Pattern Recognition by Yao Lee

PATTERN RECOGNITION

representative image of pattern, effective dpi of 300 (typical)

De Young Museum, Herzog & de Meuron, Completed in 2005 ARCH 462M | FALL 2013 CHUN LIU | YAO LI The unique faรงade pattern helps the de Young Museum and the tower to blend into nature by simulating dappled light filtering through a canopy of trees. Two Patterns, derived from the digital images of trees, are superimposed to create a layered composition in color and texture. Thus, the layered skin would manipulate light similar to environment in nature. This effect is successfully realized in the built structure, and is evident when viewing the building from virtually any vantage point. Pixel pattern is translated into an alternating grid of protruding and depressed embossings which have four depths to define different degrees of light and shadow. Copper material is chosen for its changeable quality through oxidation. When the color of copper change to a rich patina over time, the faรงade simulates mottled canopy with light and dark. The blending effect would be graceful with the surrounding natural environment. Perforation pattern is projected in six different diameters. It defines solid and void area in response to viewing, lighting and ventilation requirements. Densely wider Circular perforations are apertures which allow more light and airflow to pass through, also let tourists see through. Especially in broad staircases area, people are able to see the exterior landscape and the major square via these wider circular perforations.

SOURCE CODE

Transmittance (%)

69.0

44.1

19.7

4.8

Color to Gray Process

Based on the existing five colors showing the different porous rate, the five colors are correspondingly translated into five grayscale color to show the same meaning.

Original Unrolled Facade The original unrolled facade image was colored to show the difference of the porous density. The warmer the color is, the more open the area is. While the colder the color is, the less open the area is. To make it better to understand the perforated rate of the facade, the color vector image is going to be turned to a grayscale vector image.

SOURCE CODE DRAWING

South-East Facade

East Facade

Towards California Academy of Sciences and Music concourse

North-west Facade

West Facade

Towards Golden Gate Bridge

Towards Golden Gate Park

Orthographic Projection (elevation) with Underlying Geometric Organization from Task 2 Vector Line Drawing with Graphic Scale

8F Observation

7F Classroom

6F Education center Oﬃce

5F Artist’s studio Storage Classroom

4F Education center Storage Docent center

78%

64%

80%

Gray level

69%

18 (ft)

Grayscale Unrolled Facade

MAP 1 Set X, Y value according to the size of iamge

Grid of points

FACADE ANALYSIS facade and orientation S

Facade and Orientation

Towards California Academy of Sciences and Music concourse

Horizontal Variation The grayscale image of four elevations generated a regular curve indicating the difference between them. By pixelizing each elevation into a one-color grayscale image, we can know that the east and west face are less open and porous than the south and north. Looking at the map, it is acknowledged that the south and north elevations are more open because the south side faces towards California Academy of Sciences and Music concourse and east side faces towards Golden Gate Bridge. The higher porosity rate enables people to have a better view of the surrounding landmarks when standing on a higher level.

facade and ďŹ&#x201A;oorplan S

78%

72%

51%

68%

80%

76%

Horizontal Porosity Change GH Script 1. Get a grid of points according to the pixel amounts of the image. 2. Build a new set of points based on the X,Y value from grid and Z value from the image sampler result. 3. Generate an average set of points from the points we get in last step. 4. Build a polyline connecting all these representative average points to show the variation.

Towards Golden Gate Bridge

60%

71%

A new curve generated from an average set of points

MAP 1

facade and ďŹ&#x201A;oorplan S

78%

72%

51%

68%

80%

76%

60%

71%

Set X, Y value according to the size of iamge

Grid of points

A new curve generated from an average set of points

Horizontal Porosity Change GH Script

2.5d d

2.5d

Facade Horizontal Variation & Plan

In order to better understand the relationship between facade and building inner space, we divided the east and west elevations into three equal portions to get an average porous rate of each one. From the image, the middle part of the two elevations has the least porosity, while the elevation gets more porous on the other two sides. Compaing the facade with the plan, as the middle area is usually used as a close room with small depth, the facade can be relatively less porous. The space on the two sides are bigger and open with big depth, which refers to more light requirement, so the other two sides are more open.

