This is a design & research project that I did in summer 2019 when I was interning at Zhangzhoujie Digital Lab in Shanghai. I researched and designed generative furnitures using Rhino & Grasshopper. I spend the two months experimenting form-finding methods, exploring fabrication techniques and creating digital and physical assets for the lab. 
I learnt a lot from Zhangzhoujie, his practice, and my friends and colleagues. As an intern, I was fortunate and privileged enough to be given a lot of creative freedom in the process.

All the Images shown here are created under the guidance and supervision of Mr. Zhangzhoujie. The Copyrights of these images and scripts belong to Zhangzhoujie Digital Lab and ENDLESS FORM. This project is for employers' view only.


About Zhangzhoujie
Zhangzhoujie's work is known for being independent, experimental and futuristic. Zhang believes that objects in the digital world can grow and morph much like things found in nature, and he is dedicated to discovering and exploring the methods within these transformations. His work mainly focuses on the simplicity of logic, variety and unpredictability, which is based on his understanding of nature.
His collections have been exhibited widely around the world and selected by museums as well as individual collectors. His work has appeared in mainstream media such as Wallpaper*, the New York Times, and Vogue.
Project 1: Algorithmic form-finding of Tables and End tables, Diamond series 
The first project I had was to find forms, both manually and algorithmic-ally, of end tables.   I first studied the previous models designed and generated in the lab. My goal was to come up with designs that fit the visual language of the lab. 

Stage One: Basic Form Testing
I started by exploring the basic forms of the tables, categorizing them by the shape of both the table top and the table base. For instance, the top can be round while the bottom is a square; top is square and bottom is circle, etc. I made 4 categories: round to round, square to square, round to square, square to round. In each category I designed 5 types, then made iterations of those different types. I used Rhino to quickly model and test the basic forms to understand proportions, shapes to inspire later design.



Stage two: Discovering surface forms using control points​​​​​​​
For the second stage, I experimented with surface form-finding using control points. I These surface forms become the starting geometry for later panelization, which is the major stylist element of Zhangzhoujie's works. 
The simple script turns any surface into triangular panels

I chose my favorite iterations and make them to be more finished products. These forms I arrived are characterized by the sub-layer in the middle, which can be used for storage.
My final designs


Stage three: Using scripts to modify the original design
After going "wild" with some ideas in stage two, my colleagues told me go back to learn more about the original design. So I studied the model of Zhangzhoujie's original design, and used grasshopper and designed scripts that can generative similar versions of the original design. I let algorithms run randomly to vary the parameters and alter the overall form of the original design. However, some results didn't turn out well. 
Zhangzhoujie's Original Designs
I wrote a script that can parametrically control the form of the end-table. (From left to right: Height, Rotation, Profile Size, Sub-divison density) 
Then I used an evolutionary-solver plug-in called Bio-morpher, in which you can plug in all the parameters in the script, and the plug-in would run, randomly vary the parameters and give you thousands of different options. You can select and evolve these designs, and the next iterations would be similar to your previous selections. In this case, "natural-selection" becomes "human-selection".
However, some of the results didn't turn out great. They seem out of proportions and too random. I decided to improve the algorithm in the next stage.


Stage four: Let the script follow some "Simple logic"
I'm not satisfied with the results of the stage three, and Mr. Zhangzhoujie helped me a lot with the next stage. He said the algorithm shouldn't be too random. Too much randomness becomes "Noise" instead of "Music". Good Music follows certain patterns. Maybe I can improve the algorithm by making it follow some simple, powerful logic.

1. Rotation
In this script, I add more parameters that can control the overall form of the end-table. I used curve graphs to affect parameters like radius, height, rotation, degree, density, distribution, etc. 


2. Rotation + Mirror


Stage five: Combining with different materials
Though the lab is best at metal fabrication. I wanted to experiment with combining the iconic metallic style of the lab with other materials like glass, marble and light. These are some results.


Stage six: living room tables​​​​​​​
Inspired by some Zaha Hadid designs, I proceeded to make some low living room tables. I maintain the "rotation" idea and made what I called "blackhole" series because the forms resemble blackholes. 


Stage seven: Squarish designs​​​​​​​
Up till this point, my designs are mostly roundish largely because I was trying to maintain the overall form of the original design. For this stage, I went more unrestrained in terms of the overall form of the table. Some are rectangular, some are organic shapes. 


Project 2: Exploring form-finding and fabrication of double-curved surface tables.

One of most difficult challenges in architecture is to develop double-curved forms into manufacturable single-curved surfaces. Zhanghzhoujie digital lab's major stylistic element is metal triangular panels at that time. And I was given this challenge to explore the possibilities and potential of using curved surfaces instead of flat panels in future designs. 
I use to two major modelling methods to make sure the surfaces produces are single-curved surfaces: Loft and Sweep 1
For lofting, as long as the two profiles are straight curves, the surface produced would be single curved. For Sweep 1, as long as the profile is a line, the result would be single-curved.​​​​​​​

Fabrication Experiments


Final Outcome & Reflections:
In the end, some designs were adopted and put into production. They also end up in an exhibition. I learnt so much from this internship in terms of the application of Generative Design, parametric tools, Arts & Crafts, Contemporary Art Market, manufacturing, professional workflow, Studio environment, collaboration, business operations. I also grew tremendously as a computational designer. 



Many thanks to 

Mr. Zhangzhoujie
All Designer colleagues
My roommate and colleague Jianan Shi
All my intern friends.

You may also like

Back to Top