Hate to say it, but it's been done a bunch of times. There was a guy that did some Maya programming at Sci Arc for his thesis about 6 years ago. It didn't go over too well.

You can look at any of the blobbers and see some of what you are talking about (the 'forces' in 3D software can alter form).

That said, I am sure you can come up with something new. Just don't have the software 'design' the building, as in create the form, shape, etc., etc. It's all subjective, regardless if you sketch it with pen and ink or type some numbers in a computer (YOU have to decide those numbers, so there is always your hand in the output).
It's been discussed a bunch, try searching through the archives. Look through the old blobby books and you'll see it. Obvious ones are Greg Lynn (and his utter fascination with Maya/Alias's 'forces'), Rashid, and Karl Chu (although, admittedly, I know little of his work beyond it simply looking like abstract doodles, and there's also Marcos Novak).

Just remember, if you are making form, you have put your 'touch' on the project, even in the code written (assuming you write the scripting). You may also want to look way back at Eisenman and others and their interest in the author.

I would read per corell's thread hey all you fancy graphics lovers.

form doesn't 'auto' generate... don't be fooled into thinking you can design a building by "pumping information" into a piece of software. you have to set up the space (history/precedents/theory/site) in which the project exists... otherwise you are just tooling around and you'll look like a tool...

Been there done that.
1985 Sigma Design now called Arris
Was so cool.

You cited acoustics as an example of something that potentially could be auto-generative. Just to build some credibility on this particular topic, I am the designer at an acoustic consultant. From a 3d model, we do room modeling and are able to extrapolate data describing how sound behaves within the room. Based on that data, decisions are made as to which parts of the room require acoustic treatment. In my firm, and to my knowledge, all firms, these decisions are not made directly by a computer, which I believe at some point soon software could become sophisticated enough to do.

Now, that is where any software program will stop, it is incapable of what follows, which is practicality. Room treatment depends on the architectural character of the room - where doors are, colors, texture, material. My point is that there needs to be a human interface here, with a creative mind and an imagination.

A computer model is not needed to predict which types of rooms will inherently have good acoustics. Anyone who has studied acoustics will know the basic rules of thumb, and designs accordingly. This is the same for HVAC, electrical, plumbing, everything. If you WERE to try to create this software, it would be mind-bogglingly (is that a word?) complex, and will be extremely limited at best. I will draw the analogy with a robot. Yes, we can build them, and they perform limited, heck, even say very complex tasks. They have no soul, no creative mind. They can stamp out a series of wal-mart looking buildings in various sizes, and that's about it.

Unless of course, you would rather have a lecture on what you should & shouldn't like.

Did you look into Genetic Algorithms? There is a large body of theoretical models for "evolutionary design" and embryology. Tim de Coster is completely right when highlighting the complexity of "autogenerative" architecture. The code needed to be written to find automatic solutions of specific problems is very time-consuming, and more worryingly, a piece of subjective design itself. What you need to do is create a framework of rules of growth, evaluation and feedback which is general enough to fit any unthinkable piece of design. A designer could be involved in the process of finding an acceptable solution by steering the software in the right direction, but intuitively this feels like such an arbitray process. In my opinion, the touch of the author doesn't have to be there, why does it?

I found a really good resource the other day: "An Evolutionary Architecture" which is the out-of-print book by John Frazer, published as a couple of PDF's. Then there's the writings of Peter J. Bentley which explains genetic algorithms as a general design issue. Oh, and I'm reading D'Arcy Thompson's "On growth and form" at the moment, which is a great classic and endless source of inspiration.

...paste from website: TOPOPT is an acronym for TOPology OPTimization.........The goal of the research project is to develop systematic tools for design of Multiphysics structures using topology optimization and other structural optimization methods.

http://www.topopt.dtu.dk/

Just counting the sceconds before mr. Corell will join in...

Hi

It can be a bit off topic and I wait to add the graphic but back then when I did boat design and had to make the design tools software myself, I made an application to make a 3D morph . Now Morphing is usealy not 3D but 2D , still as I did use this program I was surprised asking myself who acturly drawn this paticular design that was number 4 of 10 inbetweens from one boat shape to another. Now what I realised was as using a typical 1920 design in one end and a typical 1980 design in the other end, that the inbetweens realy followed the design trends as how they were, that the develobment acturly showed in the stepvise changing designs, but that is the off/topic thing.
Anyway you can make an inbetween in 3D of a plane and brick.
Well almost any two opposite designs but who drawn it even it is buildable.

a translator please?

anyways, UN Studio has tried to demonstrate circulation diagrams giving rise to process driven blobiness. Nice renderings though.

So you don't know the difference between 2D morphing and 3D morphing ?

So it's all about author/no author.
Do you really think your thoughts are so completely your own?
Connect your computer to the ether.