Workflow Part 2: preparing a model for 3D printing


Before I begin Part 2, I’d like to make some updates, since such a long time has passed since I made Part 1 and I now have much more experience with using our GE v|tome|x s industrial CT scanner.


First, MakerBot has moved on to the 5th generation printers, so you can no longer purchase a Replicator 2, like we have. There are some pros and cons to this. For most users, the 5th generation machines are more likely to function directly out of the box. The con is that you cannot open up and fix your 5th generation machine without voiding the warranty. Our Replicator 2, on the other hand, can be taken apart and put back together without much difficulty, so I foresee it working for years to come, even if parts need to be occasionally replaced.


Second, I understand that Osirix is no longer a free option for segmenting CT data. However, Drishti and SPIERS are free software options.


Lastly, I have realized that there are some terms used in CT scanning have very different meanings for medical CT users and industrial CT users. The most problematic of which is ‘resolution.’ In medical CT scanning you can’t change the voxel size (are there exceptions to this?) and so resolution refers to the grey value range. That range of grey values is called contrast in industrial CT scanning. Because you can change the voxel size in industrial CT scanners (by adjusting the distance between the specimen, source, and detector), this is referred to as resolution. Because I use an industrial CT scanner, this is the terminology I will use throughout. Hopefully this prevents future confusion.


On to how to prepare a model for 3D printing!


There are two basic rules to making a printable model. First, the model must be water-tight, aka, it has no holes, and second, it must have “clean” geometry. In this case, clean requires that any given edge has only two polygons attached, normals all face the same direction, and all polygons have an area. In the animation and gaming industry, they often require a “clean” model to made of polygons that are quads instead of triangles. This is not necessary for 3D printing and all of the segmentation softwares that I have seen will export models using triangles. Unfortunately, some if not most segmentation softwares will not necessarily export clean models (omitting the quad rule). The type of geometry that is most likely, in my experience, to result in unclean geometry, is geometry that is swiss-cheesy, like bone matrix.

This complex, swiss cheese-like anatomy results in a messy model with several problems: 1) non-manifold (messy) geometry, 2) discontinuous geometry (floaters), and 3) more polygons than are necessary

In cases like this, you will need to decide where the boundary should be for your 3D print. You can clean this up at the segmentation stage, but changing the thresholding for that region or going through slice-by-slice and filling in the air. Much of the time, the human eye is better at detecting where an edge is than the software is, especially when you have two object pressed up against one another and there isn’t enough contrast for the software to detect the edge.


You can also clean up the geometry once a model has been generated. The best tool that I have seen yet is only available in Mimics or 3-matic, called the Wrap tool. In this tool, you specify what size of triangles you want the model to be wrapped in and how much of a gap closing distance you want it to detect. This eliminates internal geometry and I have not once in four years seen this result in messy geometry. Reducing tools are also really good to use for preparing a model for printing. These will reduce the number of polygons, making it easier and faster to “slice” the model for printing. Also keep in mind the resolution of the printer that you are using. If the resolution of the printer is .2mm, then it doesn’t make sense to print a model that is accurate to 20 microns.


You can also use generic 3D modeling softwares like Maya, Blender, or 3ds max (not to be confused with CT software VGStudio MAX) to clean up your geometry. Meshlab and Meshmixer offer some automated tools to clean models too.


The next step is slicing the model, which converts the polygon model into the path that the 3D printer will travel. MakerBot provides a free software formerly called MakerWare, now called MakerBot Desktop, but there are several alternatives as well. Generally these will display a virtual representation of the print platform. You drop in your model and place and orient it on this platform. Consider how much support material different orientations will require. Meshmixer can automatically calculate the orientations that will require the least amount of support material. You may also want some delicate parts of your model to be oriented up, so that you don’t risk breaking it when removing support material. After this, you can slice and print your model! If you decide that you want to share your model with the world, consider uploading it to Thingiverse or Shapeways and let me know if you do! With Shapeways you have the added benefit of being able to order your model in many different materials, even if you decide not to make the models public.


Happy printing!

3D printed Echidna shoulder girdle, with supports intact


One thought on “Workflow Part 2: preparing a model for 3D printing

  1. Pingback: Workflow Part 1: CT scan to 3D model | A Biologist's Canvas

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: