I am currently exploring complex organic topologies and trying to understand the logic behind creating “breathable,” porous structures, similar to Robert Stuart-Smith’s (rs-sdesign) Helsinki Public Library project.
What fascinates me most is how the vector flows in this work materialize into a continuous, yet very lightweight and permeable mass.
Hypothesis & Doubts: Initially, I tried to reproduce this using classic vector fields and vortices (Vortex/Spin Forces). However, I’ve started to wonder: am I on the right track? Could this be a complex programmable topology of Minimal Surfaces (like TPMS/Gyroids) adapted to an architectural form? Or is it a hybrid approach — Field-Driven Growth?
I would love to hear your thoughts on the nature of this geometry. I’ve been following discussions like this one regarding the Nautilus plugin, which seems relevant:
Current Progress: I’ve put together a basic Grasshopper script using standard vector components. I can generate the “vortices” and direct the flows, but the result is far too dense and “cluttered.” It lacks the elegant porosity and structural logic seen in the reference.
The Problem: My curves don’t form a “system.” As soon as I try to turn these lines into volume (using Dendro, ShrinkWrap, or Pipes), I end up with either a chaotic mess or a heavy, solid form with no openings or clear “voids.”
Questions for the community:
How can I effectively manage the “voids” within a vector flow to achieve a porous but structurally coherent topology?
What algorithms or plugins would you recommend for “growing” such a mesh? Should I look into Multi-Agent Systems (MAS) or perhaps Differential Growth?
If anyone has examples of similar logic or snippets (even rough drafts), I would be extremely grateful for any guidance on where to go next.
From your reference project it looks like there are 2 separate aspects to the design:
1.A continuous surface topology derived from some methodology of carving into a rectilinear volume.
2.A vector field that is applied to the surface mesh topology with the vectors and mesh faces used to orient and scale a repeating tile shape that creates a surface texture.
Others may disagree, but it doesn’t look like the vector field is a generative part of the design, just an aesthetic surface treatment.
I would focus on achieving a surface / volume that works (for what you are trying to do) as a well ordered mesh and then you can experiment with different ways to treat the surfaces (like tiling or mesh strips, for example).
Edit to add:
Upon closer inspection it looks like the surface texture is generated from mesh strips, not an individual face tiling. Something like this:
Thanks for the link, @old_taz. This is a really interesting direction!
Utilizing a strip-logic approach seems like a solid way to achieve this kind of expressive tectonics. My plan is to generate a base structure using Minimal Surfaces (TPMS) first to establish the overall mass and porosity, then use my vector fields to align the strips with the flow patterns. This might be the bridge between topology and flow I’ve been looking for.
I agree and disagree, at first glance I was OK with your statement, but a Vector Field could be transformed to a Scalar field and then you can extract an isosurface. Here I just used the length of the vector. But you can use a dot product with another field, Heteroptera has some components to do that.
So for the isosurface with Millipede or Isopod you can get that
As @old_taz say you can focus on that, I think Isopod could be very useful to mix geometry with some Minimal Surface
then add some vectors and smooth the vectors on the mesh surface
The advantage of this is that you define some vectors on points, curves … than fixed than and get all the vectors on the mesh. The advantage of my tools is that you can fix some regions not just the boundaries
How To Rhino deal with the organization, there is no link at the moment, it will be 18 and 19th of April 15h CET to 18h CET. (so 6 hours in total). It is a paid workshop.
Thanks to everyone for the insights and recommendations! A special thanks to @laurent_delrieu for the script and the wealth of shared resources! I’m going to spend some time digging into these files and running tests; once a coherent workflow takes shape, I’ll definitely share the progress here for further discussion. My primary challenge now is scaling this ‘digital corrosion’ logic to a massive legacy industrial site - specifically, the conceptual revitalization of a kilometer-long rolling mill factory. Hopefully, some compelling results will come out of this experiment
