My vision concentrated on issues relating to the poor visibility/readability of the overall area (and the necessity of a landmark) and the lack of “free” public urban space. The functional program was mainly derived from the observation that most historic centers (Lipscani being a prime example in this) loose their initial cultural and traditional economic value in favor of the over-dominant bar/cafe. This leads to a certain repetition of fluxes (economic, cultural, pedestrian) which harms the respective area.

Geometry-wise, my proposal respects local aspects of the area, namely its porosity and the way the urban tissue coagulated around narrow winding streets and small interior courtyards in a constantly surprising lattice.

In respect to this approach, i considered the lot as a solid volume on which forces are applied in respect with the openings and general director lines of the site. Performing a “structural” analysis on the site revealed the patterns by which these forces would naturally flow towards given points of rest (namely the designated openings of the courtyards).   Using topological optimization techniques these patterns were transformed into geometry which was later subtracted from the original body thus giving the overall shape of the building.

By following this design method I ensure an optimal circulation flow through the built site, encouraging interaction and furthering the development of the local urban tissue in a manner very close to its characteristics (gained by spontaneous evolution) thus fully integrating the new implant.

That’s wrapper of the last month, give or take.

Grasshopper’s scripting capabilites are now a huge super cool feature, especially if you’re coding in visual studio (c#) and then pasting the shit or making dlls. I did this some time ago when i suddenly realized that you can declare your own static variables inside scripting nodes (and these guys hold up their values during the updating procedure). And the vector math is already there too.

This is an implementation based on Craig Reynolds‘s neurons, and a little bit on Daniel Shiffman’s processing example (which was updated and used some time ago for this.

Now, if somebody will be nice and implement some collision detection…

PS: I will upload the def. soonish (i have to clean it up a bit first). Ok i didn’t clean it up; here’s the link.

Here’s a new 10 day link: http://dl.transfer.ro/boidsT-Transfer_RO-16Nov-ad56f6.zip

Final link, stable and sound: http://improved.ro/Grasshopper/boidsTCAV.zip

The line-up includes:

Ioulietta Zindrou, Maite Bravo, Verena Volger, Luis Fraguada (of Barcelona/IAAC fame);
Yours truly Dimitrie (of Bucharest fame);
Henning Rambow, Patrik Bedarf (Leipzig crew);

I haven’t been that productive lately due to the many things happening all the time, but hopefully I’ll be able to get back on the research track in summer.

Deprecated. Here’s the new version.

More as a scripting experiment, when i was mucking about trying to make the delaunay triangulation work in grasshopper i somehow found the wonderfully complex qhull library which i promptly set to push and pull to get it to work with grasshopper. As advised on their website, the best way to do it is to call it as an external program, which is exactly what i’ve done:  no files are written or read, no dos windows pop up, everything’s smooth.

Given that you don’t have many complex operations in grasshopper after the solution is generated, you’ll be able to handle quite an impressive amount of points (say 200 on my three-year old toplap) in real time. If you add the simple planarSrf operation, then say 60-70 and it gets sloppy.

What you’ll need to do to get things rolling:

0. Download the 3dvqhull definition and example file, and remember not to use it for commercial purposes, share-alike whatever you do with it and take the time to give the proper credits: 

1. Download qhull, and unzip it in a folder of your choice.

2. Get going and search for “System.dll”. What you’re interested in is the 2.0 version which you’ll usually find in here: “C:\WINDOWS\Microsoft.NET\Framework\v2.0.5[...]\”. If you can’t find it, I’m amazed grasshopper works for you. Anyway, you can find and install it from here.

3. Add the newly found System.dll version 2.o as a referenced assembly of the qhull component in the definition file.

If it turns orange, it’s cool.

4. Write in the panel that is linked to the “path” input the full path to the qhull program qvoronoi. You don’t need to add the .exe extension, but you can do it if you feel confortable.

5. There’s just one more thing you should know: facets that contain the infinite vertex are omitted altogether, without remorse. So as to have as little facets tending towards infinitum, I always add the corners of the points bounding box to the input sites.

You can scale the bounding box in respect with its center, or you can just call the whole thing off – it’s your choice.

I think this just about covers everything. Take care and have fun.

didi out.

PS: Qhull does more than voronoi. So if you have the time to explore and test, please do – the package is very powerfull and it can be used for more than this.

I talked about some projects I did (some together with Veronica Popescu) over the last two years, always insisting on the computational approach and detailing the techniques used.

I presented scripting approaches and more interactive (parametric) grasshopper based modeling. Some  Processing based approaches were presented and I insisted on the necessity of using multiple open tools that can be made to communicate between one another.

I tried to clearly mark out the relationship between fabrication and the overall design – how material properties and building techniques inform the design and viceversa.

Reactions were good, and it seems computational approaches are definetively stirring people up.

Thanks are required to Patrick Bedarf, who made this possible and is proving to be an excellent host in Leipzig, and also to prof.  Henning Rambow for all the support.

Also if you are reading this from Leipzig and are interested in joining us tommorow at 16.oo @ HTWK computer pool for some hands-on grasshopper work, there’s  no one stopping you.

