Here is an example, making circle as trim region and trimming the rectangles to separate inner and outer geometry. It worked fine for two rectangles;
But with more than two rectangles it got really messy;
3 Likes
This is actually part of my master thesis, but could be an interesting idea to explore plan solutions. I started by defining the intended relationship between rooms with pen and paper. For example Library (“Lesing”) should connect to the open space (“Void”), and so on.
I placed the rectangles in a way that roughly resembles the connections, with the Void as gravity center. Here I’m showing 2 different output; the only change being a rotation of the ellipse (Void). Pay attention to the porosity of the different spaces.
Right now it feels a little messy, but I think it could become really interesting after applying some cleanup-by-grid function. Let me know if you have any thoughts on this approach.
(Had to look up the definition for this word)
Why are the spaces all rectangular, but the one you call “void” elliptical? 
I quite like the porosity and the less regular overall layout that thus emerges.
Once you’ve placed your rectangular spaces, you could for instance only keep the circulatory core (void) and the boundary, and reevaluate the interior:
I’ve seen something like this being done before, where you extend each segment of the rectangle until it intersects the furthest other segment in the collection of rectangles.
Here, I’ve drawn it so that intersection lines can’t lie outside the overall boundary curve, which keeps the established boundary intact.
Next you want to re-establish an irregular grid of cells from the intersection lines and boundary curve.
Finally, these cells can be reassembled into bigger spaces. It’s probably easiest to first group those who fully lie inside a previously established space. The remaining ones can then be distributed with a different rule set (maximising areas, minimising sharp corners, etc.) .
This might be harder to do than perceptible at first glance. It’s easy for a human, but translating it to Grasshopper might be like opening Pandora’s box. This shouldn’t keep you from trying though. 
1 Like
Wow, that’s really interesting. I really enjoy this semi-controlled experimentation with space formation. The question of what should be simplified/controlled and what should be left irregular are super interesting and also difficult to answer.
I quite like the rule of intersecting lines inside the boundary, and reevaluate the space. As saying, this is the boundary, now let’s re-discover the interior. Really interesting idea, will definitely explore this going further🙏
@julian.riise the existing tools are frustrating, so kudos to you for making your own. This does remind me of 1831 though!
11 Likes
Frustrating in that they don’t do everything automatically and you have to think for yourself? 
It’s not as straightforward and seemingly funny as the strip portraits though!
I had plenty of people tell me you can’t do this and you should do that, but in the end everybody just knows their subjective “truth”, which is mostly objectively wrong.
I’d say the guy on laptop was right for at least trying and making it a learning experience. The other two are sheep representing the status quo that fears change and wants to hold anybody else back.
2 Likes
So true
2 Likes
I’m looking for a way to scrub through iterations on a design in a way that is useful. At this stage I have minimum and maximum roof height, and randomise them with a seed slider.
Any ideas on how I could make this more useful?
Integrate a performance metric. Is there a limit to the combined building volume? Is it related to cost? Random variations are only useful to suggest starting points
2 Likes
Excellent suggestion. This is an interesting and seemingly complex topic, and I’m excited to try out some things in the coming months.
Here is a short video of a lil workflow to measure connectivity between buildings, and use that feedback to inform me while planning. The number is just an average walking distance between the center of each building (using Delaunay edge).
Super psyched for PlanFinder, available for Rhino and Grasshopper, by @jeroenvanlith !
Check out his website: https://www.planfinder.xyz
There will be a live demo on Youtube in 8 days (5th April), catch it here:
This is the first competitor to Finch3D I believe…
2 Likes
I’m working on something really cool, this is a sample size. Ideally, the function can take a multitude of furnished rooms, with more than one gravitation point, and respect void areas. More to come soon…
Regarding furnishing, for now it seems most effective to draw up example plans first, and modify the plan realtime. However the holy grail would be for the program to suggest furnishing and walls/doors/windows based on the output of this model.
Maybe I’ll get it ready before Elon lands on Mars 
Do people want this project open source? I’m way to small for soloing this project, and tend to favour collaboration.
2 Likes
Looks cool.
Or one that would have external (e.g. sun, climate, laws, norms, renewables, etc.) and internal (e.g. furniture, plumbing, electrical, AC, heating, etc.) influences as inputs, and the flat or house would emerge as a weighted, optimized compromise between all.
It’s always a gracious idea to give something back. And keeping your projects under wraps is a dated concept anyway, if you ask me.
1 Like
PedSim (Pedestrian Simulator) seems like a really useful tool for figuring out circulation in projects.
I’m actually able to modify the plan realtime and see the changes affect how people move.
Awesome free plugin by Peng Wang.
1 Like
Check this out:
PedSim + COVID-19
1 Like
I uploaded a video demo of the latest PedSim definition. It allows Grasshopper users to define start- and end-points and simulate movement. Boundary curves will be avoided, allowing for users to test architecture plans.
Youtube video:
Sharing another project, the plan generator using clustering simulation with Kangaroo. It’s a bit of a messy definition, feel free to change at will.
Youtube:
Github:
3 Likes
This tool is one of my favourites, but also the most buggy
Input multiple curves and rectangles, and the definition outputs inner and outer walls. This is very early in the making, I would prefer it to be more stable, and also being able to include doors and windows at locations specified by the user.
Youtube:
Github:
Download, play and contribute if you like!
1 Like
Take a look at
https://twitter.com/topologicBIM
It might be of interest as it’s specifically for the hard task of intersecting room boundaries.
1 Like