Happy Holidays! 2008 holiday card design. LIVE WIRE image is used for the motif. If I show this piece to SCI-Arc folks, they would give me CREATIVE comments like...
"composition is not working" "color is not matching" "eyes won't be directed in a way you want" etc etc...
Well, it won't be the case if you are working with engineers. Regarding graphic design and compositions, they fall for whatever I do... Ha! so easy :P
Usually, CATIA parametric modeling feature is to see variety of deign or solutions, but I used it to CHECK if I modeled this complex surface as close as possible to minimal surface.
If I update sets of parameters and the surface deform undesirably, the geometrical relationships I set up should be false. By the time I get the surface update maintaining smooth tangency, it should be fairly close to minimal surface condition.
LIVE WIRE, sciarc gallery opening by OylerWu Collaborative.
Network of 1" diameter aluminum pipes -looking challenging enough to stand up on its own weight- are linking the floor level up to the walkway above the gallery space. The stair portion and wing portion are structurally independent from each other, yet all the looping pipes weave together, so live load transfers to the entire piece and vibrate the spanning wings. I think LIVE WIRE is matching name. I have been calling this piece "stairs" the whole time though.
Hardest thing for me was to prepare a CATIA model for two different purposes: one for engineering analysis and the other for fabrication. The former requires to all the bars start and end at points (point to point) to visualize the load path and deformation, while the latter has to consideroffsetfor pipe thickness and fillet at corners.
Research project on thermo bi-metal with DoSu Studio.
The problem is paracloud uniformly stretches a pattern to distribute it over grid, so a user -as far as I know- do not have control over HOW one unit should be aggregated. In this case, a pattern has start and goal condition and has to modify itself gradually.
Image1 shows how paracloud create undesirable configuration. Image2 shows how a unit should be aggregated over the surface.
Fabric pattern completed via powerCopy Automation. Method is the following three steps →
'--------------------------------------- STEP1. BeginInstanceFactory: to initialize the InstanceFactory with the reference and the document where it is stored. This step must be called once at the beginning whatever the number of instantiations are done.
STEP2. It is composed of five methods that must be called in the order. This set of five methods can be called in a loop in order to make several instantiations.
a. The method BeginInstantiate: to initialize all data of the reference. b. The method PutInputData: to set a value to any input. c. The method GetParameter: to retrieve published parameter. d. The method Instantiate: to recreate the reference. e. The method EndInstantiate: to indicate the instantiation is done.
STEP3. EndInstantiateFactory: to end instantiation and cleaning the InstanceFactory. When doing several instantiations in a loop, this step must be called just once at the end of all instantiations. '---------------------------------------
The original design (physical model) of the pattern. The geometrical performance of the pattern (in CATIA).
Wrote series of functions to set boundaries for point cloud selection. For the sake of changing colors, return is currently string points, but it's easy to change the result to array points, or even both.
INPUT: str point cloud, str target point, dbl to set a boundary RETURN: string (or array) points within a boundary
Reviewing my old script after learning Function and Vector.
This is a code I wrote originally for FEA model of a bridge. Using Rhino.VectorCrossProduct, I finally managed to draw normal line to a spline (at a point). As print screen shows above, input is simply a core spline of the bridge.
Assignment from Satoru's workshop. He provided basic For... Next framework, and by modifying conditionals and/or looping, students can create more complicated grid systems.
image from left Level1: Diagonal Lines Level2: Orthogonal Grid Level3: Diagrid Level4: Gradient Density I worked up to Level3, but for the final challenge I needed help, and was surprised how small modification I needed to make.
Aug/29 (Fri). Opening event for OylerWu Collaborative's Pendulum Plane at the LA Forum's new Gallery on Hollywood Blvd.
LA Forum hosted a competition in spring 2008 to design a display system for future exhibitions.In the open gallery space, 16 welded aluminum units are suspended from the ceiling and each allows wall and mid-room display by pivoting down hinged frames. If you grab one of the hanging frames, you can see the unit sway in counter-balance. In the network of aluminum, I liked the use of 2-inch and 1-inch diam pipes. This choice of material creates fine appearance.
Those complicated shapes are produced not by professional aluminum fabricators, but by the team of OylerWu Collaborative and sci-arc students. I was amazed at their level of craftsmanship once again since their previous exhbition, Density Field.
Pendulum Plane Working in Progress 01 Pendulum Plane Working in Progress 02
Finally got hands on CATIA scripting. Most confusing thing about CATIA vba is the tree structure. Within Part Document only, CATIA has the following structure.
