Friday, December 13, 2013

Assignment 11

A simple tube geometry was created by placing a circle on the x-z plane and another on the y-z plane. These two circles had their height, horizontal offset, and radius parametrically controlled. A line was drawn between the circles such that the degree of curvature of the line was also parametrically controlled. A "Design of Experiments" was run where each of the parametric values was varied randomly, and a spreadsheet of the value combinations was saved. Each potential geometry could be seen by then selecting a row from the spreadsheet, and applying those parameters. This spreadsheet could be output to a CSV file which could then be run through a genetic algorithm.

Thursday, December 5, 2013

Final Project

This animation shows the logic of the standard node part. It is comprised of a sphere with holes in it. These holes are at the precise angle of the members it joints, and also at the precise OD of the members. This allows the members to simply slide into the node, and become fastened at the correct angles. 

This animation shows the attempt to create a 3 dimensional lattice of these solid parts (members and nodes). The lattice is reused from assignment 7 (a grid of points from the intersection of two fans of lines projected vertically onto a surface-through-point). At this point, a solid member was made, as well as a solid node sphere. These shapes are to be used as power copies.

Difficulties arose. Not least of which being that the logic of power-copies does not lend itself well to the creation of solids in Digital Project. All Solids were required to be the PartBody portion of the tree, which lead to obvious problems, and a very tedious power copying task.

It is desired that the nodes would be output as 3d geometry would could be 3D printed. To this end, the geometry was exported to a .iges file and opened in Rhinoceros 3D. upon opening the file, it was all but unusable. It was fully of glitches, and even when the spheres did render, they were simply spherical without the desired holes.

Asignment 10

Building from the previous assignment where the lines were projected onto a surface-through-point, only lines in the y direction were used. A second UDF was put into the original catalog and through a new knowledge pattern it was used in the design file. Two of the original curves were used to make a series of rectangular facets. These facets were then populated across the entire 3 dimensional surface-through-point. As each facet was made it was assigned a color based on its location in the grid, and each color varied slightly from the color of it's nearest neighbors thus creating a color gradient.

Assignment 9

This is an exercise in creating a UDF (User Defined Feature), outputting it into a catalog, and then using it in a design file. The actual assignment involved creating a series of lines, and then projecting them 'normal' onto a surface-through-point. These lines were run in both the X and Y direction to test my ability to follow the tutorial.

Assignment 8

The power copy framework from assignment 6 was reused. This was modified such that for each power copy, the surface through point had a different point height. This made the surface more or less bumpy than its neighbors. These heights were determined by values which were imported from an MS Excel spreadsheet. Excel has the ability to generate "random" numbers. A sample of potential geometries can be seen in the animation along with the corresponding Excel values.

Assignment 6 Revisited

A wavy grid was made. and points from it were projected in the Z direction onto a 3 dimensional surface. The original Power Copy was created as a triangle with a raised middle bump (surface through point). This original geometry was power copied on the raised point grid. The shape of the underlying 3 dimensional surface was modified, and the grid was tweaked to yield dynamic changes in the overall form.

Assignment 5 Revisited

Two shapes (a square and a parallelogram) were made in separate part files, and then imported into a product file along with an angular framework. When the framework is changed (left window) The product file dynamically changes and "folds" itself up.

Assignment 2 Revisited

Starting with a revised grid, the pillars were made with a common multi-section surface creating their shared rooflines. The grid is altered, and so is the multi-section surface which yields a new column shape and roofline.

Saturday, November 2, 2013

Assignment 7

This didn't work out when implementing in Digital Project. The goal of making all of the pentagons planar was abandoned, and hexagons were adopted as well. Here we see the implementation of the hexagonal surface:

(1)Crossing Fans of lines were drawn. (2)The resultant crossing points were used to make a grid of points. (3) Based on the outside boundary of this grid along with a single passing point, a fill surface was created. (4) The appropriate grid points were then projected onto the fill surface yielding a 3-Dimensional grid. (5) The first six sided pyramid was built on this grid. (6) This Pyramid was then power-copied across the 3-D grid. (7) The underlying grid can be modified by moving the fans, changing the angle between fan blades, or changing the fan rotation. (8) A new 3-D geometry is created once the changes cascade through the tree. (9) In yet another  geometry we can see some interesting behavior; the pyramids are sometimes generated in a concave manner, and sometimes in a convex manner.  All pyramids can be switched by changing the height value from a positive(+) to a negative (-), but no easy fix could be discovered to keep all the pyramids going in the same direction.