Porosity and Vertical Space

1. Get a grid of points according to the pixel amounts of the image. 2. Build a new set of points based on the X,Y value from grid and Z value from the image sampler result. 3. Generate an average set of points from the points we get in last step. 4. Build a polyline connecting all these representative average points to show the variation.

ANALOGICAL MODEL 1 Grid of points for circles

Analogical Field: Horizontal Variation on Facade Based on the curve generated from previous pixel image, the information inside the curve is made to a visually understandable level. The different heights or Z values on the curve are reinterpretted to the radius of the circle to express the variation of facade. As seen in the field of circles, it shows that in the horizontal direction the porosity rate is getting bigger from one point to the its both sides.

Translate Y value from the curve as the radius

Visualization of Average Vertical Variation 1. Set a grid of points as a grid of center points for the circles. The size of the grid is related to the previous curve generated from the image. 2. Set a series of planes perpendicular to the curve, and get the Y value of the points on the curve. 3. Build the field of circles based the grid of points and Y value as radius.

ANALOGICAL MODEL 1

Circulation

Circulation Circulation

Space

Circulation

Space

Horizontal Variation Study

Space

Horizontal Variation Study

Based on the study of the horizontal direction variation, we found that the space face usually apears to have less porosity in the middle and gradually get more porous to the both sides. Moreover, the circulation face is in a more regular variable change. Then, we studied the situation if the proportion of the facade is changed what the overall variation of the facade will be.

Division point to control the face proportion Circulation

Space

Generate a curve expressing the variation of facade Circulation Circulation

Space

Circulation

Space

Translate the Z value on the curve as the radius

Changable overall variation GH Script 1. Set up a system for the curve expressing the variation in horizontal direction 2. Set a variable of facade proportion 3. Translate the changable curve (Z value on the curve) into a visual circle field (radius of circle).

Part Variation & Whole Variation

on & Whole Variation Whole Variation

F9 F7

F5 F4

F8 F4 F6F8

More Porous

66% 66%

Set X, Y value according to the size of iamge

69%

69% 69%

66%

More Porous

ous

MAP 2

Grid of points

A new curve generated from an average set of points

66%

66% 66%

63% 63%

63% F6

64% 64%

59% 59%

64%

59%

Less Porous

us Average Floor Change

Less Porous

Average Facade Change

Part Variation & Whole Variation Based an image pixelized in the horizontal direction to study the vertical change of the facade, it can generate a curve showing the porous rate change. The image of average floor porosity change (left one) indicates the average tendency and change of the whole unrolled facade. In order to compare the part, refering to each floor, with the whole, refering to whole facade, an average facade porosity change image (right one) is used to generate a curve with more variable details. By sliting this curve into six parts corresponding to each floor, the six curves are stretched to compare with the average floor change curve on the left. It can be noticed that there are three curves sharing a similar tendency with the average floor change curve, while the others are in an opposite direction. Meanwhile, it is obvious to tell that the dominant porosity change on the whole facade becomes more porous from low to high and slightly reverses at the top floor. Three floors have the same skin change features, which means the space and light experience on these floors are similar to the whole building.

Vertical Porosity Change GH Script

68% 68%

57% 57%

74% 74%

70% 70%

76% 76%

65% 65%

75% 75%

68% 68%

73% 73%

65% 65%

73% 73%

61% 61%

MorePorous Porous More

LightRequirement Requirement Light

W W

LessPorous Porous Less

N N

ContinuousProgram Program Continuous

MorePorous Porous More

LightRequirement Requirement Light

LessPorous Porous Less

Continuous ContinuousProgram Program

MAP 2

Set X, Y value according to the size of iamge

75% 75%

58% 58%

71% 71%

59% 59%

75% 75%

54% 54%

74% 74%

63% 63%

75% 75%

49% 49%

76% 76%

53% 53%

Grid of points

A new curve generated from an average set of points

W W

Observation Observation Floor Floor

Vertical Porosity Change GH Script

Classroom Classroom

1. Get a grid of points according to the pixel amounts of the image. Set the number slider according to each elevation image size. 2. Build a new set of points based on the X,Y value from grid and Z value from the image sampler result. 3. Generate an average set of points from the points we get in last step. 4. Build a polyline connecting all these representative average points to show the variation.

