bed. The bed temperature is very sensitive to changes in room temperature. For larger print you have to use glue. Brim" is another great way to help keep your print from warping. This feature puts what looks like the brim of a hat on the bottom layer of your print to help fight against the pulling forces of the cooling print

  • 3D printing trouble shooting guide
  • Top tips for 3D printing.

  • Test Prints

    I started to design a shrink test in antimony. I made it all parametric and nice with embossed and extruded text. Unfortunately I could not export the file as I get errors in the extrude node c output property. I could not figure out why so I ended up exporting the shape without fanciness. How ever, here I bumped in to the second problem. Finding a balance between file size and quality of the mesh.

  • 8 voxel = 3MB
  • 16 voxel = 12.7MB
  • 32 voxel = 51.7MB
  • 64 voxel = 207.4MB
  • Found at 64 voxel was getting the quality i wanted. All in all a bit anoying not to get what you see.

    Export to .stl from Antimoy. Low voxal gave me chamfers on all the edges.

    Shrink test results

    This was actually a shrink test and press fit test combined. The bed was not calibrated properly, a bit to high, this resulted in that there where small gaps between the lines on the bottom layer and the print did not stick (the cubes fell over before the print was finished) and probably the cause for the wrapping taking place.

    I found out that in general the cube held its dimension quiet well. the smaller one shrink more 0.1mm. As for the holes the width was quiet accurate while the length suffered quiet a lot 0.2mm. This is most likely the result of the wrapping.

    Heat test results

    Heat test was designed in Blender and was aimed to see how the filament behaves within the recommended range 195 -220. And then one where i really heat it up. As with the previous test the calibration of the nozzle head was also to high so the result is not so valid but interesting to see that the hottest temperature did the best. I did notice a lot of wrapping around the edges while printing so the higher levels did not have as good start starting point as the first.

    Heat test for PLA.

    Ball Bearing

    I wanted to see if my my ball baring design I did in week 2 would actually work. And then from there start building on top of that adding more rings for pivoting on several axis.

    The pint went well and but needed a bit of clean up to remove the support material and raft. The balls where still stuck, but with a bit of prying with a screwdriver and suddenly it all came loose. It was definitely a Kodak moment but could not get my hands down from the excitement of success.

    Top and bottom side. I used raft as platform adhesive type which left some really visible marks .

    Working but not efficient 3D printed ball bearing. Design in Antimony and printed with PLA on a Ultimaker 2

    Ball Bearing v2

    Next step in the spiral development cycle was to make the next ring for pivoting and rotation. I designing it with the same principals as the bearing it self in antimony. For support I choose brim to get it to stick and to be able to fine tune the settings if necessary, for support I choose touching build plate. Both of theses decisions was a mistake. Since I'm basically printing a C- shape for the outer ring, having the support structure set to "touching build plate" it ONLY build support for over hangs and not put any support inside of the model. Antimony design. Coming back to Cura and choosing everywhere for support gave me horizontal support, but I want vertical support as I can't go within the shape and for clean up. Seems like custom built support is the only way to go. I will also I will skip brim on next print as it was impossible to remove... maybe a raft?

    Cura will only give me horizontal support Where as I need vertical support only in the opening.
    Result after failed support settings.


    When digitizing physical objects, there are two primary strategies. Surface scanning uses structured light, lasers, etc. to produce a point cloud describing an objects surface. These points can then be stitched together into a mesh, often with color information, to produce a triangulated object model. In contrast, volumetric scanning (e.g. CT scanning) produces a volumetric data set with millions of object density samples stored on a three-dimensional lattice.

    Skanect and OpenKinect

    Skanect impressions. The resolution on a Skanect is not that high but we where spray painting with the sensor all over the place. Getting multiple data for the same body parts. I think this confused the software as it inevitable that the person moves a little bit. We clean the mesh and made it water tight within Skanect and export to obj. Here we should have been more aggressive reducing polygons. We had meshes with millions of polygons and that could not my Mac Book Air core i7 with 8gb RAM handle. We cleaned up and reduced the polycount in Rhino and then I went on for further modification in Blender. The next day I noticed that Skanect saves all the scans by default so I could have come back and exported again.

    How I will approaching scanning next time.

  • Start with capturing the face and make sure I'm not scanning it again.
  • If possible have the person on a turn wheel or bar chair.
  • reduce polygon count to 40 000-100 000 but do no further editing in skanect.
  • Export my .OBJ and repair the mesh in Meshmixer.

  • Mesh Labs reduce polygons. My Mac Book Air could not handle the large amount of data

    Printing result. You can see some of the infill shows through to the outside layer. This is because I put the shell thickness to only 0.4,one layer, I think at least two outer shells with standard nozzle (0.8) would be preferable

    Strait out of the printer
    Cleaned up the print with a high precision tool called exacto knife.

    123D Catch mobile app

    Having played with Skanect I wanted to try another scanning platform. I chose 123D Catch and the mobile app. I thought i was very clever with my improvised turn wheel. But spinning the object is not how 123d Catch works! It relies on tracking points to figure out the rotation and the angle of the object. So after that complete failure I did a quick test in my living room. Not so impressive result but the light and setup was also not beneficial.

    All in all it is a nice tool and with right preparation you could get some nice results. The mobile version is super slow so I would definitely use the desktop version next time.


    Spinning setup for 123D catch capture

    Quick test with map of Barcelona as tracking reference.

    Download Assets:

  • Assets files for week 5.

  • Google+

    Follow me on Google +.

    Click me


    Twitter madness.

    Click me


    Connect with me on LinkedIn.

    Click me