Final Thesis

Thought I would share this with everyone, seeing much of it was done with grasshopper, python, and other tools. It is a look into data driven architecture and how we can start to change the process of design from one of forms to one of frameworks. Meaning, rather than the architect designing forms, the architect designs a framework that produces form based on contextual and user data.

Please feel free to comment or ask questions on the document page as I would love the input. The online document & video can be found at:

http://zachsoflin.com/2012/04/20/data-driven-architecture-2/

sorry if this is the wrong place to post this, let me know if I need to change. :)

Replies to This Discussion

Permalink Reply by Sebastian L. on April 26, 2012 at 1:53am

Hey,
wow, this is amazing, a really cool approach for using parametric design! I like the idea with the data streams and beeing surrounded by them at all the time and making sth out of them! It is also really good presented and animated in the video and also the brochure! The video, is it done with After Effects? Good job!

One question, the base consists of 6 modules and is expandable if the user want to do this, right? Don't you get some problems with the light?

Anyways, hope you passed your final thesis well? ;)

Permalink Reply by David Stasiuk on April 26, 2012 at 3:07am

Outstanding work, from concept through investigation, development and presentation. Really intriguing ideas!

Permalink Reply by Zachary Soflin on April 26, 2012 at 11:48am

Thanks David, it feels good to have it done! :)

Permalink Reply by RWNB on April 26, 2012 at 3:33am

very interesting project.

what i miss, is the documentation and designprozess with grasshopper.

Permalink Reply by Zachary Soflin on April 26, 2012 at 12:01pm

RWNB:

I am actually in the process of producing this as I decided that it didn't simplify the idea to include it in the Final Thesis Document. See my response below for the general process (for the tower portion) and the resulting component library. The library connects grasshopper to Fusion Tables which is a online spatial database program. The components enable you to perform spatial queries from within grasshopper to return values or geometry from the database. For example: "return all values within .1 miles of 41.5235,-96.3432". You can also upload SHP files which enables you to actually use values from GIS databases and map the geometry within Grasshopper. The components also allow you to write data to the database. This is the basis for the project in that it provides a universal platform to store and access data about the users and the context.

Any questions, let me know.

Thanks for the interest and kind words,

Zach

Permalink Reply by RWNB on April 27, 2012 at 2:51am

dear zachary,

many thanks for your detailed response.
i am now convinced, that a comprehensive documentation
of the parametricismus of the work is necessary.
absolutely, that's the point of your work. (:

many thanks for this interesting work and insights.

Permalink Reply by Zachary Soflin on April 27, 2012 at 8:01am

RWNB,

I agree, that is my goal the beginning of this summer. Thanks for your interest.

Zach

Permalink Reply by Zachary Soflin on April 26, 2012 at 11:46am

Sebastian:

Thanks for the kind words. The way it works is a user first starts by designing their unit (which is a android app for tablets you can find here: https://www.dropbox.com/s/4hkknjpznnvaxxv/CondoCreator2.apk). Through this process the user lays out and programs his/her unit, and selects priorities that are important to them. Once they are done they submit their unit. This uploads the unit to an online database (Fusion Tables) which is connected to grasshopper through GeoCloud (A grasshopper component library I will soon release: See Image Below). Grasshopper then pulls the newly submitted unit from the database and places it on the tower. Once the majority of units are received, the units are optimized using the priorities specified in the application by the user and other constraints. Since each unit is limited on either side, each unit maintains its view outward no matter what. Views to either side are controlled through a bidding process. As users specify there preference for views outward, a bid amount is also submitted. Through the optimization process, the first goal is to meet every units preferences. But if this isn't possible then the optimization process prioritizes the unit with the highest bid for that particular view. No matter what though the unit still maintains its view outward, providing the unit with at least 36 linear feet of view outward. Therefore light shouldn't be a problem on any of the units. The process is much more involved, but hope this sheds some light on the idea Sebastian. Let me know if you have other questions.

Zach

Permalink Reply by Sebastian L. on April 26, 2012 at 2:04pm

I am speechless! :-) amazing! I will check out the app later on and give you some feedback! This is marvelous!!! Thumbs up man! This poroject seems to be reall really well-thought-out!

Permalink Reply by Michael Pryor on April 26, 2012 at 2:21pm

impressive

Permalink Reply by Oliver Sjöberg on April 26, 2012 at 6:35pm

Is the optimization in the video just a mockup or is it actually working? If the latter, is it based on Galapagos, any own algorithm or some external multitarget evolutionary solver?

Fantastic piece of work you've achieved! :-)

Permalink Reply by Zachary Soflin on April 26, 2012 at 9:31pm

Oliver,

Are you talking about the drawing of distances to the surrounding buildings?(base) or the optimization of the units in the tower? Both are a capture of a period of time of the actual process. Obviously, the process doesn't really play out in quite as graphically interesting a way. But it is a graphical depiction of what the program is doing. Part of the challenges of describing how this project works is that most of the logic is non graphical.

1. The base process shown in the video isn't an optimization process, it is just showing the live submittal of surverys. So as people submit their preferences for walking distances, it is measuring the distances to surrounding buildings, searching the area for similar programs, and determining the programs that are needed. The process shows how the definition is determining the size of the user group, and what types of programs needed and how big they should be.

2. The tower unit optimization process is a combination of a python algorithm and galapagos. Each tower is submitted with certain parameters and priorities. Grasshopper connects to the database and assembles all of the units. Those are then placed at a default location on the tower. Each unit is measured for its performance. This assigns a rating to each unit on how it fulfills the users specified priorities. All of the units ratings are added and that number is what galapagos is trying to maximize.

Hope this answers some of your questions, like I said above, I am working on documenting this process as it is quite complex, and divided into many pieces.

Thank you for your interest and kind words,

Zach