Education Education Center Center

Classroom/ Classroom/ Studio Studio

Education Education Center Center

Volunteer Volunteer Center Center

Facade Vertival Variation & Section To get a sense of the relationship between function and facade porosity, each elevation of the unrolled facade is pixelized to an average grayscale image to generate a curve indicating porosity variation. From the facade variation image, the east and west elevations have a similar variation tendency. The facade gets more open when the level goes up. The south and north elevations stay in a relatively stable variation conditon, which means they almost have a same porous rate on the whole elevation. To have a closer understanding of the strategy for the elevation, we studied the section of the building. The function on each floor shows a relationship with the treatment of the facade. The upper floors are mainly classroom, studio and observation space which require more light, while the lower floor is mainly for function that requires less light. Meanwhile, the staircases are set on both norht and south sides, so the elevations on these direction are made to an open and perforated skin with less variation. People walking up and down can have a very clear and open view of the outside environment.

ANALOGICAL MODEL 2 Grid of points for circles

Analogical Field: Vertical Variation on Facade Based on previous curve generated from the image of vertical variation, we get the geometric field indicating an average openning area on each floor of the facade. The value on the curve corresponds to the radius of the circle, which make the information of the curve visually understandable. This field shows that the facade gets more open and porous according to the height goes up and reverses slightly on the top floor.

Translate Y value from the curve as the radius

ANALOGICAL MODEL 2

Circulation

Space

Circulation Space

Circulation

Space

Vertical Variation Study Based on the previous study of the facade variation in the vertical direction,this step is to generate a GH model which can produce different results according the basic logic of the facade. By studying the facade, we found that the average vertical variation on the facade is mainly influenced by the area of elevation relating to different functions: circulation and space. It is clear to see that the average variation tendency is close to the dominant elevation tendency of the whole facade.

Circulation

Space

Vertical Variation Study We set the mathematical relationship of the curve generated from the elevation as a known proportion factor and use the area of each elevation as the variable to generate the average curve. The average curve will change its appearance corresponding to the change of the area factors.

Circulation

Space

Translate Y value as the radius of circle

Circulation

Space

A variable to control the face proportion

Generate a new curve

Translate Y value of the curve to radius

Curve expressing the variation of circulation/ space face Chagable Overall Variation GH Script 1. Set up the basic curve generated from previous map which contains the essential information expressing the feature of space and circulation face. 2. Translate the Y value as the radius of circle to visualize the information of the curve

Chagable Overall Variation GH Script 1. Set up a variable to control the proportion of the circulation face and space face 2. Connect the variable with previous two curves representing the circulaiton and space faces 3. Generate an average curve based on the two curves 4. Translate the Y value of the new curve as the radius

Similarity to space or facade

Circulation- levels

INTEGRATED ANALOGICAL MODEL

ace- levels space- levels

CirculationCirculationlevels levels

Similarity to space or facade

Circulation

Similarity to space or facade

60%

Space Face

60%

Space

facade relate to space and circulation

Circulation Face

facade relate to space and circulation

Similarity to Similarity space ortofacade space or facade Integrated Field of Facade

Circulation 40%

Circulation

60%

Space

Space 60%

Circulation 40%

Overall Face Related to Proportion Based on the analogical model 2 of two directions study, we generate a integratSpace Circulation Circulation ed model to show the relationship between face proportion and overall 60% porosity 60% rate in horizontal and vertical directions.

Space 60% Overall Face Related to Circulation & Space

Space

facade relate to space and circulation

facade relate facade to space relateand to space circulation and circulation

INTEGRATED ANALOGICAL MODEL

Space 60%

Circulation 80% Space 20%

Circulation 80%

Space 90% Circulation 40%

Space 60%

Circulation 40%

Space 20%

Differen Proportion Condition When the dominant face is circulation face, the overall variation in the vertial direction will appear more close to the circulaiton face variation. In the horizontal direction, the dominant tendency is more about circulation face.