And also, since CATIA has multiple ways to create a geometry, "Dim ...... As ......" statements get a little tricky. Point Geometry for example, has 9 types including intersection point. HybridShapePointBetween HybridShapePointCenter HybridShapePointCoord HybridShapePointExplicit HybridShapePointOnCurve HybridShapePointOnPlane HybridShapePointOnSurface HybridShapePointTangent HybridShapeInterSection
Anyway, a lot more to learn. I hope to post powerCopy script next time. What I have now is just CONTEXT of what I want to acheive.
Here is the link to the script tasks in order. Here is the link to ONE MODULE (one of the tasks).
コンサルティングを傍で眺めながら私も CATIA model development に参加した Voussoir Cloud. 8/8 Fri のオープニング、BH でこの installation を担当したエンジニアも来ていて満足そうでした。次の sci-arc installation では CATIA 担当が私なので10月の完成が楽しみです。
Current sci-arc exhibition Voussoir Cloud by IwamotoScott Architecture. Voussoirs means the wedge-shaped stones that form an arch or vaulted ceiling. It seems like that people's first reaction to the piece was "...but how ?" An arch or vault usually consists of compression taking materials. But the vaults of Voussoir Cloud are made up with paper-thin wood veneer, looking fragile enough not to take any compression forces. Going through various form finding simulations, Buro Happold optimized the geometric performance of the global surface so that the internal tension enables the surface to hold itself within the material property.
Here is the design process of each triangle shaped component.
Working on CATIA model for next installation by OylerWu Collaborative. This project is sequential to their previous two installations, Pendulum Plane, and Density Field. Their highly sculptural stairs with enormous wings -again, made out of aluminum pipes (1 inch diameter)- seems challenging enough to support its own weight, but it has to support live load this time.
Since the architects designed the form going through hundreds of itterations, I simply focused on angle optimization, rather than producing veriations. Their original form consists of totally unique angles, so I constrained series of angles that appear to be similar.
But... how much decimal-place-dead-on accuracy in idealized computer world matters if those aluminum pipes are cut, bent, welded by human hands ?
This optimazation is not ment for reducing construction tolerance to achieve perfection, but for the preparation of frameworks (jigs). At least some sets of angles are constrained to be same, fabricators (students) can produce templetes for identical bending.
技術経験の面で雲の上の存在のエンジニアの方が “プロジェクト用の Web アップした” とおっしゃるので sign-in してみると、まっしろ back に まっくろ Times New Roman で箇条書き、とか。ソフトを operate しているウィンドウから PrintScreen したものを (crop すらせずに) diagram として使っているとか。もう少し絵心といいますか ...Focus が違うのね、というギャップによく遭遇します。建築とフィールドはものすごく似ているのにね。
With a few arrangements, trees become Hanabi !! (Fireworks in Japanese) They are all generated from the same script, only by changing input numbers. The pointcloud on the left corner is geometrical inputs.
Playing around with the find closest script I posted previously, I have been growing bunch of trees. If I look only the final result, it looks sort of intimidating. But I start with the simple code like find closest, and kept adding more functions and intelligence. I also work this way when I have to figure out complicated powercopy.
If you look close, tangency of branches appear more smoothly in type02. This regards the order that code generates branches. It makes sense that more points the code takes into consideration in the beginning (like in type02), the more naturally tangency can grow.
I have to learn how to call function for some part of the script, otherwise, scripts I write are getting too long to post... If somebody is interested, please let me know.
Branching is defined by multiple closest points. Each Loop finds first closest, second closest, third closest, so on and so forth. I thought it would be cool to set a variable that defines how many closest point each target has to find, but I couldn't figure it out. Let's just start with recycling the same Loop over and over again.
* this post was published after the work was opened for public * images by IwamotoScott Architecture
Unfold 前と後で surface area が一定か否かをそのつど確認。
Coming up sci-arc exhibition by IwamotoScott Architecture. Buro Happold is consulting the installation.
Challenge for this installation is the exploration on the relationship between digital model and physical result. Each triangle shaped component is given three dimensional curvature by creasing the material. But since the installation is designed initially in Rhino, when unfolded in Rhino or CATIA, it gets pretty hard to create laser-cut patterns that guarantee the same surface curvature estimated in the original Rhino model. Here is the completed installation.
The problem was the method I obtained normal direction. There is a command available to draw a line "perpendicular to curve/ from first point," but there is no ScriptMethod equivalent to it. In the previous Script, normal concerns the accuracy (tolerance) of offset, therefore not stable. So it's easier to start with normal line already built in the model. When prompted, you have to select normal curves in order.