Right now we’re forced to draw these little circuit paths by hand every single time in Illustrator, though I’m doing it more in Rhino these days because it’s so easy to trim paths. Then I typically sweep these paths in Cinema4D and render them there.
I have this dream of having some kind of rig, where I could:
Aibrush a blurry black/white image in Photoshop, almost like a noise/heightmap.
The white areas would inform and influence where randomized circuit lines appear. The brighter the intensity, the more activity. You could even paint a few thin tangent lines to influence circuit lines that shoot off.
Load it into Grasshopper and use some combination of techniques (eg: L-system? Unsure…) to actually generate all the lines.
How plausible do you think this is? Any ideas on how I might get started even simply, which I can then explore/experiment/iterate on to chase the kind of look I’m going for?
There are (always) multiple ways, this is one of them:
You will need somehow a point grid
→ you could start with a regular rectangular point grid and cull some points based on your heightmap ( not so nice )
→ or you start with a rectangular grid and use the circle-packing of kangaroo based on your heightmap to get a nicer point grid
now you have to clusterize somehow the points to create a polyline
→ you could cluster your points & values with k-means ( not so easy, and maybe not useful )
→ you could use proximity lines & topologizer & shortest paths to get branch-like-lines ( easier)
→ the next step would be to " discretize" poylines to specific angles with the great plugin pufferfish
→ the last step is to turn your polylines to a 3d-geometry like pipes etc.
I did not tried these steps, but with a little effort and research, it should be possible!
A while back I played with a funny way of generating some patterns that look a little bit circuit-board-ish
The idea came from this tweet by Ntsutae.
It’s just an incredibly short one line function - x+t^x-t)**3%1729&3?0:w
but varying the input constants gives some surprisingly varied and complex patterns.
Feeding in brightness values from an image sampler as one of the constants gave this: