Tuesday, 16 November 2010


Dear internet, 

     Without being too long winded about it, I'd like to introduce the blog as an ongoing repository of computational design sketches. Most of the content will likely be early stage works in progress in an effort to prioritize driving logics and underlying processes of spatial organization. To get the ball rolling, heres a link to some of my recent work on Vimeo.



  1. Hey Davideo,
    This simulation is great and I am very interested in the logic behind it. I will tell you what i think is happening and I would really appreciate if you could correct my assumptions. You have a brep on which you place random points, then you connect those points x distance toward the center of the brep if they are in a certain sphere of influence, you repeat that process which yields fewer lines everytime until there are two and finally those two are connected. My question is, how can you be sure that you will have the appropriate number of points per generation? Thanks

  2. Hi Paula,

    You're really close. Rather than a sphere of influence, each point finds it's closest neighbour in the current set. The two then converge towards the brep centroid with one of them being passed to the next iteration. So to answer your question, as long as there are multiple points within the current set, it can proceed.

    To expand a bit on your question, things can get a bit trickier if you wanted to pair or couple points. In the case above each point is free to converge more than once per iteration. For example if ptA's nearest neighbour is ptB and ptC's nearest neighbour is ptA then ptA ends up converging with both ptB and ptC. With coupling on the other hand, ptC wouldn't be allowed to converge with ptA because ptA is already with ptB. This requires considerably more smarts within the algorithm to avoid distant stragglers coupling when everyone else nearby is spoken for (http://www.lavalife.com/).