Finally, an error message always appeared when using global variables:

What can be done to eliminate this?

Saturday, October 19, 2013

Assignment 6

My power copy can be seen on the top lower left. Instantiated through the pionts in the background. The next instances of it do not incorporate the bump. When I modify the underlying framework, the bump gets lost on the original power copy as well.

Thursday, October 10, 2013

Assignment 5

This is what I'm trying to accomplish in Digital Project

This is as far as I got. I'm having problems getting the pieces to snap to the framework. I need to constrain the framework line (1) to the plane that the main square is also constrained to. Then I need to constrain line (2) of the framwork to the line edge of the parallelogram (3). So far, nothing attempted yields anything but error messages.

Friday, October 4, 2013

On Emergent Formalism

Greatly simplified, the idea that a complicated thing can be created by the interaction of many simple processes is referred to as emergence. From complicated ripple patterns in sand, to the quasi-synchronized movement of a flock of starlings simple rules make for complicated patterns. In parametrically designed architecture the a reverence has arisen for the emergent complexity. When simple rule sets (parameters) are tweaked until a complex form appears which the designer considers aesthetic and/or functional the design is deemed a success. The critique against this is that emergent design is just formalism masquerading as intelligent process.

Then there is the practice of iteration: Design, test/critique, and repeat. This always yields successively better results. Natural selection is a type of iteration. Emergence and iteration are entangled, but they are entirely separate forces. In the cognitive space between these two concepts is the idea of a feedback loop where two forces act to change each other. Each exerts an influence on the other, and each is changed by the other's influence. A Feedback loop can be present in emergent systems, but it is not necessary. Feedback is necessary for iteration, but a feedback loop may not exist. When talking about such a tangled and nebulous subject as emergence many things which are not truly in the subject matter can tend to get conflated.

Thursday, October 3, 2013

Assignment 4

Week 4 - Multi-Part Assembly

An animated version of the final functional version

What I was trying to Achieve, as modeled in Rhinoceros

The problem I ran into in Digital Project

Thursday, September 26, 2013

Assignment 3

Responses to a few passages from the text "How Do Simulations Know?" by Loukissas

“Being in control of the machine can seem like low-level work.”
Yep, but being in full control yields some pretty great designs. Just like everything else, the devil (and the value) is in the details

“architects try to reconcile the dispersed knowledge of many specialized designers.”
Being able to do this well is the mark of a good designer, or team. The Day that computers can hold a holistic picture in memory, and simulate all repercussions of a building’s design (including political ramifications, and sliding scale of funding based on stylistic considerations) to arrive at a final design with an ambiguous definition of "optimal" will be a day that humans become largely obsolete.

 “…in most cases, a computer simulation is just a verification of what he already knows”
The magic happens when you don’t completely know what the simulation will tell you, and your next step could not have happened without the simulation. Being able to simulate the simulation in your head helps you be more efficient, but in the end there is no substitute for iteration when it comes to designing.

Regarding the passage about engineers spending six months building: The word Architect is derived from ancient words meaning “master builder” it is a shame that in this day and age knowledge has become so specialized that a person finds it difficult to both design and build. Most people don’t have the time to be a builder and a designer. They choose one or the other as a career.

Regarding the tailoring of simulations to the client: Simulations are sometimes sales aids. This is what renderings are for. An architect might use a rendering as a final check to see if the design looks good when represented photorealistically, but it is a final check and not really a design aid.

Regarding “total architecture”: Achieving it is kind of a siren song. If something labeled such comes into existence in the future, I suspect it will just mean that the realm of possibilities has been so constrained through building codes and industry practices that an entire building can be simulated in a virtual environment. Eventually (hopefully) someone will break out of that box, and the “total architecture” model will fail to encompass the new way of thinking.

Friday, September 20, 2013

Assignment 2

The first image is as far as I got. The next image shows a weird error I get when trying to extrude "up to element" and it all breaks down. The third image shows how some of my associations don't stay associated. When I try to edit the associations I usually have catastrophic failure of the model, and have to pres ctrl-z a bunch of times. The fourth image shows the success I had in remaking the original assignment to be used in this model. However, sometimes weird stuff happens. On the right side of the green model we see that two points didn't move after the parameters were changed. It was working fine, and then seemingly without reason these points just didn't move with the rest of them.

Thursday, September 19, 2013

Thoughts on “Design Development Environments” by Daniel Schodek

CAD, in the hands of an unskilled operator can be used to make some very interesting forms, but as the article states “It is often more difficult to understand what has been created than to generate the surface itself.” Meaning, the model is supposed to represent a real-actual object that is going to be built. What does your CAD shape represent? Can it be built using materials at hand?