We are using the excellent space-partitioning properties of the Voronoi algorithm to create a lattice of interior and exterior courtyards that progressively make a transition from public space to semi-public, semi-private, and, in the end private space. This way we propose the creation of a coherent urban lattice encouraging interaction with the busy city outside while in the same time offering various degrees of protection/”cosiness”. We are not proposing a new type of urban tissue in itself, but merely adapting and reinterpreting some qualities of sponaneous developement and local tradition(“fundatura”) in a flexible (yet highly accurate) digital framework.

As I was saying at the begging of the post, this is just a technical showdown – everything you see rendered above is far from what the end product will probably be  (since also the way the assignement was formulated forms that evade from some general bar-variation are quite difficult to fit in the judging criteria). Yet it indeed demonstrates the versatility of parametric/generative architecture and the capabilities of the digital framework proposed to control and manipulate with accuracy all aspects of the project (from raw geometry and algorithm to technical: floor height, floor areas, built/unbuilt ratio, sun exposure and so on – the possibility to expand the definition to apartment level details is definitively there).

That above graphic  is a watered down version of the grasshopper definition (I will not release it under any license for now).

For those who really like to play with voronoi regions and stuff, here’s the final grasshopper definition file (right click, save target as – else you’ll get a ~250kb of useless xml in your browser window).

The vcell component outputs now individual cells as closed polylines and closed nurbs curves. This is useful if you are using this for some urban project like i am i would have liked to, mostly due to easy offsetting and area calculation possiblities – included in the definition file – or height extrusion – included as well. You can even go further and color (using shaders) each cell coresponding with its mass/area/height etc.

We started out on this by analyzing the area and its surroundings from the point of view of circulation (more on how we did this). How would people move around? The conclusions from this study where integrated later on in the project.

Next we took on the task of (re)creating the landscape – modify the terrain in a functional yet unobtrusive way. Formal concepts revolved around radiolaria, foam, water, waves etc. Luckily we had a working circle packing applet in processing ready (which was used for this). From here on, it was quite straightforward: expand the circles to smooth spherical caps and smartly transform them into urban furniture.

The interstitial space that remained between the”bubbles” is packed with wood alongside the routes we discovered to be ideal using the circulation study. The rest is English lawn pure green smartly-cut grass.

This project is a collaboration between Veronica and me.

Tools we used include:
Processing was VERY important early on in the design process – we used it for the circulation study as well as for the early circle packing experiments (size, density, spread etc).

Rhino and RhinoScript: RhinoScript is great – we used it extensively (create spherical caps based on the generating circle’s radius, expand circles, contract circles, import circles which the processing applet generated, etc.). Rhino was used for everything else – 3D modelling and, of course, making valid STL files for the 3D printer and also exporting the right things for the laser printer.

Laser cutting was done here (as always).

3D wax models were kindly made for us by mazarom (at the moment the only 3D printing service in Bucharest). If you need a complicated model, don’t hesitate to contact them!

Plotting the final presentation was done at studio spot. They don’t have a webiste…

parametric variation of given frame (lateral walls) in given limits. each house becomes uniqe, escaping repetition. hints of emergence.

Basically the facades are independent from the core. Each lateral wall (calcan/blind wall) starts from an ideal shape that is subtly altered by a small script. This creates variation, as opposed to the usual repetition of a terraced housing lot. Mass production of 18 identical houses becomes mass production of 18 identical and UNIQUE houses. I would like to stress that: Mass production of 18 identical houses becomes mass production of 18 identical and, in the same time, UNIQUE houses.

There. Now I feel better.

The project’s challenge was to transpose the notion of perfume in an architectural form. Wikipedia suggests many trails. Chevalier’s Dictionary of Symbols suggests some more. Balzac: “Tout parfum est une combinaison d’air et de lumiere.”; Hugo: “Le parfum est de la lumiere.”. Floral motifs, blobitecture, incense and religion, drugs – it was up to you to decide on how to visually speculate perfume.

I was fed up of the meanings and interpretations of perfume. Perfume nowadays is synthetic. Leave me be. Smell chemistry. You can’t contradict that.

I broke the visual representations of the chemical components of a standard perfume (chemical representations have very strict geometric rules – 120 deg angles, pentagons and hexagons) and got voronoi regions.

Ahhh, sweet metaphor of computational architecture. Isn’t it beautiful? Elegant? Yet it has nothing to do with architecture. Form is detached from function and we are approaching a similar crisis of that of modernism, when the exact opposite happened. People don’t find themselves included in such designs. Computation that strictly refers to form/ellegance alienates architecture from its goals. I don’t trust the starchitect theory as being the solution; it won’t last or it will split us into sculptors and engineers. 

(Please do read that article. It’s a must. It shows past (as in blobitecture and zaha), present(not the “so last year” deconstructivism or – worse even – post-modernism) and pushes through to the question posed by the future.)

Enough theoretical blabber. I’m giving all this computational business some thought. Code in architecture is nice, as long as it serves the people, not just ellegance. (Please do notice I didn’t use the term “function”. I prefer the much more accurate and revelatory word “people” = complex emergent system.)

Here are some pics of the model:

Programmed in Rhino. Finished in Sketchup. Model made in ManualManufacturyCAD(hand glued). Laser cutting here.