ITP Blog

Subtraction - Week 13: A different approach to sculpture


When I started to know more about digital fabrication, especially in art and design field, I always have a question. What's the point of using machine to create an art piece? Computers, softwares and CNC, they create perfect and precise shapes. I admit that they are doing great jobs, but I feel a little bit boring when I saw those perfect and mass-produced pieces. Now we can have two sculptures looks exactly the same, but back to the ancient, those master pieces in MET, only had a unique copy existing in this world.

I didn't mean that the amount of existing copies matters. I like seeing a randomly scratch, a broken spot, or a specific change according to the materials in a work.

I think this explain why I like Sebstian's works in Pier 9. I mean, drawing some thing in CAD software basically just a transplanting of traditional manufacture production without the beautiful mistakes and great insight during the physical touch moment. What I like his work is that he use calculation to generate the shape, which adds new insight to the process of creating sculpture and is creative and generative way to do digital fabrication. With generative concept, digital fabrication can create unique piece and simulate the diversity of nature.

The link of floating wave:



I made two pieces for the final assignment. The first one is more like a compromise to the time slots. I couldn't get the wave generator work at first, so I chose my plan B, I made a curly organic shape in Fusion 360. It's a combination of two piece(#dont ask me why it looks like a spoon...).

The process of making it went well at first, but when cutting the other side, it suddenly went wild and aggressive. So I hit the emergency stop button. I talked to my friend about what happened to my shape, who is a product designer, she said the reason might be there is some gaps between the two part when I joined them. 

Unfortunately, she was right, I checked my file and find a very tiny slot between my two parts.


The emergency stop made it looked like a broken bow tie, which I sort of like it. I sand it then make it a milestone of my subtraction failure collection.



I thought I got no time slots and there was no way for me to make a second piece. Under this condition, I became peaceful enough to figure out how to realize the generative way of digital fabrication.

  • Generative shape from processing

According to the tutorial, I used a processing application to generate the vertices of a composition of different tide waves, which is more realistic.


The output of processing wave generator is a vertices collection txt file and not machining ready. (There is two bugs in the wave generator, the gradient function and the floor vertices function don't work)

The next step is to create a solid shape using the vertices with meshlab, rhino and fusion 360.

  • Create mesh from vertices in Meshlab

In meshlab, import the txt file using default settings and adjust the point size so we can see the vertices. Then use filters/normals, curvatures, orientation/compute normals from point sets(re) and filters/remeshing, simplification, reconstruction/surface reconstruction: Pivoting Ball to construct the mesh surface from the vertices I have.


Sometimes, the mesh has holes on it, I would go back to the processing application and remake another set of vertices.

  • Create prototype mesh in Rhino

Then I export the mesh as a STL file, and import the mesh to Rhino to do further construction. I like to try different softwares.

In Rhino, I imported the mesh twice to create both the top and bottom side of the wave, and added some distance between them.

The next step is created a 3D prototype. The tool I used was ExtractMeshEdges, which would create a new outline of our mesh. I would recommend import one mesh first, then ExtractMeshEdges, hide the original mesh, join all the edges into one group. Then create another layer and do the some process again. In the end, it would be easier to do next steps.

Hide both of the top and bottom meshes, only show two groups of edges.  Then, I used Loft function and created a polysurface as the side surface between top and bottom sides. It took me a little time to figure out why I can't join my side, top and bottom surface together. It turned out that we can't join mesh(top and bottom) and polysurface(side) in Rhino directly. The solution was easy, Mesh/from NURBS object function will turn polysurface to a mesh, then we can join them together to make a solid shape, theoretically.

The last step in Rhino is to export the shape as a STL file.

  • Make a machining ready model in Fusion 360

In fact, right now I'm not sure if it's necessary to do this step. I'm just curious about what it would be in Fusion 360, which turned to be my favorite CAD software.

In the latest version of Fusion 360, we can do mesh editing in a mesh workshop. To turn on the mode, we need go to the preferences/preview and check the mesh workshop.

I don't really want to risk destroying my model, what I did in Fusion 360 is to convert a mesh to a Brep Object, which I think is a safe choice for 4 axis machining.

Save the model into a STL file, and the 3D modeling is done.


  • Machining

Luckily I got a time slot on Sunday afternoon after Jackie.

The material I used is the rest frame piece of my first try. I cut it into a block shape.

To cut short the time, I did the CAM first and got the minimal size I need, and adjust the Z size of my material.

  • Presentation and finish