One way of building involves developable and nondevelopable surfaces. “A nondevelopable surface…requires cutting and/or stretching if it is to be flattened out into a planar sheet.” A developable one being the opposite, and being easily turned into a flat 2-dimensional sheet. Of course this terminology is indicative of the pervasive building paradigm of the last hundred or so years since the mass adoption of mass produced sheeting materials such as plywood and sheet metal.

Sheet metal unlike plywood can be stretched into nondevelopable surfaces. This entails Gaussian curvature. Positive Gaussian curvature (synclastic) means concave or convex shapes, and negative Gaussian curvature (anticlastic) means the curvature switches from concave or convex like a saddle. These terms seem to be from the math discipline, and not really used by designers, but used extensively by the smart, SMART, folks who program our CAD software.

No matter how smart they are, programmers are not designers. And it takes a special kind of mind to be a good designer when using CAD. Many of the modeling techniques talked about (especially Feature Based Model Building) require an extensive foresight. “…building of the model and the definition of key dimensions must frequently anticipate design changes that inevitably occur during the development process.” The model must be set up correctly in the beginning as future limitations insofar as altering/modifying the model will be present. Of course not all things can be predicted by the mistake prone human brain, and hence “In practice it is not uncommon to rebuild a model completely as the design progresses, because detailed technical questions necessitate changes that were not initially anticipated.”

There is another job that good Designers do. I’m talking about designers who bridge the gap between design and engineering. They design with process in mind. They draw CAD models while always having in mind the real-world procedures and tools which they will use to actually build the thing. This is the kind of thing the software is beginning to get to, and it was talked about in the section called Application-Oriented Modeling Techniques. This is the area of the greatest deficit in CAD, and improvements here will embody the Future of CAD. I imagine a day when the CAD Designer steps into a virtual first person environment and dictates a stock material, and its dimensions. This then opens a toolbox of actions which he/she can perform on the material. These actions will be direct allegories to real-world processes. If the designer tries to do actions which would lead to manufacturing problems, the program will either model these problems (Like cracks in the material) or it will display an error warning (Like “the draft angle of your mold makes the part unable to release from the mold”). Of course, when the software does all that, what is to keep it from doing the actual designing as well? With the software designing the self-driving cars, will there be any use for us human beings? At least we’ll make great pets.

Note: All quotes came from the chapter itself.

Thursday, September 12, 2013

Thoughts on “Keepers of the Geometry” by Yanni A. Loukissas

Digital processes have changed creativity.  First of all digital modeling is not simply simulation. The designing process happens through the modeling process. This feeds into the psychology of the practitioner. People can potentially see the digital model as fixed, and unchangeable. They may say things like “That’s how I drew it, so that’s how it will be.” But everything is changeable, and this attitude speaks to the human resistance to change and not the nature of the technology. Especially in designs which we conceived we are reluctant to discard aspects despite the likelihood of finding better solutions. Seeing a design manifest on the screen makes it that much more solidified, and iterating is hard. Knowing what aspects to attack, and what to leave in is hard, but that has more to do with human nature than the software. Change is easy in a digital model.

Digital modeling is an epistemology. It is a worldview. It liberates through its functionality and capacity to represent, but it is also limiting. The way that you must plan to build things in a program shapes the way you think of the thing you are building. Each program has a certain paradigm driving its digital modeling environment. This shapes everything that comes after it. The dividing line where most programs start is Nurbs vs. Polygons. Rhinoceros is a Nurbs Modeler, while Catia is a Polygon based program. If a person where to approach a loosely defined conceptual project in Rhinoceros the eventual outcome would vary greatly from what they would create if they had started with Catia. This is the same as an oil painting looking very different from a watercolor painting of the same scene.

Is the digital model the final product, or is it the built architecture? Could a sufficiently skilled and cognitively gifted designer get to a final built architecture without the models? If they could, it would not be as rigorous or well thought out as the design modeled on the computer. That being said, it appears that being a technician with a highly developed digital skillset precludes you from working on more general ‘managerial’ tasks. In other words, by being an expert at 3D modeling you pigeon-hole yourself away from becoming the boss. Then again, there are such things as “Masters of the Virtual.” This term refers to a specific type of competition architect who does not actually implement their designs. So you could be the boss in such a practice based on your digital skills. Despite it being a shared goal by most, we can’t all be the boss, so becoming as skilled as you can at the digital is probably a prudent endeavor in this ever changing technological world.