Circulation 80%

tion Space 20%

Differen Proportion Condition When the dominant face is space face, the overall variation in the vertial direction will appear more close to the space face variation. In the horizontal direction, the dominant tendency is more about space face.

Space 60%

Space 60% Circulation 40%

Circulation 40%

Space 90%

Circulation 10%

Circulation 80% Space 20%

Circulation 10%

Space 60%

Space 90% Circulation 40% Circulation 10%

Chagable Integrated Variation GH Script

Differen Proportion Condition This situation is the original proportion of the facade. The vertical and horizontal variation are the same as shown in the map 2 study.

Based on the previous study, this is the integrated model to show when the proportion of the circulaiton and space faces change, the variation of the facade will change correspondingly in an intrinsic logic, which we got from the mapping study.

TRANSLATION TESTING

Study Model- Evolution Sequence Based on the analysis from the previous study of source code, the strategy is to create openings and solid on the surface according to the variation logic of the study facade. Starting from a two-layer model, the way to manipulate the surfaces became more three-dimensional and variable. Lastly, we found a triangular shape as the basic element to generate a pattern for the model. By folding and assembling different sizes of triangles on the surface, it creates more possibilities of form.

TRANSLATION TESTING

Final Model 1 Study Model- Prototype toward Final After a sequece of tesing and modeling of the triangular shape pattern, the final models tend to use a diamond shape as the basic element for the pattern. By folding the diamond on the surface, it generates connection of two layers and creates multilevel space, which has a big potential to be used as a prototype for a real building in the city.

Final Model 2

TRANSLATION PRODUCTION 1

Cut Score

Circulation 10% Space 20% Circulation 80%

Space 60%

Circulation 40% Space 90%

Model Photo

Unrolled AutoCAD File

With three-layer folding and attaching object, it creates the possibility of circulaiton and space. Moreover, the connections can work as the structure system in a real architecture integrating with circulation system and enclosing parts. At this point, this form or object give the potential to be a public space such as market, parking building or landscape.

t Cu re o Sc

+ Application of the analysis from map 2

Way of folding and attaching

EXPRESSION 1

Site Plan

Section

One possibility of this geometrical form is to be an additional part of a market located in Spain. This form has its potential to be a roof surface within a thickness of space. The shape of diamond flips to connect upper level as a structural geometry, at the same time, the diamond figure on the plan suggests the area of space can be occupied by people as a shop in the market. The size of the diamond can be controlled by different shop program to meet different need of area.

Shop Area

Middle Level Plan

EXPRESSION 1

Circulation

Shop Space

Aerial View

Model

The interesting thing of a single unit is that it encloses and defines a space and stands to be a part of structural system integrating with circulation. In addition, this kind of unit could be repeated and used as simple as a parking lot unit.

Higher Level Circulation

Circulation

Shop Space Shop Space

Lower Level Diagram of a single unit

Diagram of a single unit

TRANSLATION PRODUCTION 2 surface to be glued 2

2 1

1 surface to be glued

folding diagram

Score Cut

Unrolled surface to lasercut

EXPRESSION 2

Plan or Elevation (depending on final orientation & demonstrating scale and context)

Section (demonstrate scale and context)

Plan of Trestle Shell

Section The different size of pieces are both structures and the facade of Trestle shell, in the middle the pieces become larger to offer the seats for people.

The pattern is translated into spatial expression of one part of trestle park in St. louis. Map 1 controls the width of the grid to adapt the structure of the trestle and map 2 controls the views of different angles from the Arch and the Greenland in the east of Mississippi River.

map2

views more void space towards Arch

map1 Circulation 80%

Space 60%

Space 20%

Space 90%

Trestle structure

Circulation 40% Circulation 10%

Circulation 80%

Space 60%

Space 20%

Grid

Map 1, 2 imput

Space 90% Circulation 40% Circulation 10%

Function and view

EXPRESSION 2

Include a 3d visualization of your Expression in context (Where renderings are difficult, Photoshopped versions of your final Expression model may be of great benefit... and are strongly encouraged)

Perspective view The structure offers opennings for people to see through in different height and angles.

Final model