ARCH 655 - Project 02

 

Project 02 - Confluence Park

April 22, 2024

Project 02 | ARCH 655 | Professor: Dr. Wei Yan

Alberto Ibarra (M.ARCH Student)

Texas A&M University

01 - Project Brief

Figure 1| Confluence Park by Rhino Display.

Project 02 builds on the development of of Project 01, where I worked with the design of the Confluence Pavilion by Lake|Flato. In Project 01 I developed parametric 2D reference lines that allowed me to adjust dimensions, scale, and height. After I utilized the polar array component to create the surface. This approach facilitated the projection of a voronoi pattern which was inspired by the design conversations with AI such as ChatGPT.

In this project, I am working with genetic algorithms in Grasshopper, specifically utilizing Galapagos, to optimize the design process and facilitate multiple design iterations. The primary objective of this project is to enhance the efficiency of design with the help of genetic algorithms to adjust the scale, height, and positioning of the structure in aims to achieve optimal shading in a given area.

02 - Case Study

Figure 2| Confluence Park. (Source: Lake|Flato, 2018).

    Confluence Park 

    Architects: Lake|Flato + Matsys Design

    Location: San Antonio, TX

    Date of Completion: 2018

    Typology: Park, Pavilion

        

    The project is situated in San Antonio, Texas and is designed as an educational park focusing on the critical role of water in the regional ecosystem. The central pavilion is composed of 22 concrete freeform forms that collectively make a network of vaults.

Figure 3| Roof Plan. (Source: Lake|Flato, 2018).

    Complemented by an education center, the design of the form was inspired by the way many plants in the South Texas region direct rainwater to their root systems. Each petal is cast-in-place concrete fabricated with fiberglass composite molds.

Figure 4| Geometry Design. (Source: Projects by Matsys, 2018)

 

03 - Project 01 Summary

    The design process began by asking the AI chatbot to: retain the original geometry of the concrete structure but add a triangular pattern that creates perforations on the surface + providing an image of the Confluence Park. 

Figure 5| Confluence Park Pavilion Interior. (Source: Lake|Flato, 2018).

This prompt resulted in the creation of the following visualization:

Figure 6 | Result 04. (Source: ChatGPT 4, 2024).

Parametric Design

    I began the modeling process by creating the linework for for the geometry that facilitated controls for scale, pedal geometry, and height.  To recreate a pattern similar to the design visualized by the AI, I utilized the 'Surface Morph' node to create a base surface where I could modify the Voronoi pattern to create a more organic and fluid pattern.

Figure 7 | Rhino Render of  'Baked' Model

Script

Figure 8 | Final Script

Render

Figure 9 | Beach Render 01

Project 2

04 - Parametric Design

    The Project begins with designating a space where our forms will be placed. With the Populate 2D node we can dictate the amount of Flower structures we want as well as add a random seed value that will serve as a gene for Galapagos. 

Figure 10 | Initial Area & Points(Grasshopper).

    Next I connect the points created to a circle component. These sets of circles will serve as base geometry to loft the surface of the form. 

Figure 11 | Geometry Sets for Loft (Grasshopper).

In this step, I utilize the gene pool that will group parametric sliders that change the form.

I established three distinct gene pools, each corresponding to a variable parameter: the radius, height, and scale proportions of the roof of each structure.

Figure 12 | Gene Pools for Galapagos (Grasshopper).

 I encountered a problem with data organization during this process. Originally, the gene pool copied the set of circles according to the gene count, which in this case was 10. However, by grafting the data at specific parameters, I was able to resolve the issue and move on.

    Finally I am able to connect the geometry to the loft component in order to visualize the surfaces of each ‘flower’.

Figure 13 | Loft (Grasshopper).

Figure 14 | Lofted Surfaces (Rhino).

05 - Genetic Algorithm (Galapagos)

    For the Galapagos component I need to set a goal to optimize to. For this project, I want to optimize the shape of the ‘flower’ form, therefore I created an expression that will compare the area of the initial rectangle to the area of the surface. This will ensure the forms will provide the needed shade with the desired freeform aesthetic of the structure.

Figure 15 | Expression for Surface Area & Galapagos Fitness (Grasshopper).

The Galapagos solver will work to adjust the genes provided: radius, height, scale, and position in order to maximize the shade provided.

Figure 16 | Genomes for Galapagos (Grasshopper).

 Galapagos Interface Solver

Figure 17 | Galapagos Solver (Galapagos).

Galapagos at Work

06 - Pattern

    Utilizing the optimized design provided by Galapagos, I wanted to apply the organic pattern from Project 01. Therefore, similar to project 01, I used the Surface Morph component to project the interpolated voronoi pattern and extrude the final surface.

Figure 18 | Voronoi Pattern Design (Grasshopper).

Figure 19 | Surface Morph & Pattern Integration (Grasshopper).

07 - Script

Figure 20 | Final Script (Grasshopper).

08 - Renders

Figure 21 | Beach Render 01 (Lumion).

        Part of the reason for Project 2 is to investigate different ways AI, Genetic Algorithms, and Parametric design can optimize the design process. Therefore to recreate renders similar to Project 1 I utilized AI rendering software to streamline the visualization workflow. This also allows me to explore variations and experiment with different materials to test design alternatives. 
The software I utilized to render the project was Visoid, a render engine that creates high quality visualizations based on an image file you provide.

Figure 22 | Visoid.

The Workflow
The workflow for this part of the project required me to take screenshots of the ‘baked’ model in Rhino render display. These images were then inputted into the application where you select a style and provide a prompt that will visualize the environment of the render. 

Figure 23 | VISOID Interface (VISOID).

These three images were utilized as input/reference images.

View 01

Figure 24 | Rhino Render View 01 (Rhino).

View 02

Figure 25 | Rhino Render View 02 (Rhino).

View 03

Figure 26 | Rhino Render View 03 (Rhino).

Then I manipulated the renders with the use of the prompt to place the design in different environments and visualize different materials.

View 01

Figure 27 | Suburbs Render 01 (VISOID).

Figure 28 | Park Render 02 (VISOID).

Figure 29 | Urban Park Render 03 (VISOID).

Figure 30 | Urban Park at Night Render 04 (VISOID).

Figure 31 | Urban Park at Night Render 05 (VISOID).

Figure 32 | Desert Render 06 (VISOID).

Figure 33 | Urban Park at Night Render 06 (VISOID).

View 02

Figure 34 | Beach Render 01 (VISOID).

Figure 35 | Urban Render 02 (VISOID).

Figure 36 | City Render 03 (VISOID).

Figure 37 | Suburbs Render 04 (VISOID).

Figure 38 | Wood Render 05 (VISOID).

Figure 39 | Render 06 (VISOID).

Figure 40 | Night Render 07 (VISOID).

Figure 41 | Wood Render 08 (VISOID).

Figure 42 | Edited Render 09 (VISOID).

View 03

Figure 43 | City Render 01 (VISOID).

Figure 44 | Park Render 02 (VISOID).

Figure 45 | City Park Render 03 (VISOID).

Figure 46 | Wood Material Render 04 (VISOID).

Figure 47 | Material Render 05 (VISOID).

Figure 48 | Metal Material Render 06 (VISOID).

Figure 49 | Render 07 (VISOID).

Figure 50 | Render 08 (VISOID).

09 - Video