Jeffery makes (almost) Anything

Project 3 : 3d printing and scanning

project photo

3d printing and scanning opens huge possibilities. i wanted to probe a a few limits concerning how porous the objects created and scanned can be.

3D Printing

I started considering cristal structures for printing. Cristals naturally assemble in regular and fascinating geometric shapes. These often result from many instances of one simple and regular shapes; a list of different possibilities can be found in "Bravais lattices".


I imagined representing a lattice with thin cylindrical rods of material representing the edges and inner structure of several adjacent lattices and small cubes representing the corresponding vertices. How thin and intricate would I be able to make these rods?

basic element

I started programming the basic shape in Antimony. It is composed first of a cube that is rotated so that the ground is tangent to one of its vertices. It is then inscribed in a hexagon of corresponding proportions. The diameter and length of the rods are all parametrized to enable testing of rod thickness. I then added complications to this basic shape: I duplicated the shape four times so a to create a more complex cristal structure. Finally I distorted the shapes so that the edges do not go straight up but progress at an angle. This is an additional touch to test the use of support structures by the 3D printer.


Once the design is complete, Antimony allows exports of meshes in ".stl" files. I can then see the mesh in the software Meshlab to make sure the file is ready. Indeed at first I had to make the resolution a lot higher for the shape to be recognizable.


Finally I uploaded the project onto the sotware of the 3D printer. I used the MakerBot printer, and hence the software Makerware. I had to make sure that the rods would be thicker then 1mm, size below which the machine had trouble printing. I can then upload the output file onto an SD-card and upload it onto the printer. Here is the printer starting with a removable support structure.


After an hour and a half approximately, the printing job was finished:


We can see that a huge amount of support structures (the perfectly vertical walls) were used by the printer in order to print the top layer. The rods that I intended to print are mostly hidden by these structures. So I tried to remove the support structures, which turned out to be very tough. I ended up breaking the fragile rods of the actual shape while trying to remove these structures


To make up for this I then removed some of troublesome factors: I made the scale of the cristal greater, in particular the rod thickness. I also only printed one of the basic shapes and did not distort the final shape.


The MakerBot printer had no problem doing printing this version, and did not even use support structures:


The finished version was successful and sturdy:


However I have merely drawn an upper bound and lower bound on the kinds of prints that are at the limit of the machine's capacities. With more time given, it would have been interesting to test this boundary much more tightly.

3D Scanning

For the scanning portion of the assignment I used the "Sense" scanner and software available. I originally intended to scan small objects with it but it turned out that the best results were of human scale. So I tried scanning myself. The following were useful tactics to make the scan a lot more succesful:


The above scan was the final version I chose. The software was quite straightforward, and allowed me to truncate off unwanted pieces or noise off the scan. I made a secon version that I could 3D print by truncating most of the chest off:


For this printing job I wanted to use the Ultimaker. The software, "Cura", allowed me to easily import the .stl mesh and place it flat on the ground. However the Ultimaker had printing issues.

I had to wait for a later week to finally print myself; the extruder had been fixed. The print was made in fast mode but not the fastest that had been used by others: it had given to their objects an unwanted stringy texture.

The print took around an hour, but was quite successful.

I had a final surprise two weeks later when I saw a familiar face in the Ultimaker. It turns out I had left my .gcode the file on the Ultimaker machine, and Rob had used it to make a demonstration. As a result I now have two prints of the scan I made.


Tools used