I really enjoy procedural generation, especially systems designed to work with hardware outputs. After starting work at Formlabs in September of 2023 and gaining access to much nicer printers than I was used to, I started wanting to tackle some large algorithmic structure projects. Complexity is free in 3d printing, the limit of design geometry is mostly how much time you’re willing to spend in CAD. I wanted to print the most complicated art piece I could think of. Marble Fountain is what I came up with.
Tracks
My initial system came together quickly. Randomly placing spaced out points, drawing a spline through them, and setting a constant slope just works. My first draft was just subtracting a tube from the solid support structure which worked but was super limited. I wanted to add more parts and so I started working on a path solver. I wanted to fit as much motion into the volume of the printer as possible. This turned out to be extremely challenging.
The solver starts by making a random series of line segments connecting the top and bottom of the lift. There are several different algorithms to generate this guess, as the initial conditional has a noticeable impact on the shape of the structure for lower path counts. It’s interesting to play with different variants of the starting conditions and see how they change during generation.
A series of functions update the positions over time to “pull” the points into a followable path. The points making up each path:
Stay in the bounding box
Evenly space themselves out
Pull towards a fixed height to enforce a constant slope
Enforce min and max turning radius of the path
Repel away from other tracks
... continue reading